Strategy_by_technical_indicator_v6_4(+ve&_ve)_Training.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Strategy by technical indicator v6.4(+ve&-ve)_Training.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.4"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/bradbradlin/Machine-Learning-Projects/blob/main/Strategy_by_technical_indicator_v6_4(%2Bve%26_ve)_Training.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nxRqHTpwr6VH"
      },
      "source": [
        "To use TI signal and TI\n",
        "\n",
        "Two classes:  Positive return or Negative return\n",
        "\n",
        "Weight entropy loss function\n",
        "\n",
        "Use tf.feature_column\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NxydUF46r6VI",
        "outputId": "3d183ca9-8b99-4ecd-e88e-330e191e478b"
      },
      "source": [
        "import tensorflow as tf\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "import math\n",
        "import keras.backend as K\n",
        "from tensorflow import feature_column\n",
        "from tensorflow.keras.layers.experimental import preprocessing\n",
        "from tensorflow.keras.utils import to_categorical\n",
        "from tensorflow.keras.callbacks import EarlyStopping\n",
        "from tensorboard.plugins.hparams import api as hp\n",
        "print(tf.__version__)"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2.4.1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "eLS0Sflrj16D",
        "outputId": "21aaa6bb-c28a-40a7-9bb9-20fba3e1e7af"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')\n",
        "\n",
        "%cd /content/gdrive/My Drive/Colab Notebooks/AI trading project"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/gdrive\n",
            "/content/gdrive/My Drive/Colab Notebooks/AI trading project\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7eF-bxr8r6VN"
      },
      "source": [
        "class myCallback(tf.keras.callbacks.Callback):\n",
        "    def on_epoch_end(self, epoch, logs={}):\n",
        "        if(logs.get('accuracy')>0.8):\n",
        "            print(\"Reached 80% accuracy so cancelling training!\")\n",
        "            self.model.stop_training = True"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OEB0qJfpr6VN"
      },
      "source": [
        "def tf_dataset(df, target, batch_size, shuffle=True):\n",
        "    # To convert to tf dataset\n",
        "    df2 = df.copy()\n",
        "    #target = df2.pop(TARGET)\n",
        "    dataset = tf.data.Dataset.from_tensor_slices((dict(df2), target))\n",
        "    if shuffle:\n",
        "        dataset = dataset.shuffle(len(df2))\n",
        "        dataset = dataset.batch(batch_size)\n",
        "    else:\n",
        "        dataset = dataset.batch(batch_size)\n",
        "    return dataset"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9T49C2cv7fQe"
      },
      "source": [
        "def get_focal_params(y_pred):\n",
        "    epsilon = tf.constant(1e-9)\n",
        "    gamma = tf.constant(3.)\n",
        "    y_pred = y_pred + epsilon\n",
        "    pinv = 1./y_pred\n",
        "    pos_weight_f = (pinv - 1)**gamma\n",
        "    weight_f = y_pred**gamma\n",
        "    return pos_weight_f, weight_f\n",
        "\n",
        "def custom_loss(y_true,y_pred):\n",
        "    # the custom loss function for imbalance dataset\n",
        "    y_pred_prob = tf.keras.backend.sigmoid(y_pred)    \n",
        "    pos_weight_f, weight_f = get_focal_params(y_pred_prob)\n",
        "    alpha = tf.constant(.35)\n",
        "    alpha_ = 1 - alpha\n",
        "    alpha_div = alpha / alpha_\n",
        "    pos_weight = pos_weight_f * alpha_div\n",
        "    weight = weight_f * alpha_\n",
        "    l2 = weight * tf.nn.weighted_cross_entropy_with_logits\\\n",
        "    (labels=y_true, logits=y_pred, pos_weight=weight)\n",
        "    return l2"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fN6cHDJu7xK-"
      },
      "source": [
        "# def custom_loss2(y_true,y_pred):\n",
        "#     class_count_df = raw_df.groupby('return').count()\n",
        "#     n_0, n_1 = class_count_df.iloc[0, 0], class_count_df.iloc[1,0]\n",
        "#     w_0 = (n_0 + n_1) / (2.0 * n_0)\n",
        "#     w_1 = (n_0 + n_1) / (2.0 * n_1)\n",
        "#     class_weights=torch.tf.tensor([w_0,w_1])"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XsawHk-_r6VO"
      },
      "source": [
        "def get_f1(y_true, y_pred): #taken from old keras source code\n",
        "    # to calculate F1 \n",
        "    true_positives = K.sum(K.round(K.clip(y_true * y_pred, 0, 1)))\n",
        "    possible_positives = K.sum(K.round(K.clip(y_true, 0, 1)))\n",
        "    predicted_positives = K.sum(K.round(K.clip(y_pred, 0, 1)))\n",
        "    precision = true_positives / (predicted_positives + K.epsilon())\n",
        "    recall = true_positives / (possible_positives + K.epsilon())\n",
        "    f1_val = 2*(precision*recall)/(precision+recall+K.epsilon())\n",
        "    return f1_val"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0iGQlPA1ABIZ"
      },
      "source": [
        "def get_normalization_layer(name, dataset):\n",
        "  # Create a Normalization layer for our feature.\n",
        "    normalizer = preprocessing.Normalization()\n",
        "\n",
        "    # Prepare a Dataset that only yields our feature.\n",
        "    feature_ds = dataset.map(lambda x, y: x[name])\n",
        "\n",
        "    # Learn the statistics of the data.\n",
        "    normalizer.adapt(feature_ds)\n",
        "\n",
        "    return normalizer"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "njQ1PaKgSdyA"
      },
      "source": [
        "def get_category_encoding_layer(name, dataset, dtype, max_tokens=None):\n",
        "      # Create a StringLookup layer which will turn strings into integer indices\n",
        "    if dtype == 'string':\n",
        "        index = preprocessing.StringLookup(max_tokens=max_tokens)\n",
        "    else:\n",
        "        index = preprocessing.IntegerLookup(max_values=max_tokens)\n",
        "\n",
        "    # Prepare a Dataset that only yields our feature\n",
        "    feature_ds = dataset.map(lambda x, y: x[name])\n",
        "\n",
        "      # Learn the set of possible values and assign them a fixed integer index.\n",
        "    index.adapt(feature_ds)\n",
        "\n",
        "    # Create a Discretization for our integer indices.\n",
        "    encoder = preprocessing.CategoryEncoding(max_tokens=index.vocab_size())\n",
        "\n",
        "    # Prepare a Dataset that only yields our feature.\n",
        "    feature_ds = feature_ds.map(index)\n",
        "\n",
        "    # Learn the space of possible indices.\n",
        "    encoder.adapt(feature_ds)\n",
        "\n",
        "    # Apply one-hot encoding to our indices. The lambda function captures the\n",
        "    # layer so we can use them, or include them in the functional model later.\n",
        "    return lambda feature: encoder(index(feature))"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1jQXAGZkr6VN",
        "scrolled": true
      },
      "source": [
        "callbacks = myCallback()\n",
        "#@tf.function\n",
        "def train_model(train_ds, val_ds, epoch, a_inputs, enc_features): \n",
        "    L1L2 = tf.keras.regularizers.l1_l2(l1=0.1, l2=0.1)\n",
        "\n",
        "    tf.keras.backend.clear_session()\n",
        "    tf.keras.backend.set_floatx('float32')\n",
        "  \n",
        "    merged =  tf.keras.layers.concatenate(enc_features)\n",
        "    dense1 =  tf.keras.layers.Dense(128, activation=tf.nn.relu)(merged)\n",
        "    dense2 =  tf.keras.layers.Dense(128, activation=tf.nn.relu)(dense1)\n",
        "    dropout1 = tf.keras.layers.Dropout(0.05)(dense2)\n",
        "    # dense3 =  tf.keras.layers.Dense(512, activation=tf.nn.relu)(dense2)\n",
        "    # dense4 =  tf.keras.layers.Dense(1028, activation=tf.nn.relu)(dense3)    \n",
        "    # dense5 =  tf.keras.layers.Dense(512, activation=tf.nn.relu)(dense4)\n",
        "    #dense6 =  tf.keras.layers.Dense(256, activation=tf.nn.relu)(dense5)\n",
        "    #dense7 =  tf.keras.layers.Dense(64, activation=tf.nn.relu)(dense6)\n",
        "    output =  tf.keras.layers.Dense(num_category, activation=tf.nn.softmax)(dropout1)#)(dense5)\n",
        "                                   #kernel_regularizer=L1L2,)(dense5)\n",
        "\n",
        "    model =  tf.keras.models.Model(a_inputs, output)\n",
        "\n",
        "    opt = tf.keras.optimizers.Adam(\n",
        "        learning_rate=1e-4, beta_1=0.9, beta_2=0.999, epsilon=1e-07, amsgrad=False,\n",
        "        name='adam',)\n",
        "    es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=5)\n",
        "    \n",
        "    model.compile(optimizer=opt,\n",
        "                  #loss = 'binary_crossentropy', \n",
        "                  loss='categorical_crossentropy',\n",
        "                  #loss= custom_loss,\n",
        "                  metrics=['accuracy',get_f1])\n",
        "\n",
        "    history = model.fit(train_ds,\n",
        "          epochs=epoch, \n",
        "          validation_data=val_ds,\n",
        "          verbose = 2,\n",
        "          callbacks=[es])\n",
        "    \n",
        "    return model, history"
      ],
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1Ga7z8cI6Esm"
      },
      "source": [
        "def create_fea_col(train_ds):\n",
        "    all_inputs = []\n",
        "    encoded_features = []\n",
        "    # Numeric features.\n",
        "    for header in CONT_NAMES:\n",
        "        numeric_col = tf.keras.Input(shape=(1,), name=header)\n",
        "        normalization_layer = get_normalization_layer(header, train_ds)\n",
        "        encoded_numeric_col = normalization_layer(numeric_col)\n",
        "        all_inputs.append(numeric_col)\n",
        "        encoded_features.append(encoded_numeric_col)\n",
        "\n",
        "    # Categorical features encoded as int.\n",
        "    for header in CAT_NAMES:\n",
        "        categorical_col = tf.keras.Input(shape=(1,), name=header, dtype='string')\n",
        "        encoding_layer = get_category_encoding_layer(header, train_ds, dtype='string', max_tokens=5)\n",
        "        encoded_categorical_col = encoding_layer(categorical_col)\n",
        "        all_inputs.append(categorical_col)\n",
        "        encoded_features.append(encoded_categorical_col)\n",
        "\n",
        "    return all_inputs, encoded_features"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cb0RJa5vr6VL"
      },
      "source": [
        "# load technical indicator from csv\n",
        "filename = 'tech_GSPC_v2_20201227.csv'\n",
        "raw_df = pd.read_csv(filename)[200:]\n",
        "# raw_df2 = pd.read_csv('tech_NDX_v2_20D.csv')[200:-1]\n",
        "# raw_df3 = pd.read_csv('tech_HSI_v2_20D.csv')[200:-1]\n",
        "# raw_df4 = pd.read_csv('tech_HSCE_v2_20D.csv')[200:-1]\n",
        "# raw_df5 = pd.read_csv('tech_DJI_v2_20D.csv')[200:-1]\n",
        "# raw_df6 = pd.read_csv('tech_GDAXI_v2_20D.csv')[200:-1]\n",
        "# raw_df7 = pd.read_csv('tech_RUT_v2_20D.csv')[200:-1]\n",
        "# raw_df8 = pd.read_csv('tech_STI_v2_20D.csv')[200:-1]\n",
        "# raw_df9 = pd.read_csv('tech_N225_v2_20D.csv')[200:-1]\n",
        "# raw_df10 = pd.read_csv('tech_FCHI_v2_20D.csv')[200:-1]\n",
        "# raw_df11 = pd.read_csv('tech_AXJO_v2_20D.csv')[200:-1]\n",
        "\n",
        "# raw_df = raw_df.append(raw_df2, ignore_index=True)\n",
        "# raw_df = raw_df.append(raw_df3, ignore_index=True)\n",
        "# raw_df = raw_df.append(raw_df4, ignore_index=True)\n",
        "# raw_df = raw_df.append(raw_df5, ignore_index=True)\n",
        "# raw_df = raw_df.append(raw_df6, ignore_index=True)\n",
        "# raw_df = raw_df.append(raw_df7, ignore_index=True)\n",
        "# raw_df = raw_df.append(raw_df8, ignore_index=True)\n",
        "# raw_df = raw_df.append(raw_df9, ignore_index=True)\n",
        "# raw_df = raw_df.append(raw_df10, ignore_index=True)\n",
        "# raw_df = raw_df.append(raw_df11, ignore_index=True)\n",
        "\n",
        "# VIX_df = pd.read_csv('VIX.csv')\n",
        "# SKEW_df = pd.read_csv('SKEW.csv')\n",
        "\n",
        "raw_df.drop(['Unnamed: 0'], axis=1, inplace=True)\n",
        "raw_df.reset_index(inplace=True)\n",
        "raw_df.drop(['index'], axis=1, inplace=True)\n",
        "raw_df['date'] = pd.to_datetime(raw_df['date'])\n",
        "# VIX_df['date'] = pd.to_datetime(VIX_df['date'])\n",
        "# SKEW_df['date'] = pd.to_datetime(VIX_df['date'])"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sTw2h_fIME4o"
      },
      "source": [
        "#get VIX index\n",
        "VIX_df = pd.read_csv('^VIX.csv')\n",
        "VIX_df['Date'] = pd.to_datetime(VIX_df['Date'])\n",
        "# get probability of recession by new york bank\n",
        "recess_df = pd.read_excel('prob_recess.xls')\n",
        "recess_df['Date'] = pd.to_datetime(recess_df['Date'])\n",
        "recess_df['month'] = recess_df['Date'].dt.month.astype('str')\n",
        "recess_df['year'] = recess_df['Date'].dt.year.astype('str')\n",
        "# #get USD dollar index\n",
        "# stock1 = Ticker('DX-Y.NYB')\n",
        "# USD_df = stock1.history(period=period).reset_index()\n",
        "# USD_df['date'] = pd.to_datetime(USD_df['date'])\n",
        "\n",
        "raw_df = pd.merge(raw_df, VIX_df, how=\"inner\", left_on='date', right_on='Date',suffixes=(\"\",\"_VIX\"))\n",
        "raw_df['month'] = raw_df['date'].dt.month.astype('str')\n",
        "raw_df['year'] = raw_df['date'].dt.year.astype('str')\n",
