lstm.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2587,
     "status": "ok",
     "timestamp": 1597492072008,
     "user": {
      "displayName": "Expss Xu",
      "photoUrl": "",
      "userId": "17480852382145563764"
     },
     "user_tz": -480
    },
    "id": "5NCGpNCVotjd"
   },
   "outputs": [],
   "source": [
    "#导入必须的库\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.metrics import mean_absolute_error\n",
    "plt.style.use('fivethirtyeight')#画图风格\n",
    "np.random.seed(100)\n",
    "window=12#时间窗\n",
    "label_dim=8#标签纬度\n",
    "scale=100#归一化参数\n",
    "amount_of_features=1#特征数量\n",
    "d=0.05#dropout系数\n",
    "batch_size=12#批训练数量\n",
    "epoch=200#总迭代次数\n",
    "train_file='international-airline-passengers.csv'#读取训练集地址\n",
    "test_file='example.csv'#读取训练集地址\n",
    "save_file='predict_result_lstm.csv'#储存预测值地址"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "data=pd.read_csv(train_file,header=0) \n",
    "data=data['passengers']#选数据集的哪一列\n",
    "#构造针对机器学习模型的数据集\n",
    "data=data.values \n",
    "dataset=data\n",
    "for i in range(window+label_dim-1):\n",
    "    zero=np.zeros(i+1)\n",
    "    temp=np.append(data[i+1:],zero)\n",
    "    dataset=np.row_stack((dataset,temp))\n",
    "dataset=pd.DataFrame(dataset).T\n",
    "dataset=dataset.iloc[:-(window+label_dim)]\n",
    "dataset=dataset/scale#进行伪归一化\n",
    "import seaborn as sns\n",
    "plt.figure(figsize=(10,5))\n",
    "sns.heatmap(dataset,annot=False)\n",
    "plt.show()#展示下原始数据集"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "#划分特征与标签\n",
    "X=dataset.iloc[:,:window]\n",
    "y=dataset.iloc[:,-label_dim:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 89
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2576,
     "status": "ok",
     "timestamp": 1597492072008,
     "user": {
      "displayName": "Expss Xu",
      "photoUrl": "",
      "userId": "17480852382145563764"
     },
     "user_tz": -480
    },
    "id": "UqmQarL8otjj",
    "outputId": "068dd905-8a38-40c6-c417-1407aca65c65"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(124, 12)\n",
      "(124, 8)\n"
     ]
    }
   ],
   "source": [
    "#构造训练集测试集\n",
    "X,y=X.values,y.values\n",
    "print(X.shape)#通过输出训练集测试集的大小来判断数据格式正确。\n",
    "print(y.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 89
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2570,
     "status": "ok",
     "timestamp": 1597492072009,
     "user": {
      "displayName": "Expss Xu",
      "photoUrl": "",
      "userId": "17480852382145563764"
     },
     "user_tz": -480
    },
    "id": "Z14xqX3gotjn",
    "outputId": "54248355-3e0a-4f2d-84e0-1adfaa2dca4c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(124, 12, 1)\n",
      "(124, 8)\n"
     ]
    }
   ],
   "source": [
    "#LSTM神经网络的数据集要稍微做一些修改\n",
    "X= np.reshape(X, (X.shape[0],X.shape[1], amount_of_features))\n",
    "print(X.shape)\n",
    "print(y.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 269
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 3600,
     "status": "ok",
     "timestamp": 1597492073045,
     "user": {
      "displayName": "Expss Xu",
      "photoUrl": "",
      "userId": "17480852382145563764"
     },
     "user_tz": -480
    },
    "id": "i0dUbIgGotjq",
    "outputId": "9a38af4d-6447-4c0c-c7b0-d6b70db24324"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "lstm_3 (LSTM)                (None, 32)                4352      \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 16)                528       \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 8)                 136       \n",
      "=================================================================\n",
      "Total params: 5,016\n",
      "Trainable params: 5,016\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "#建立LSTM模型 训练\n",
    "from keras.models import Sequential\n",
    "from keras.layers.core import Dense, Dropout, Activation,Dropout\n",
    "from keras.layers.recurrent import LSTM\