        "raw_df = pd.merge(raw_df, recess_df, how=\"inner\", left_on=['month','year'], right_on=['month','year'],suffixes=(\"\",\"_recess\"))\n",
        "# raw_df = pd.merge(raw_df, USD_df, how=\"inner\", left_on='date', right_on='date',suffixes=(\"\",\"_USDIdx\"))\n",
        "\n",
        "raw_df = raw_df.drop(['Date','Open','High','Low','Adj Close','Volume'\n",
        "                            ,'month','year','Date_recess','NBER_Rec','10 Year Treasury Yield',\n",
        "                            '3 Month Treasury Yield','3 Month Treasury Yield (Bond Equivalent Basis)'],axis=1)\n",
        "#'symbol_USDIdx','open_USDIdx','high_USDIdx','volume_USDIdx','low_USDIdx','adjclose_USDIdx'],axis=1)\n",
        "\n",
        "raw_df = raw_df.rename(columns={'Close': 'Close_VIX'})"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qt9UH7nir6VM",
        "outputId": "75428997-580a-4dc7-d806-e17738bc81b9"
      },
      "source": [
        "#choose forcast days\n",
        "day = 10\n",
        "\n",
        "day_dict = {1: 'marking date return 1D', 5: 'marking date return 5D', 10: 'marking date return 10D'}\n",
        "\n",
        "week_dict = {1: 'weekday1', 5: 'weekday5', 10: 'weekday10'}\n",
        "month_dict = {1: 'month1', 5: 'month5', 10: 'month10'}\n",
        "\n",
        "# to_catergorical input factors\n",
        "CAT_NAMES = list(raw_df.loc[:, raw_df.dtypes == 'int64'].columns)[:-3]\n",
        "CAT_NAMES.extend(['volatility_bbhi','volatility_bbli','volatility_kchi', \n",
        "                  'volatility_kcli', 'trend_psar_up_indicator', 'trend_psar_down_indicator',\n",
        "                 week_dict[day], month_dict[day]])\n",
        "# CAT_NAMES = ['CDLHAMMER',\n",
        "#  'CDLHIKKAKE',\n",
        "#  'trend_psar_up_indicator',\n",
        "#  'trend_psar_down_indicator',\n",
        "#  'CDLSEPARATINGLINES',\n",
        "#  'CDLHARAMI',\n",
        "#  'CDL3OUTSIDE',\n",
        "#  'CDLINVERTEDHAMMER',\n",
        "#  'CDL3INSIDE',\n",
        "#  'CDLADVANCEBLOCK',\n",
        "#  'CDLEVENINGSTAR',\n",
        "#  'CDLXSIDEGAP3METHODS',\n",
        "#  'CDLSHOOTINGSTAR',\n",
        "#  'CDLTHRUSTING',\n",
        "#  'CDLTASUKIGAP']\n",
        "\n",
        "print(CAT_NAMES)\n",
        "\n",
        "a = list(raw_df.loc[:, raw_df.dtypes == 'float64'].columns)\n",
        "CONT_NAMES = [x for x in a if x not in ['open','volume','low','high','close','adjclose', \n",
        "                                        'volatility_bbhi', 'volatility_bbli', 'volatility_kchi', \n",
        "                                        'volatility_kcli', 'trend_psar_up_indicator', 'trend_psar_down_indicator',\n",
        "                                       'weekday1','month1','weekday5','month5','weekday10','month10','marking date return 1D','marking date return 5D','marking date return 10D']]\n",
        "\n",
        "# raw_df[CAT_NAMES]= raw_df[CAT_NAMES].replace(-1, 'Sell')\n",
        "# raw_df[CAT_NAMES]= raw_df[CAT_NAMES].replace(0, 'NA')\n",
        "# raw_df[CAT_NAMES]= raw_df[CAT_NAMES].replace(1, 'Buy')\n",
        "raw_df[CAT_NAMES] = raw_df[CAT_NAMES].astype('string')\n",
        "#raw_df = raw_df.dropna(how=\"any\", axis=0)\n",
        "\n",
        "\n",
        "#convert to catergorical output\n",
        "TARGET = ['negative', 'no' , 'positive']\n",
        "#y = raw_df.apply(lambda x: 1 if x[day_dict[day]]>=0.005 else 0, axis=1)\n",
        "y = raw_df.apply(lambda x: 2 if x[day_dict[day]]>=0.003 else (0 if x[day_dict[day]]<=-0.003 else 1), axis=1)\n",
        "raw_df[TARGET] = pd.DataFrame(to_categorical(y), columns=TARGET)\n",
        "raw_df[TARGET] = raw_df[TARGET].astype('int')\n",
        "print(raw_df[TARGET].value_counts())\n",
        "# raw_df = raw_df.fillna(0)\n",
        "#raw_df = raw_df.dropna(how=\"any\", axis=0)\n",
        "# raw_df = pd.merge(raw_df, VIX_df, how='left', left_on='date', right_on='date')\n",
        "# raw_df = pd.merge(raw_df, SKEW_df, how='left', left_on='date', right_on='date')\n",
        "raw_df2 = raw_df.copy()\n",
        "\n",
        "Input_col = []\n",
        "Input_col.extend(CAT_NAMES)\n",
        "Input_col.extend(CONT_NAMES)\n",
        "Input_col.extend(TARGET)\n",
        "Input_col.append('date')\n",
        "Input_col.append('symbol')\n",
        "\n",
        "raw_df = raw_df[raw_df.columns[raw_df.columns.isin(Input_col)]]\n",
        "raw_df = raw_df.fillna('0')\n",
        "org_df = raw_df.copy()"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "['CDL2CROWS', 'CDL3BLACKCROWS', 'CDL3INSIDE', 'CDL3LINESTRIKE', 'CDL3OUTSIDE', 'CDL3STARSINSOUTH', 'CDL3WHITESOLDIERS', 'CDLABANDONEDBABY', 'CDLADVANCEBLOCK', 'CDLBELTHOLD', 'CDLBREAKAWAY', 'CDLCLOSINGMARUBOZU', 'CDLCONCEALBABYSWALL', 'CDLDARKCLOUDCOVER', 'CDLDOJI', 'CDLDOJISTAR', 'CDLDRAGONFLYDOJI', 'CDLENGULFING', 'CDLEVENINGDOJISTAR', 'CDLEVENINGSTAR', 'CDLGAPSIDESIDEWHITE', 'CDLGRAVESTONEDOJI', 'CDLHAMMER', 'CDLHANGINGMAN', 'CDLHARAMI', 'CDLHARAMICROSS', 'CDLHIGHWAVE', 'CDLHIKKAKE', 'CDLHIKKAKEMOD', 'CDLHOMINGPIGEON', 'CDLIDENTICAL3CROWS', 'CDLINNECK', 'CDLINVERTEDHAMMER', 'CDLKICKING', 'CDLLADDERBOTTOM', 'CDLLONGLEGGEDDOJI', 'CDLMARUBOZU', 'CDLMATCHINGLOW', 'CDLMATHOLD', 'CDLMORNINGDOJISTAR', 'CDLMORNINGSTAR', 'CDLONNECK', 'CDLPIERCING', 'CDLRICKSHAWMAN', 'CDLRISEFALL3METHODS', 'CDLSEPARATINGLINES', 'CDLSHOOTINGSTAR', 'CDLSPINNINGTOP', 'CDLSTICKSANDWICH', 'CDLTAKURI', 'CDLTASUKIGAP', 'CDLTHRUSTING', 'CDLTRISTAR', 'CDLUNIQUE3RIVER', 'CDLUPSIDEGAP2CROWS', 'CDLXSIDEGAP3METHODS', 'volatility_bbhi', 'volatility_bbli', 'volatility_kchi', 'volatility_kcli', 'trend_psar_up_indicator', 'trend_psar_down_indicator', 'weekday10', 'month10']\n",
            "negative  no  positive\n",
            "0         0   1           4280\n",
            "1         0   0           2760\n",
            "0         1   0            768\n",
            "dtype: int64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "f0LscRU1tdRY",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bde74287-543b-4d7a-b73e-45dd3cd3936d"
      },
      "source": [
        "batch_size = 32\n",
        "eval_start = '20070101'\n",
        "val_start = '20180101'\n",
        "tf.keras.backend.clear_session()\n",
        "tf.random.set_seed(51)\n",
        "np.random.seed(51)\n",
        "\n",
        "raw_df = raw_df[raw_df['date']>=eval_start].reset_index()\n",
        "train_df = raw_df[raw_df['date']<val_start]\n",
        "train_df = train_df.set_index('date')\n",
        "train_df = train_df.drop(['index'], axis=1)\n",
        "train_df[CONT_NAMES] = train_df[CONT_NAMES].astype('float32')\n",
        "train_df['symbol'] = train_df['symbol'].astype('string')\n",
        "train_df[CAT_NAMES] = train_df[CAT_NAMES].astype('string')\n",
        "\n",
        "val_df = raw_df[raw_df['date']>=val_start]\n",
        "val_df = val_df.set_index('date')\n",
        "val_df = val_df.drop(['index'], axis=1)\n",
        "val_df[CONT_NAMES] = val_df[CONT_NAMES].astype('float32')\n",
        "val_df['symbol'] = val_df['symbol'].astype('string')\n",
        "val_df[CAT_NAMES] = val_df[CAT_NAMES].astype('string')\n",
        "\n",
        "temp_train = train_df.loc[:, Input_col[:len(CAT_NAMES)+len(CONT_NAMES)]]\n",
        "temp_val = val_df.loc[:, Input_col[:len(CAT_NAMES)+len(CONT_NAMES)]]\n",
        "# only top15 important fesatures\n",
        "COL = []\n",
        "COL.extend(CAT_NAMES)\n",
        "COL.extend(CONT_NAMES)\n",
        "\n",
        "# CAT_NAMES = ['month10']\n",
        "# CONT_NAMES = ['volatility_kcw','volume_nvi', 'trend_visual_ichimoku_b','volatility_atr',\n",
        "# 'volume_obv', 'trend_adx', 'trend_kst_sig', 'trend_mass_index',\n",
        "# 'volume_mfi', 'trend_aroon_ind', 'volatility_dcw', 'trend_adx_neg', 'trend_kst_diff']\n",
        "   \n",
        "temp_train = temp_train.loc[:,COL]\n",
        "temp_val = temp_val.loc[:,COL]\n",
        "\n",
        "\n",
        "train1_ds = tf_dataset(temp_train, train_df[TARGET], batch_size, True)\n",
        "train_ds = train1_ds.take(int(len(train_df)/batch_size*0.9))\n",
        "# # train_dev dataset for validate variance of model\n",
        "train_dev_ds = train1_ds.skip(int(len(train_df)/batch_size*0.9))\n",
        "val_ds = tf_dataset(temp_val, val_df[TARGET], batch_size, False)\n",
        "train2_ds = tf_dataset(temp_train, train_df[TARGET], batch_size, False)\n",
        "\n",
        "allinputs, encodedfeatures = create_fea_col(train_ds)\n",
        "\n",
        "print(train_df[TARGET].value_counts())"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "negative  no  positive\n",
            "0         0   1           1539\n",
            "1         0   0            956\n",
            "0         1   0            274\n",
            "dtype: int64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VGEOuSEer6VO",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# num_category = len(TARGET)\n",
        "# model2, history = train_model(train_ds, val_ds, 500, allinputs, encodedfeatures)\n",
        "# model2.evaluate(train_dev_ds)\n",
        "# model2.evaluate(val_ds)"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2oY49g6yLUds",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "#model2.summary()"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OEzGwOabr6VO",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# y_true = []\n",
        "# y_pred = []\n",
        "# val_inputs = []\n",
        "# thres = 0.5\n",
        "# #index = '^GSPC'\n",
        "\n",
        "# for x, y in val_ds:\n",
        "#     for j in y:\n",
        "#         y_true.append(j.numpy())\n",
        "\n",
        "# y_pred = pd.DataFrame(model2.predict(val_ds))\n",
        "# #y_pred = pd.DataFrame(np.argmax(model2.predict(val_ds), axis=1))\n",
        "# y_true = pd.DataFrame(np.argmax(y_true, axis=1))\n",
        "# result = pd.concat([y_true, y_pred], axis=1, ignore_index=True)\n",
        "# result.columns=['true_return', 'pred_neg', 'na','pred_pos']\n",
        "# result['true(+/-)'] = result.apply(lambda x: '+ve' if x['true_return']==2 else ('-ve' if x['true_return']==0 else'no') , axis=1)\n",
        "# result['pred(+/-)']  = result.apply(lambda x: '+ve' if x['pred_pos']>thres else ('-ve' if x['pred_neg']>thres else'no'), axis=1)\n",
        "\n",
        "# # result['true(+/-)'] = y_true.apply(lambda x: '++ve' if x[0]==3 else ('+ve' if x[0]==2 else ('no' if x[0]==1 else ('-ve' if x[0]==0 else '--ve'))), axis=1)\n",
        "# # result['pred(+/-)']  = y_pred.apply(lambda x: '++ve' if x[0]==3 else ('+ve' if x[0]==2 else ('no' if x[0]==1 else ('-ve' if x[0]==0 else '--ve'))), axis=1)\n",
        "# result.to_excel('result.xls')\n",
        "# #result['acc'] = result.apply(lambda x: 1 if result['true(+/-)'] == result['pred(+/-)']  else 0, axis=1)\n",
        "# print(result['pred(+/-)'].value_counts())\n",
        "\n",
        "# from sklearn.metrics import classification_report\n",
        "# print(CAT_NAMES)\n",
        "# print(classification_report(result['true(+/-)'] , result['pred(+/-)']))\n",
        "\n",
        "# result.to_csv('output_pos_val.csv')\n",
        "# result.to_excel('output_pos_val.xls')"
      ],
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8Qh6dOS1jnC6",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# y_true = []\n",
        "# y_pred = []\n",
        "# val_inputs = []\n",
        "# thres = 0.5\n",
        "# #index = '^GSPC'\n",
        "\n",
        "# for x, y in train2_ds:\n",
        "#     for j in y:\n",
        "#         y_true.append(j.numpy())\n",
        "\n",
        "# y_pred = pd.DataFrame(model2.predict(train2_ds))\n",
        "# #y_pred = pd.DataFrame(np.argmax(model2.predict(val_ds), axis=1))\n",
        "# y_true = pd.DataFrame(np.argmax(y_true, axis=1))\n",
        "# result = pd.concat([y_true, y_pred], axis=1, ignore_index=True)\n",
        "# result.columns=['true_return', 'pred_neg', 'na','pred_pos']\n",
        "# result['true(+/-)'] = result.apply(lambda x: '+ve' if x['true_return']==2 else ('-ve' if x['true_return']==0 else'no') , axis=1)\n",
        "# result['pred(+/-)']  = result.apply(lambda x: '+ve' if x['pred_pos']>thres else ('-ve' if x['pred_neg']>thres else'no'), axis=1)\n",
        "\n",
        "# # result['true(+/-)'] = y_true.apply(lambda x: '++ve' if x[0]==3 else ('+ve' if x[0]==2 else ('no' if x[0]==1 else ('-ve' if x[0]==0 else '--ve'))), axis=1)\n",
        "# # result['pred(+/-)']  = y_pred.apply(lambda x: '++ve' if x[0]==3 else ('+ve' if x[0]==2 else ('no' if x[0]==1 else ('-ve' if x[0]==0 else '--ve'))), axis=1)\n",
        "# #result.to_excel('result.xls')\n",
        "# #result['acc'] = result.apply(lambda x: 1 if result['true(+/-)'] == result['pred(+/-)']  else 0, axis=1)\n",
        "# print(result['pred(+/-)'].value_counts())\n",
        "\n",
        "# from sklearn.metrics import classification_report\n",
        "# print(CAT_NAMES)\n",
        "# print(classification_report(result['true(+/-)'] , result['pred(+/-)']))\n",
        "\n",
        "# result.to_csv('output_pos.csv')"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eW25wTjGpalx",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# import matplotlib.image  as mpimg\n",
        "# import matplotlib.pyplot as plt\n",
        "\n",
        "# #-----------------------------------------------------------\n",
        "# # Retrieve a list of list results on training and test data\n",
        "# # sets for each training epoch\n",
        "# #-----------------------------------------------------------\n",
        "# mae=history.history['get_f1']\n",
        "# loss=history.history['loss']\n",
        "# v_loss=history.history['val_loss']\n",
        "# v_mae = history.history['val_get_f1']\n",
        "# epochs=range(len(loss)) # Get number of epochs\n",
        "\n",
        "# #------------------------------------------------\n",
        "# # Plot MAE and Loss\n",
        "# #------------------------------------------------\n",
        "# plt.plot(epochs, mae, 'r')\n",
        "# plt.plot(epochs, loss, 'b')\n",
        "# plt.title('Train F1 and Loss')\n",
        "# plt.xlabel(\"Epochs\")\n",
        "# plt.ylabel(\"F1\")\n",
        "# plt.legend([\"F1\", \"Loss\"])\n",
        "\n",
        "# plt.figure()\n",
        "\n",
        "# epochs_zoom = epochs[:]\n",
        "# mae_zoom = v_mae[:]\n",
        "# loss_zoom = v_loss[:]\n",
        "\n",
        "# #------------------------------------------------\n",
        "# # Plot Zoomed MAE and Loss\n",
        "# #------------------------------------------------\n",
        "# plt.plot(epochs_zoom, mae_zoom, 'r')\n",
        "# plt.plot(epochs_zoom, loss_zoom, 'b')\n",
        "# plt.title('Val F1 and Loss')\n",
        "# plt.xlabel(\"Epochs\")\n",
        "# plt.ylabel(\"F1\")\n",
        "# plt.legend([\"F1\", \"Loss\"])\n",
        "\n",
        "# plt.figure()"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "00sGDdaF5Cy1"
      },
      "source": [
        "Testing expanding windows"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kuSlErhpLdwL",
        "outputId": "e610e2db-21f2-4cb8-cfe3-41b7967a8a56"
      },
      "source": [
        "model1 = tf.keras.models.load_model('saved_model/my_model')\n",
        "\n",
        "# Check its architecture\n",
        "model1.summary()"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:5 out of the last 5 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7b85f0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:6 out of the last 6 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7e2cb0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:7 out of the last 7 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf787e60> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:8 out of the last 8 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf797200> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:9 out of the last 9 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7488c0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:10 out of the last 10 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7dc5f0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7e2170> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf748710> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf756830> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf756560> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7b8830> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf787710> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf75b7a0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf76ac20> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7f8170> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf748ef0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf76ab00> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7777a0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf76aef0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf756680> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7f8cb0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf77b710> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7ed5f0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf748170> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf76af80> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf708320> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf718c20> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf777cb0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf75b3b0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf797f80> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf71b7a0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf71b9e0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7fb680> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf77b200> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7270e0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf733b90> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf718b00> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7b8b90> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf727290> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf73ac20> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7278c0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf73add0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf733dd0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf75b0e0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf6c4290> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7974d0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf71bb90> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7183b0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf6d7560> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf6d78c0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf718440> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf7774d0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf6e5290> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf6f2950> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf827950> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf73a950> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf6e5170> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf6fe9e0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf6fe7a0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "WARNING:tensorflow:11 out of the last 11 calls to <function recreate_function.<locals>.restored_function_body at 0x7f35bf6e5f80> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "Model: \"model\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "CDL2CROWS (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDL3BLACKCROWS (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDL3INSIDE (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDL3LINESTRIKE (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDL3OUTSIDE (InputLayer)        [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDL3STARSINSOUTH (InputLayer)   [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDL3WHITESOLDIERS (InputLayer)  [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLABANDONEDBABY (InputLayer)   [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLADVANCEBLOCK (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLBELTHOLD (InputLayer)        [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLBREAKAWAY (InputLayer)       [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLCLOSINGMARUBOZU (InputLayer) [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLCONCEALBABYSWALL (InputLayer [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLDARKCLOUDCOVER (InputLayer)  [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLDOJI (InputLayer)            [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLDOJISTAR (InputLayer)        [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLDRAGONFLYDOJI (InputLayer)   [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLENGULFING (InputLayer)       [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLEVENINGDOJISTAR (InputLayer) [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLEVENINGSTAR (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLGAPSIDESIDEWHITE (InputLayer [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLGRAVESTONEDOJI (InputLayer)  [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLHAMMER (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLHANGINGMAN (InputLayer)      [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLHARAMI (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLHARAMICROSS (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLHIGHWAVE (InputLayer)        [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLHIKKAKE (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLHIKKAKEMOD (InputLayer)      [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLHOMINGPIGEON (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLIDENTICAL3CROWS (InputLayer) [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLINNECK (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLINVERTEDHAMMER (InputLayer)  [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLKICKING (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLLADDERBOTTOM (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLLONGLEGGEDDOJI (InputLayer)  [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLMARUBOZU (InputLayer)        [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLMATCHINGLOW (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLMATHOLD (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLMORNINGDOJISTAR (InputLayer) [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLMORNINGSTAR (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLONNECK (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLPIERCING (InputLayer)        [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLRICKSHAWMAN (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLRISEFALL3METHODS (InputLayer [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLSEPARATINGLINES (InputLayer) [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLSHOOTINGSTAR (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLSPINNINGTOP (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLSTICKSANDWICH (InputLayer)   [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLTAKURI (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLTASUKIGAP (InputLayer)       [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLTHRUSTING (InputLayer)       [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLTRISTAR (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLUNIQUE3RIVER (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLUPSIDEGAP2CROWS (InputLayer) [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "CDLXSIDEGAP3METHODS (InputLayer [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_bbhi (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_bbli (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_kchi (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_kcli (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_psar_up_indicator (InputL [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_psar_down_indicator (Inpu [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "weekday10 (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "month10 (InputLayer)            [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volume_adi (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volume_obv (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volume_cmf (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volume_fi (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volume_mfi (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volume_em (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volume_sma_em (InputLayer)      [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volume_vpt (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volume_nvi (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volume_vwap (InputLayer)        [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_atr (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_bbm (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_bbh (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_bbl (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_bbw (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_bbp (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_kcc (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_kch (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_kcl (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_kcw (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_kcp (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_dcl (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_dch (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_dcm (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_dcw (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_dcp (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "volatility_ui (InputLayer)      [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_macd (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_macd_signal (InputLayer)  [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_macd_diff (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_sma_fast (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_sma_slow (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_ema_fast (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_ema_slow (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_adx (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_adx_pos (InputLayer)      [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_adx_neg (InputLayer)      [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_vortex_ind_pos (InputLaye [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_vortex_ind_neg (InputLaye [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_vortex_ind_diff (InputLay [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_trix (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_mass_index (InputLayer)   [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_cci (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_dpo (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_kst (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_kst_sig (InputLayer)      [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_kst_diff (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_ichimoku_conv (InputLayer [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_ichimoku_base (InputLayer [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_ichimoku_a (InputLayer)   [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_ichimoku_b (InputLayer)   [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_visual_ichimoku_a (InputL [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_visual_ichimoku_b (InputL [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_aroon_up (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_aroon_down (InputLayer)   [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_aroon_ind (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_psar_up (InputLayer)      [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_psar_down (InputLayer)    [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "trend_stc (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_rsi (InputLayer)       [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_stoch_rsi (InputLayer) [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_stoch_rsi_k (InputLaye [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_stoch_rsi_d (InputLaye [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_tsi (InputLayer)       [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_uo (InputLayer)        [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_stoch (InputLayer)     [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_stoch_signal (InputLay [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_wr (InputLayer)        [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_ao (InputLayer)        [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_kama (InputLayer)      [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_roc (InputLayer)       [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_ppo (InputLayer)       [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_ppo_signal (InputLayer [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "momentum_ppo_hist (InputLayer)  [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "others_dr (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "others_dlr (InputLayer)         [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "others_cr (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "Close_VIX (InputLayer)          [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "Spread (InputLayer)             [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "Rec_prob (InputLayer)           [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup (StringLookup)    (None, 1)            0           CDL2CROWS[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_1 (StringLookup)  (None, 1)            0           CDL3BLACKCROWS[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_2 (StringLookup)  (None, 1)            0           CDL3INSIDE[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_3 (StringLookup)  (None, 1)            0           CDL3LINESTRIKE[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_4 (StringLookup)  (None, 1)            0           CDL3OUTSIDE[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_5 (StringLookup)  (None, 1)            0           CDL3STARSINSOUTH[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_6 (StringLookup)  (None, 1)            0           CDL3WHITESOLDIERS[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_7 (StringLookup)  (None, 1)            0           CDLABANDONEDBABY[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_8 (StringLookup)  (None, 1)            0           CDLADVANCEBLOCK[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_9 (StringLookup)  (None, 1)            0           CDLBELTHOLD[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_10 (StringLookup) (None, 1)            0           CDLBREAKAWAY[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_11 (StringLookup) (None, 1)            0           CDLCLOSINGMARUBOZU[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_12 (StringLookup) (None, 1)            0           CDLCONCEALBABYSWALL[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_13 (StringLookup) (None, 1)            0           CDLDARKCLOUDCOVER[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_14 (StringLookup) (None, 1)            0           CDLDOJI[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_15 (StringLookup) (None, 1)            0           CDLDOJISTAR[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_16 (StringLookup) (None, 1)            0           CDLDRAGONFLYDOJI[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_17 (StringLookup) (None, 1)            0           CDLENGULFING[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_18 (StringLookup) (None, 1)            0           CDLEVENINGDOJISTAR[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_19 (StringLookup) (None, 1)            0           CDLEVENINGSTAR[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_20 (StringLookup) (None, 1)            0           CDLGAPSIDESIDEWHITE[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_21 (StringLookup) (None, 1)            0           CDLGRAVESTONEDOJI[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_22 (StringLookup) (None, 1)            0           CDLHAMMER[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_23 (StringLookup) (None, 1)            0           CDLHANGINGMAN[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_24 (StringLookup) (None, 1)            0           CDLHARAMI[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_25 (StringLookup) (None, 1)            0           CDLHARAMICROSS[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_26 (StringLookup) (None, 1)            0           CDLHIGHWAVE[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_27 (StringLookup) (None, 1)            0           CDLHIKKAKE[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_28 (StringLookup) (None, 1)            0           CDLHIKKAKEMOD[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_29 (StringLookup) (None, 1)            0           CDLHOMINGPIGEON[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_30 (StringLookup) (None, 1)            0           CDLIDENTICAL3CROWS[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_31 (StringLookup) (None, 1)            0           CDLINNECK[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_32 (StringLookup) (None, 1)            0           CDLINVERTEDHAMMER[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_33 (StringLookup) (None, 1)            0           CDLKICKING[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_34 (StringLookup) (None, 1)            0           CDLLADDERBOTTOM[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_35 (StringLookup) (None, 1)            0           CDLLONGLEGGEDDOJI[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_36 (StringLookup) (None, 1)            0           CDLMARUBOZU[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_37 (StringLookup) (None, 1)            0           CDLMATCHINGLOW[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_38 (StringLookup) (None, 1)            0           CDLMATHOLD[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_39 (StringLookup) (None, 1)            0           CDLMORNINGDOJISTAR[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_40 (StringLookup) (None, 1)            0           CDLMORNINGSTAR[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_41 (StringLookup) (None, 1)            0           CDLONNECK[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_42 (StringLookup) (None, 1)            0           CDLPIERCING[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_43 (StringLookup) (None, 1)            0           CDLRICKSHAWMAN[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_44 (StringLookup) (None, 1)            0           CDLRISEFALL3METHODS[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_45 (StringLookup) (None, 1)            0           CDLSEPARATINGLINES[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_46 (StringLookup) (None, 1)            0           CDLSHOOTINGSTAR[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_47 (StringLookup) (None, 1)            0           CDLSPINNINGTOP[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_48 (StringLookup) (None, 1)            0           CDLSTICKSANDWICH[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_49 (StringLookup) (None, 1)            0           CDLTAKURI[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_50 (StringLookup) (None, 1)            0           CDLTASUKIGAP[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_51 (StringLookup) (None, 1)            0           CDLTHRUSTING[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_52 (StringLookup) (None, 1)            0           CDLTRISTAR[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_53 (StringLookup) (None, 1)            0           CDLUNIQUE3RIVER[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_54 (StringLookup) (None, 1)            0           CDLUPSIDEGAP2CROWS[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_55 (StringLookup) (None, 1)            0           CDLXSIDEGAP3METHODS[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_56 (StringLookup) (None, 1)            0           volatility_bbhi[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_57 (StringLookup) (None, 1)            0           volatility_bbli[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_58 (StringLookup) (None, 1)            0           volatility_kchi[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_59 (StringLookup) (None, 1)            0           volatility_kcli[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_60 (StringLookup) (None, 1)            0           trend_psar_up_indicator[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_61 (StringLookup) (None, 1)            0           trend_psar_down_indicator[0][0]  \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_62 (StringLookup) (None, 1)            0           weekday10[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "string_lookup_63 (StringLookup) (None, 1)            0           month10[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "normalization (Normalization)   (None, 1)            3           volume_adi[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "normalization_1 (Normalization) (None, 1)            3           volume_obv[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "normalization_2 (Normalization) (None, 1)            3           volume_cmf[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "normalization_3 (Normalization) (None, 1)            3           volume_fi[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "normalization_4 (Normalization) (None, 1)            3           volume_mfi[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "normalization_5 (Normalization) (None, 1)            3           volume_em[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "normalization_6 (Normalization) (None, 1)            3           volume_sma_em[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "normalization_7 (Normalization) (None, 1)            3           volume_vpt[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "normalization_8 (Normalization) (None, 1)            3           volume_nvi[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "normalization_9 (Normalization) (None, 1)            3           volume_vwap[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "normalization_10 (Normalization (None, 1)            3           volatility_atr[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_11 (Normalization (None, 1)            3           volatility_bbm[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_12 (Normalization (None, 1)            3           volatility_bbh[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_13 (Normalization (None, 1)            3           volatility_bbl[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_14 (Normalization (None, 1)            3           volatility_bbw[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_15 (Normalization (None, 1)            3           volatility_bbp[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_16 (Normalization (None, 1)            3           volatility_kcc[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_17 (Normalization (None, 1)            3           volatility_kch[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_18 (Normalization (None, 1)            3           volatility_kcl[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_19 (Normalization (None, 1)            3           volatility_kcw[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_20 (Normalization (None, 1)            3           volatility_kcp[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_21 (Normalization (None, 1)            3           volatility_dcl[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_22 (Normalization (None, 1)            3           volatility_dch[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_23 (Normalization (None, 1)            3           volatility_dcm[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_24 (Normalization (None, 1)            3           volatility_dcw[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_25 (Normalization (None, 1)            3           volatility_dcp[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_26 (Normalization (None, 1)            3           volatility_ui[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "normalization_27 (Normalization (None, 1)            3           trend_macd[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "normalization_28 (Normalization (None, 1)            3           trend_macd_signal[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "normalization_29 (Normalization (None, 1)            3           trend_macd_diff[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "normalization_30 (Normalization (None, 1)            3           trend_sma_fast[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_31 (Normalization (None, 1)            3           trend_sma_slow[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_32 (Normalization (None, 1)            3           trend_ema_fast[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_33 (Normalization (None, 1)            3           trend_ema_slow[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_34 (Normalization (None, 1)            3           trend_adx[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "normalization_35 (Normalization (None, 1)            3           trend_adx_pos[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "normalization_36 (Normalization (None, 1)            3           trend_adx_neg[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "normalization_37 (Normalization (None, 1)            3           trend_vortex_ind_pos[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "normalization_38 (Normalization (None, 1)            3           trend_vortex_ind_neg[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "normalization_39 (Normalization (None, 1)            3           trend_vortex_ind_diff[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "normalization_40 (Normalization (None, 1)            3           trend_trix[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "normalization_41 (Normalization (None, 1)            3           trend_mass_index[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "normalization_42 (Normalization (None, 1)            3           trend_cci[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "normalization_43 (Normalization (None, 1)            3           trend_dpo[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "normalization_44 (Normalization (None, 1)            3           trend_kst[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "normalization_45 (Normalization (None, 1)            3           trend_kst_sig[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "normalization_46 (Normalization (None, 1)            3           trend_kst_diff[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_47 (Normalization (None, 1)            3           trend_ichimoku_conv[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "normalization_48 (Normalization (None, 1)            3           trend_ichimoku_base[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "normalization_49 (Normalization (None, 1)            3           trend_ichimoku_a[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "normalization_50 (Normalization (None, 1)            3           trend_ichimoku_b[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "normalization_51 (Normalization (None, 1)            3           trend_visual_ichimoku_a[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "normalization_52 (Normalization (None, 1)            3           trend_visual_ichimoku_b[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "normalization_53 (Normalization (None, 1)            3           trend_aroon_up[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_54 (Normalization (None, 1)            3           trend_aroon_down[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "normalization_55 (Normalization (None, 1)            3           trend_aroon_ind[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "normalization_56 (Normalization (None, 1)            3           trend_psar_up[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "normalization_57 (Normalization (None, 1)            3           trend_psar_down[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "normalization_58 (Normalization (None, 1)            3           trend_stc[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "normalization_59 (Normalization (None, 1)            3           momentum_rsi[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "normalization_60 (Normalization (None, 1)            3           momentum_stoch_rsi[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "normalization_61 (Normalization (None, 1)            3           momentum_stoch_rsi_k[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "normalization_62 (Normalization (None, 1)            3           momentum_stoch_rsi_d[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "normalization_63 (Normalization (None, 1)            3           momentum_tsi[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "normalization_64 (Normalization (None, 1)            3           momentum_uo[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "normalization_65 (Normalization (None, 1)            3           momentum_stoch[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "normalization_66 (Normalization (None, 1)            3           momentum_stoch_signal[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "normalization_67 (Normalization (None, 1)            3           momentum_wr[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "normalization_68 (Normalization (None, 1)            3           momentum_ao[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "normalization_69 (Normalization (None, 1)            3           momentum_kama[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "normalization_70 (Normalization (None, 1)            3           momentum_roc[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "normalization_71 (Normalization (None, 1)            3           momentum_ppo[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "normalization_72 (Normalization (None, 1)            3           momentum_ppo_signal[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "normalization_73 (Normalization (None, 1)            3           momentum_ppo_hist[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "normalization_74 (Normalization (None, 1)            3           others_dr[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "normalization_75 (Normalization (None, 1)            3           others_dlr[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "normalization_76 (Normalization (None, 1)            3           others_cr[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "normalization_77 (Normalization (None, 1)            3           Close_VIX[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "normalization_78 (Normalization (None, 1)            3           Spread[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "normalization_79 (Normalization (None, 1)            3           Rec_prob[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding (CategoryEnco (None, 3)            0           string_lookup[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_1 (CategoryEn (None, 3)            0           string_lookup_1[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_2 (CategoryEn (None, 5)            0           string_lookup_2[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_3 (CategoryEn (None, 5)            0           string_lookup_3[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_4 (CategoryEn (None, 5)            0           string_lookup_4[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_5 (CategoryEn (None, 3)            0           string_lookup_5[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_6 (CategoryEn (None, 4)            0           string_lookup_6[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_7 (CategoryEn (None, 3)            0           string_lookup_7[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_8 (CategoryEn (None, 4)            0           string_lookup_8[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_9 (CategoryEn (None, 5)            0           string_lookup_9[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_10 (CategoryE (None, 3)            0           string_lookup_10[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_11 (CategoryE (None, 5)            0           string_lookup_11[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_12 (CategoryE (None, 3)            0           string_lookup_12[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_13 (CategoryE (None, 4)            0           string_lookup_13[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_14 (CategoryE (None, 4)            0           string_lookup_14[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_15 (CategoryE (None, 5)            0           string_lookup_15[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_16 (CategoryE (None, 4)            0           string_lookup_16[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_17 (CategoryE (None, 5)            0           string_lookup_17[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_18 (CategoryE (None, 4)            0           string_lookup_18[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_19 (CategoryE (None, 4)            0           string_lookup_19[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_20 (CategoryE (None, 4)            0           string_lookup_20[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_21 (CategoryE (None, 4)            0           string_lookup_21[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_22 (CategoryE (None, 4)            0           string_lookup_22[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_23 (CategoryE (None, 4)            0           string_lookup_23[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_24 (CategoryE (None, 5)            0           string_lookup_24[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_25 (CategoryE (None, 5)            0           string_lookup_25[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_26 (CategoryE (None, 5)            0           string_lookup_26[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_27 (CategoryE (None, 5)            0           string_lookup_27[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_28 (CategoryE (None, 5)            0           string_lookup_28[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_29 (CategoryE (None, 4)            0           string_lookup_29[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_30 (CategoryE (None, 4)            0           string_lookup_30[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_31 (CategoryE (None, 4)            0           string_lookup_31[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_32 (CategoryE (None, 4)            0           string_lookup_32[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_33 (CategoryE (None, 3)            0           string_lookup_33[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_34 (CategoryE (None, 4)            0           string_lookup_34[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_35 (CategoryE (None, 4)            0           string_lookup_35[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_36 (CategoryE (None, 5)            0           string_lookup_36[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_37 (CategoryE (None, 4)            0           string_lookup_37[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_38 (CategoryE (None, 3)            0           string_lookup_38[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_39 (CategoryE (None, 4)            0           string_lookup_39[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_40 (CategoryE (None, 4)            0           string_lookup_40[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_41 (CategoryE (None, 4)            0           string_lookup_41[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_42 (CategoryE (None, 4)            0           string_lookup_42[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_43 (CategoryE (None, 4)            0           string_lookup_43[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_44 (CategoryE (None, 4)            0           string_lookup_44[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_45 (CategoryE (None, 5)            0           string_lookup_45[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_46 (CategoryE (None, 4)            0           string_lookup_46[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_47 (CategoryE (None, 5)            0           string_lookup_47[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_48 (CategoryE (None, 4)            0           string_lookup_48[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_49 (CategoryE (None, 4)            0           string_lookup_49[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_50 (CategoryE (None, 5)            0           string_lookup_50[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_51 (CategoryE (None, 4)            0           string_lookup_51[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_52 (CategoryE (None, 5)            0           string_lookup_52[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_53 (CategoryE (None, 4)            0           string_lookup_53[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_54 (CategoryE (None, 4)            0           string_lookup_54[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_55 (CategoryE (None, 5)            0           string_lookup_55[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_56 (CategoryE (None, 4)            0           string_lookup_56[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_57 (CategoryE (None, 4)            0           string_lookup_57[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_58 (CategoryE (None, 4)            0           string_lookup_58[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_59 (CategoryE (None, 4)            0           string_lookup_59[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_60 (CategoryE (None, 4)            0           string_lookup_60[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_61 (CategoryE (None, 4)            0           string_lookup_61[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_62 (CategoryE (None, 5)            0           string_lookup_62[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "category_encoding_63 (CategoryE (None, 5)            0           string_lookup_63[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "concatenate (Concatenate)       (None, 348)          0           normalization[0][0]              \n",
            "                                                                 normalization_1[0][0]            \n",
            "                                                                 normalization_2[0][0]            \n",
            "                                                                 normalization_3[0][0]            \n",
            "                                                                 normalization_4[0][0]            \n",
            "                                                                 normalization_5[0][0]            \n",
            "                                                                 normalization_6[0][0]            \n",
            "                                                                 normalization_7[0][0]            \n",
            "                                                                 normalization_8[0][0]            \n",
            "                                                                 normalization_9[0][0]            \n",
            "                                                                 normalization_10[0][0]           \n",
            "                                                                 normalization_11[0][0]           \n",
            "                                                                 normalization_12[0][0]           \n",
            "                                                                 normalization_13[0][0]           \n",
            "                                                                 normalization_14[0][0]           \n",
            "                                                                 normalization_15[0][0]           \n",
            "                                                                 normalization_16[0][0]           \n",
            "                                                                 normalization_17[0][0]           \n",
            "                                                                 normalization_18[0][0]           \n",
            "                                                                 normalization_19[0][0]           \n",
            "                                                                 normalization_20[0][0]           \n",
            "                                                                 normalization_21[0][0]           \n",
            "                                                                 normalization_22[0][0]           \n",
            "                                                                 normalization_23[0][0]           \n",
            "                                                                 normalization_24[0][0]           \n",
            "                                                                 normalization_25[0][0]           \n",
            "                                                                 normalization_26[0][0]           \n",
            "                                                                 normalization_27[0][0]           \n",
            "                                                                 normalization_28[0][0]           \n",
            "                                                                 normalization_29[0][0]           \n",
            "                                                                 normalization_30[0][0]           \n",
            "                                                                 normalization_31[0][0]           \n",
            "                                                                 normalization_32[0][0]           \n",
            "                                                                 normalization_33[0][0]           \n",
            "                                                                 normalization_34[0][0]           \n",
            "                                                                 normalization_35[0][0]           \n",
            "                                                                 normalization_36[0][0]           \n",
            "                                                                 normalization_37[0][0]           \n",
            "                                                                 normalization_38[0][0]           \n",
            "                                                                 normalization_39[0][0]           \n",
            "                                                                 normalization_40[0][0]           \n",
            "                                                                 normalization_41[0][0]           \n",
            "                                                                 normalization_42[0][0]           \n",
            "                                                                 normalization_43[0][0]           \n",
            "                                                                 normalization_44[0][0]           \n",
            "                                                                 normalization_45[0][0]           \n",
            "                                                                 normalization_46[0][0]           \n",
            "                                                                 normalization_47[0][0]           \n",
            "                                                                 normalization_48[0][0]           \n",
            "                                                                 normalization_49[0][0]           \n",
            "                                                                 normalization_50[0][0]           \n",
            "                                                                 normalization_51[0][0]           \n",
            "                                                                 normalization_52[0][0]           \n",
            "                                                                 normalization_53[0][0]           \n",
            "                                                                 normalization_54[0][0]           \n",
            "                                                                 normalization_55[0][0]           \n",
            "                                                                 normalization_56[0][0]           \n",
            "                                                                 normalization_57[0][0]           \n",
            "                                                                 normalization_58[0][0]           \n",
            "                                                                 normalization_59[0][0]           \n",
            "                                                                 normalization_60[0][0]           \n",
            "                                                                 normalization_61[0][0]           \n",
            "                                                                 normalization_62[0][0]           \n",
            "                                                                 normalization_63[0][0]           \n",
            "                                                                 normalization_64[0][0]           \n",
            "                                                                 normalization_65[0][0]           \n",
            "                                                                 normalization_66[0][0]           \n",
            "                                                                 normalization_67[0][0]           \n",
            "                                                                 normalization_68[0][0]           \n",
            "                                                                 normalization_69[0][0]           \n",
            "                                                                 normalization_70[0][0]           \n",
            "                                                                 normalization_71[0][0]           \n",
            "                                                                 normalization_72[0][0]           \n",
            "                                                                 normalization_73[0][0]           \n",
            "                                                                 normalization_74[0][0]           \n",
            "                                                                 normalization_75[0][0]           \n",
            "                                                                 normalization_76[0][0]           \n",
            "                                                                 normalization_77[0][0]           \n",
            "                                                                 normalization_78[0][0]           \n",
            "                                                                 normalization_79[0][0]           \n",
            "                                                                 category_encoding[0][0]          \n",
            "                                                                 category_encoding_1[0][0]        \n",
            "                                                                 category_encoding_2[0][0]        \n",
            "                                                                 category_encoding_3[0][0]        \n",
            "                                                                 category_encoding_4[0][0]        \n",
            "                                                                 category_encoding_5[0][0]        \n",
            "                                                                 category_encoding_6[0][0]        \n",
            "                                                                 category_encoding_7[0][0]        \n",
            "                                                                 category_encoding_8[0][0]        \n",
            "                                                                 category_encoding_9[0][0]        \n",
            "                                                                 category_encoding_10[0][0]       \n",
            "                                                                 category_encoding_11[0][0]       \n",
            "                                                                 category_encoding_12[0][0]       \n",
            "                                                                 category_encoding_13[0][0]       \n",
            "                                                                 category_encoding_14[0][0]       \n",
            "                                                                 category_encoding_15[0][0]       \n",
            "                                                                 category_encoding_16[0][0]       \n",
            "                                                                 category_encoding_17[0][0]       \n",
            "                                                                 category_encoding_18[0][0]       \n",
            "                                                                 category_encoding_19[0][0]       \n",
            "                                                                 category_encoding_20[0][0]       \n",
            "                                                                 category_encoding_21[0][0]       \n",
            "                                                                 category_encoding_22[0][0]       \n",
            "                                                                 category_encoding_23[0][0]       \n",
            "                                                                 category_encoding_24[0][0]       \n",
            "                                                                 category_encoding_25[0][0]       \n",
            "                                                                 category_encoding_26[0][0]       \n",
            "                                                                 category_encoding_27[0][0]       \n",
            "                                                                 category_encoding_28[0][0]       \n",
            "                                                                 category_encoding_29[0][0]       \n",
            "                                                                 category_encoding_30[0][0]       \n",
            "                                                                 category_encoding_31[0][0]       \n",
            "                                                                 category_encoding_32[0][0]       \n",
            "                                                                 category_encoding_33[0][0]       \n",
            "                                                                 category_encoding_34[0][0]       \n",
            "                                                                 category_encoding_35[0][0]       \n",
            "                                                                 category_encoding_36[0][0]       \n",
            "                                                                 category_encoding_37[0][0]       \n",
            "                                                                 category_encoding_38[0][0]       \n",
            "                                                                 category_encoding_39[0][0]       \n",
            "                                                                 category_encoding_40[0][0]       \n",
            "                                                                 category_encoding_41[0][0]       \n",
            "                                                                 category_encoding_42[0][0]       \n",
            "                                                                 category_encoding_43[0][0]       \n",
            "                                                                 category_encoding_44[0][0]       \n",
            "                                                                 category_encoding_45[0][0]       \n",
            "                                                                 category_encoding_46[0][0]       \n",
            "                                                                 category_encoding_47[0][0]       \n",
            "                                                                 category_encoding_48[0][0]       \n",
            "                                                                 category_encoding_49[0][0]       \n",
            "                                                                 category_encoding_50[0][0]       \n",
            "                                                                 category_encoding_51[0][0]       \n",
            "                                                                 category_encoding_52[0][0]       \n",
            "                                                                 category_encoding_53[0][0]       \n",
            "                                                                 category_encoding_54[0][0]       \n",
            "                                                                 category_encoding_55[0][0]       \n",
            "                                                                 category_encoding_56[0][0]       \n",
            "                                                                 category_encoding_57[0][0]       \n",
            "                                                                 category_encoding_58[0][0]       \n",
            "                                                                 category_encoding_59[0][0]       \n",
            "                                                                 category_encoding_60[0][0]       \n",
            "                                                                 category_encoding_61[0][0]       \n",
            "                                                                 category_encoding_62[0][0]       \n",
            "                                                                 category_encoding_63[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "dense (Dense)                   (None, 480)          167520      concatenate[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "dense_1 (Dense)                 (None, 704)          338624      dense[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "dense_2 (Dense)                 (None, 160)          112800      dense_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dropout (Dropout)               (None, 160)          0           dense_2[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dense_3 (Dense)                 (None, 3)            483         dropout[0][0]                    \n",
            "==================================================================================================\n",
            "Total params: 619,667\n",
            "Trainable params: 619,427\n",
            "Non-trainable params: 240\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "csZXd8c34_HL",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "2ea3e1b4-35ab-43b3-ec9a-da940d5ce5c8"
      },
      "source": [
        "from pandas.tseries.offsets import *\n",
        "BDay()\n",
        "# #create rolling windows dataset\n",
        "n_train = 5*(52) # training period \n",
        "val_size = 10 # validation period (next n days)\n",
        "window_shift = 10 # period for re-train model\n",
        "n_records = len(raw_df[raw_df['symbol']=='^GSPC'])\n",
        "fq = 'B'\n",
        "batch_size = 32\n",
        "num_category = len(TARGET)\n",
        "ts_date= []\n",
        "te_date = []\n",
        "vt_date = []\n",
        "ve_date = []\n",
        "td_acc = []\n",
        "t_acc = []\n",
        "v_acc = []\n",
        "t_F1 = []\n",
        "td_F1 = []\n",
        "v_F1 = []\n",
        "y_true = []\n",
        "y_pred = pd.DataFrame({})\n",
        "\n",
        "\n",
        "train_start =  pd.Timestamp('2016-01-02')\n",
        "for j, i in enumerate(np.arange(1, 255, window_shift)):\n",
        "    #split train and test\n",
        "    \n",
        "    train_range = pd.date_range(train_start, periods=n_train, freq=fq)\n",
        "    test_start = train_range[-1]+BDay()\n",
        "    test_range = pd.date_range(test_start, periods=val_size, freq=fq)\n",
        "    \n",
        "    train_df, val_df = raw_df[raw_df['date'].isin(train_range)], raw_df[raw_df['date'].isin(test_range)]\n",
        "    train_df[CONT_NAMES] = train_df[CONT_NAMES].astype('float32')\n",
        "    val_df[CONT_NAMES] = val_df[CONT_NAMES].astype('float32')\n",
        "    \n",
        "    #y_train, y_val = raw_df[raw_df['date'].isin(train_range)][TARGET], raw_df[raw_df['date'].isin(test_range)][TARGET]\n",
        "    print(i, 'Training window: {} - {}'.format(train_range[0], train_range[-1]))\n",
        "    if len(val_df)==0:\n",
        "        # train_start = train_range[0]+BDay()\n",
        "        n_train = n_train + window_shift\n",
        "        continue\n",
        "    #print('train=%d, test=%d' % (len(train_df), len(test_df)))\n",
        "    temp_train = train_df.iloc[:, :len(CAT_NAMES)+len(CONT_NAMES)].reset_index()\n",
        "    temp_train = train_df.loc[:, COL].reset_index()\n",
        "    temp_train = temp_train.drop(['index'], axis=1)\n",
        "    temp_val = val_df.iloc[:, :len(CAT_NAMES)+len(CONT_NAMES)].reset_index()\n",
        "    temp_val = val_df.loc[:, COL].reset_index()\n",
        "    temp_val = temp_val.drop(['index'], axis=1)\n",
        "    \n",
        "    train1_ds = tf_dataset(temp_train, train_df[TARGET], batch_size, True)\n",
        "    train_ds = train1_ds.take(int(len(train_df)/batch_size*0.9))\n",
        "    # # train_dev dataset for validate variance of model\n",
        "    train_dev_ds = train1_ds.skip(int(len(train_df)/batch_size*0.9))\n",
        "    val_ds = tf_dataset(temp_val, val_df[TARGET], batch_size, False)\n",
        "\n",
        "#     train1_ds = tf_dataset(train_df.loc[:, Input_col[:-1]], train_df[TARGET], batch_size, True)\n",
        "#     train_ds = train1_ds.take(int(len(train_df)/batch_size*0.9))\n",
        "#     # # train_dev dataset for validate variance of model\n",
        "#     train_dev_ds = train1_ds.skip(int(len(train_df)/batch_size*0.9))\n",
        "#     val_ds = tf_dataset(val_df.loc[:, Input_col[:-1]], val_df[TARGET], batch_size, False)\n",
        "    \n",
        "    #allinputs, encodedfeatures = create_fea_col(train_ds)\n",
        "\n",
        "    #model training\n",
        "    #model1, history = train_model(train_ds, train_dev_ds, 200, allinputs, encodedfeatures)\n",
        "    #model1 =  train_model(train_ds, val_ds, 50, encoded_features)\n",
        "    \n",
        "    loss, acc, F1 = model1.evaluate(train_ds, verbose=0)\n",
        "    print(i, 'Model train accuracy: {:5.2f}% , F1: {:5.2f}'.format(100 * acc, F1))\n",
        "\n",
        "    \n",
        "    #model validation\n",
        "    loss, acc, F1= model1.evaluate(train_dev_ds, verbose=0)\n",
        "    print(i, 'Model train-dev accuracy: {:5.2f}% , F1: {:5.2f}'.format(100 * acc, F1))\n",
        "\n",
        "    \n",
        "    print('Val window: {} - {}'.format(test_range[0], test_range[-1]))\n",
        "\n",
        "    loss, acc, F1 = model1.evaluate(val_ds, verbose=0)\n",
        "    y_pred = y_pred.append(pd.DataFrame(model1.predict(val_ds)), ignore_index=True, sort=False)\n",
        "    #print(y_pred)\n",
        "    \n",
        "    for x, y in val_ds:\n",
        "        for k in y:\n",
        "            y_true.append(k.numpy())\n",
        "            t_acc.append(acc)\n",
        "            t_F1.append(F1)\n",
        "            td_acc.append(acc)\n",
        "            td_F1.append(F1)\n",
        "            v_acc.append(acc)\n",
        "            v_F1.append(F1)\n",
        "            ts_date.append(train_start)\n",
        "            te_date.append(train_range[-1])\n",
        "            vt_date.append(test_start)\n",
        "            ve_date.append(test_range[-1])\n",
        "            test_start = test_start + BDay()\n",
        "\n",
        "    print(i, 'Model val accuracy: {:5.2f}% , F1: {:5.2f} \\n'.format(100 * acc, F1))\n",
        "    \n",
        "    \n",
        "    # train_start = train_range[0]+BDay(window_shift)\n",
        "    n_train = n_train + window_shift"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "1 Training window: 2016-01-04 00:00:00 - 2016-12-30 00:00:00\n",
            "1 Model train accuracy: 89.80% , F1:  0.79\n",
            "1 Model train-dev accuracy: 90.00% , F1:  0.64\n",
            "Val window: 2017-01-02 00:00:00 - 2017-01-13 00:00:00\n",
            "1 Model val accuracy: 62.50% , F1:  0.56 \n",
            "\n",
            "11 Training window: 2016-01-04 00:00:00 - 2017-01-13 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "11 Model train accuracy: 87.76% , F1:  0.77\n",
            "11 Model train-dev accuracy: 94.12% , F1:  0.86\n",
            "Val window: 2017-01-16 00:00:00 - 2017-01-27 00:00:00\n",
            "11 Model val accuracy: 66.67% , F1:  0.22 \n",
            "\n",
            "21 Training window: 2016-01-04 00:00:00 - 2017-01-27 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "21 Model train accuracy: 89.53% , F1:  0.76\n",
            "21 Model train-dev accuracy: 81.58% , F1:  0.67\n",
            "Val window: 2017-01-30 00:00:00 - 2017-02-10 00:00:00\n",
            "21 Model val accuracy: 90.00% , F1:  0.90 \n",
            "\n",
            "31 Training window: 2016-01-04 00:00:00 - 2017-02-10 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "31 Model train accuracy: 86.32% , F1:  0.73\n",
            "31 Model train-dev accuracy: 90.20% , F1:  0.82\n",
            "Val window: 2017-02-13 00:00:00 - 2017-02-24 00:00:00\n",
            "31 Model val accuracy: 55.56% , F1:  0.56 \n",
            "\n",
            "41 Training window: 2016-01-04 00:00:00 - 2017-02-24 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "41 Model train accuracy: 87.84% , F1:  0.76\n",
            "41 Model train-dev accuracy: 89.29% , F1:  0.76\n",
            "Val window: 2017-02-27 00:00:00 - 2017-03-10 00:00:00\n",
            "41 Model val accuracy: 90.00% , F1:  0.90 \n",
            "\n",
            "51 Training window: 2016-01-04 00:00:00 - 2017-03-10 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "51 Model train accuracy: 85.97% , F1:  0.74\n",
            "51 Model train-dev accuracy: 89.19% , F1:  0.77\n",
            "Val window: 2017-03-13 00:00:00 - 2017-03-24 00:00:00\n",
            "51 Model val accuracy: 88.89% , F1:  0.80 \n",
            "\n",
            "61 Training window: 2016-01-04 00:00:00 - 2017-03-24 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "61 Model train accuracy: 85.33% , F1:  0.75\n",
            "61 Model train-dev accuracy: 84.00% , F1:  0.79\n",
            "Val window: 2017-03-27 00:00:00 - 2017-04-07 00:00:00\n",
            "61 Model val accuracy: 70.00% , F1:  0.70 \n",
            "\n",
            "71 Training window: 2016-01-04 00:00:00 - 2017-04-07 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "71 Model train accuracy: 85.46% , F1:  0.76\n",
            "71 Model train-dev accuracy: 83.02% , F1:  0.70\n",
            "Val window: 2017-04-10 00:00:00 - 2017-04-21 00:00:00\n",
            "71 Model val accuracy: 100.00% , F1:  1.00 \n",
            "\n",
            "81 Training window: 2016-01-04 00:00:00 - 2017-04-21 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "81 Model train accuracy: 85.77% , F1:  0.75\n",
            "81 Model train-dev accuracy: 72.22% , F1:  0.65\n",
            "Val window: 2017-04-24 00:00:00 - 2017-05-05 00:00:00\n",
            "81 Model val accuracy: 100.00% , F1:  0.70 \n",
            "\n",
            "91 Training window: 2016-01-04 00:00:00 - 2017-05-05 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "91 Model train accuracy: 86.40% , F1:  0.75\n",
            "91 Model train-dev accuracy: 82.22% , F1:  0.74\n",
            "Val window: 2017-05-08 00:00:00 - 2017-05-19 00:00:00\n",
            "91 Model val accuracy: 70.00% , F1:  0.70 \n",
            "\n",
            "101 Training window: 2016-01-04 00:00:00 - 2017-05-19 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "101 Model train accuracy: 87.75% , F1:  0.77\n",
            "101 Model train-dev accuracy: 82.00% , F1:  0.68\n",
            "Val window: 2017-05-22 00:00:00 - 2017-06-02 00:00:00\n",
            "101 Model val accuracy: 88.89% , F1:  0.89 \n",
            "\n",
            "111 Training window: 2016-01-04 00:00:00 - 2017-06-02 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "111 Model train accuracy: 87.54% , F1:  0.77\n",
            "111 Model train-dev accuracy: 83.33% , F1:  0.81\n",
            "Val window: 2017-06-05 00:00:00 - 2017-06-16 00:00:00\n",
            "111 Model val accuracy: 62.50% , F1:  0.50 \n",
            "\n",
            "121 Training window: 2016-01-04 00:00:00 - 2017-06-16 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "121 Model train accuracy: 85.82% , F1:  0.76\n",
            "121 Model train-dev accuracy: 89.74% , F1:  0.74\n",
            "Val window: 2017-06-19 00:00:00 - 2017-06-30 00:00:00\n",
            "121 Model val accuracy: 70.00% , F1:  0.70 \n",
            "\n",
            "131 Training window: 2016-01-04 00:00:00 - 2017-06-30 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "131 Model train accuracy: 86.69% , F1:  0.75\n",
            "131 Model train-dev accuracy: 87.76% , F1:  0.75\n",
            "Val window: 2017-07-03 00:00:00 - 2017-07-14 00:00:00\n",
            "131 Model val accuracy: 100.00% , F1:  1.00 \n",
            "\n",
            "141 Training window: 2016-01-04 00:00:00 - 2017-07-14 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "141 Model train accuracy: 85.41% , F1:  0.75\n",
            "141 Model train-dev accuracy: 81.82% , F1:  0.68\n",
            "Val window: 2017-07-17 00:00:00 - 2017-07-28 00:00:00\n",
            "141 Model val accuracy: 55.56% , F1:  0.50 \n",
            "\n",
            "151 Training window: 2016-01-04 00:00:00 - 2017-07-28 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "151 Model train accuracy: 84.89% , F1:  0.75\n",
            "151 Model train-dev accuracy: 87.18% , F1:  0.77\n",
            "Val window: 2017-07-31 00:00:00 - 2017-08-11 00:00:00\n",
            "151 Model val accuracy: 90.00% , F1:  0.90 \n",
            "\n",
            "161 Training window: 2016-01-04 00:00:00 - 2017-08-11 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "161 Model train accuracy: 86.44% , F1:  0.78\n",
            "161 Model train-dev accuracy: 91.49% , F1:  0.80\n",
            "Val window: 2017-08-14 00:00:00 - 2017-08-25 00:00:00\n",
            "161 Model val accuracy: 100.00% , F1:  1.00 \n",
            "\n",
            "171 Training window: 2016-01-04 00:00:00 - 2017-08-25 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "171 Model train accuracy: 85.30% , F1:  0.76\n",
            "171 Model train-dev accuracy: 83.33% , F1:  0.62\n",
            "Val window: 2017-08-28 00:00:00 - 2017-09-08 00:00:00\n",
            "171 Model val accuracy: 100.00% , F1:  1.00 \n",
            "\n",
            "181 Training window: 2016-01-04 00:00:00 - 2017-09-08 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "181 Model train accuracy: 86.94% , F1:  0.78\n",
            "181 Model train-dev accuracy: 84.13% , F1:  0.73\n",
            "Val window: 2017-09-11 00:00:00 - 2017-09-22 00:00:00\n",
            "181 Model val accuracy: 90.00% , F1:  0.90 \n",
            "\n",
            "191 Training window: 2016-01-04 00:00:00 - 2017-09-22 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "191 Model train accuracy: 86.59% , F1:  0.77\n",
            "191 Model train-dev accuracy: 90.91% , F1:  0.78\n",
            "Val window: 2017-09-25 00:00:00 - 2017-10-06 00:00:00\n",
            "191 Model val accuracy: 100.00% , F1:  0.90 \n",
            "\n",
            "201 Training window: 2016-01-04 00:00:00 - 2017-10-06 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "201 Model train accuracy: 85.13% , F1:  0.76\n",
            "201 Model train-dev accuracy: 81.82% , F1:  0.74\n",
            "Val window: 2017-10-09 00:00:00 - 2017-10-20 00:00:00\n",
            "201 Model val accuracy: 50.00% , F1:  0.50 \n",
            "\n",
            "211 Training window: 2016-01-04 00:00:00 - 2017-10-20 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "211 Model train accuracy: 85.96% , F1:  0.77\n",
            "211 Model train-dev accuracy: 84.62% , F1:  0.77\n",
            "Val window: 2017-10-23 00:00:00 - 2017-11-03 00:00:00\n",
            "211 Model val accuracy: 70.00% , F1:  0.70 \n",
            "\n",
            "221 Training window: 2016-01-04 00:00:00 - 2017-11-03 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "221 Model train accuracy: 84.37% , F1:  0.75\n",
            "221 Model train-dev accuracy: 87.50% , F1:  0.71\n",
            "Val window: 2017-11-06 00:00:00 - 2017-11-17 00:00:00\n",
            "221 Model val accuracy: 60.00% , F1:  0.60 \n",
            "\n",
            "231 Training window: 2016-01-04 00:00:00 - 2017-11-17 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "231 Model train accuracy: 84.13% , F1:  0.76\n",
            "231 Model train-dev accuracy: 81.13% , F1:  0.73\n",
            "Val window: 2017-11-20 00:00:00 - 2017-12-01 00:00:00\n",
            "231 Model val accuracy: 88.89% , F1:  0.89 \n",
            "\n",
            "241 Training window: 2016-01-04 00:00:00 - 2017-12-01 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "241 Model train accuracy: 84.35% , F1:  0.76\n",
            "241 Model train-dev accuracy: 81.36% , F1:  0.71\n",
            "Val window: 2017-12-04 00:00:00 - 2017-12-15 00:00:00\n",
            "241 Model val accuracy: 100.00% , F1:  1.00 \n",
            "\n",
            "251 Training window: 2016-01-04 00:00:00 - 2017-12-15 00:00:00\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/pandas/core/frame.py:3069: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  self[k1] = value[k2]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "251 Model train accuracy: 85.60% , F1:  0.77\n",
            "251 Model train-dev accuracy: 83.58% , F1:  0.72\n",
            "Val window: 2017-12-18 00:00:00 - 2017-12-29 00:00:00\n",
            "251 Model val accuracy: 100.00% , F1:  1.00 \n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7p4cUzwd9TSf"
      },
      "source": [
        "thres = 0.5\n",
        "\n",
        "a = pd.DataFrame(ts_date, columns=['train start date'])\n",
        "b = pd.DataFrame(te_date, columns=['train end date'])\n",
        "c = pd.DataFrame(t_acc, columns=['train accuracy'])\n",
        "d = pd.DataFrame(vt_date, columns=['val start date'])\n",
        "e = pd.DataFrame(ve_date, columns=['val end date'])\n",
        "f = pd.DataFrame(v_acc, columns=['val accuracy'])\n",
        "g = pd.DataFrame(td_acc, columns=['train-dev accuracy'])\n",
        "h = pd.DataFrame(y_true, columns=['true_' + sub for sub in TARGET])\n",
        "i = y_pred\n",
        "i.columns = ['pred_-ve', 'pred_no', 'pred_+ve']\n",
        "\n",
        "out_df = pd.concat([a, b, c, g, d, e, f, h, i], axis=1)\n",
        "out_df['true(+/-)'] = out_df.apply(lambda x: '+ve' if x['true_positive']==1 else ('-ve' if x['true_negative']==1 else'no') , axis=1)\n",
        "out_df['pred(+/-)']  = out_df.apply(lambda x: '+ve' if x['pred_+ve']>thres else ('-ve' if x['pred_-ve']>thres else'no'), axis=1)\n",
        "\n",
        "out_df.tail(20)\n",
        "\n",
        "out_df.to_excel('output.xls')"
      ],
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uPafhLtWjnC8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 157
        },
        "outputId": "42ab3f7c-45d1-4318-d223-ff2a31550212"
      },
      "source": [
        "out_df.head(2)"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>train start date</th>\n",
              "      <th>train end date</th>\n",
              "      <th>train accuracy</th>\n",
              "      <th>train-dev accuracy</th>\n",
              "      <th>val start date</th>\n",
              "      <th>val end date</th>\n",
              "      <th>val accuracy</th>\n",
              "      <th>true_negative</th>\n",
              "      <th>true_no</th>\n",
              "      <th>true_positive</th>\n",
              "      <th>pred_-ve</th>\n",
              "      <th>pred_no</th>\n",
              "      <th>pred_+ve</th>\n",
              "      <th>true(+/-)</th>\n",
              "      <th>pred(+/-)</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2016-01-02</td>\n",
              "      <td>2016-12-30</td>\n",
              "      <td>0.625</td>\n",
              "      <td>0.625</td>\n",
              "      <td>2017-01-02</td>\n",
              "      <td>2017-01-13</td>\n",
              "      <td>0.625</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0.225112</td>\n",
              "      <td>0.035607</td>\n",
              "      <td>0.739281</td>\n",
              "      <td>+ve</td>\n",
              "      <td>+ve</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2016-01-02</td>\n",
              "      <td>2016-12-30</td>\n",
              "      <td>0.625</td>\n",
              "      <td>0.625</td>\n",
              "      <td>2017-01-03</td>\n",
              "      <td>2017-01-13</td>\n",
              "      <td>0.625</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.312698</td>\n",
              "      <td>0.103336</td>\n",
              "      <td>0.583965</td>\n",
              "      <td>-ve</td>\n",
              "      <td>+ve</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "  train start date train end date  train accuracy  ...  pred_+ve true(+/-) pred(+/-)\n",
              "0       2016-01-02     2016-12-30           0.625  ...  0.739281       +ve       +ve\n",
              "1       2016-01-02     2016-12-30           0.625  ...  0.583965       -ve       +ve\n",
              "\n",
              "[2 rows x 15 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-xVC8ZgdjnC9",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "fba91ce4-b830-49a0-8ebc-b02a1ac25834"
      },
      "source": [
        "from sklearn.metrics import classification_report\n",
        "print(CAT_NAMES)\n",
        "print(classification_report(out_df['true(+/-)'] , out_df['pred(+/-)']))\n",
        "print(out_df['pred(+/-)'].value_counts())\n",
        "print(out_df['true(+/-)'].value_counts())"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "['CDL2CROWS', 'CDL3BLACKCROWS', 'CDL3INSIDE', 'CDL3LINESTRIKE', 'CDL3OUTSIDE', 'CDL3STARSINSOUTH', 'CDL3WHITESOLDIERS', 'CDLABANDONEDBABY', 'CDLADVANCEBLOCK', 'CDLBELTHOLD', 'CDLBREAKAWAY', 'CDLCLOSINGMARUBOZU', 'CDLCONCEALBABYSWALL', 'CDLDARKCLOUDCOVER', 'CDLDOJI', 'CDLDOJISTAR', 'CDLDRAGONFLYDOJI', 'CDLENGULFING', 'CDLEVENINGDOJISTAR', 'CDLEVENINGSTAR', 'CDLGAPSIDESIDEWHITE', 'CDLGRAVESTONEDOJI', 'CDLHAMMER', 'CDLHANGINGMAN', 'CDLHARAMI', 'CDLHARAMICROSS', 'CDLHIGHWAVE', 'CDLHIKKAKE', 'CDLHIKKAKEMOD', 'CDLHOMINGPIGEON', 'CDLIDENTICAL3CROWS', 'CDLINNECK', 'CDLINVERTEDHAMMER', 'CDLKICKING', 'CDLLADDERBOTTOM', 'CDLLONGLEGGEDDOJI', 'CDLMARUBOZU', 'CDLMATCHINGLOW', 'CDLMATHOLD', 'CDLMORNINGDOJISTAR', 'CDLMORNINGSTAR', 'CDLONNECK', 'CDLPIERCING', 'CDLRICKSHAWMAN', 'CDLRISEFALL3METHODS', 'CDLSEPARATINGLINES', 'CDLSHOOTINGSTAR', 'CDLSPINNINGTOP', 'CDLSTICKSANDWICH', 'CDLTAKURI', 'CDLTASUKIGAP', 'CDLTHRUSTING', 'CDLTRISTAR', 'CDLUNIQUE3RIVER', 'CDLUPSIDEGAP2CROWS', 'CDLXSIDEGAP3METHODS', 'volatility_bbhi', 'volatility_bbli', 'volatility_kchi', 'volatility_kcli', 'trend_psar_up_indicator', 'trend_psar_down_indicator', 'weekday10', 'month10']\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "         +ve       0.89      0.82      0.85       163\n",
            "         -ve       1.00      0.68      0.81        44\n",
            "          no       0.46      0.75      0.57        44\n",
            "\n",
            "    accuracy                           0.78       251\n",
            "   macro avg       0.78      0.75      0.74       251\n",
            "weighted avg       0.83      0.78      0.79       251\n",
            "\n",
            "+ve    150\n",
            "no      71\n",
            "-ve     30\n",
            "Name: pred(+/-), dtype: int64\n",
            "+ve    163\n",
            "-ve     44\n",
            "no      44\n",
            "Name: true(+/-), dtype: int64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "collapsed": true,
        "jupyter": {
          "outputs_hidden": true
        },
        "id": "OSnwtkyKjnC9",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# num_category = len(TARGET)\n",
        "# train_range = pd.date_range('2019-12-23', periods=254, freq='B')\n",
        "# print(train_range)\n",
        "# train_df = raw_df[raw_df['date'].isin(train_range)]\n",
        "# train_df[CONT_NAMES] = train_df[CONT_NAMES].astype('float32')\n",
        "# temp_train = train_df.loc[:, COL].reset_index()\n",
        "# #temp_train = temp_train.drop(['index','date','level_0','symbol','CDLXSIDEGAP3METHODS'], axis=1)\n",
        "# train1_ds = tf_dataset(temp_train, train_df[TARGET], batch_size, True)\n",
        "# train_ds = train1_ds.take(int(len(train_df)/batch_size*0.9))\n",
        "# # # train_dev dataset for validate variance of model\n",
        "# train_dev_ds = train1_ds.skip(int(len(train_df)/batch_size*0.9))\n",
        "# allinputs, encodedfeatures = create_fea_col(train_ds)\n",
        "# model1, history = train_model(train_ds, train_dev_ds, 300, allinputs, encodedfeatures)"
      ],
      "execution_count": 26,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QkpN3OUqjnC-",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# val_df = raw_df[raw_df['date']=='2020-12-24']\n",
        "# val_df[CONT_NAMES] = val_df[CONT_NAMES].astype('float32')\n",
        "# temp_val = val_df.loc[:, COL].reset_index()\n",
        "# #temp_val = temp_val.drop(['index','date','level_0','symbol'], axis=1)\n",
        "# temp_val = temp_val.drop('index', axis=1)\n",
        "# val_ds = tf_dataset(temp_val, val_df[TARGET], batch_size, False)\n",
        "# model1.predict(val_ds)"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9lsohj_djnC_"
      },
      "source": [
        "# "
      ]
    }
  ]
}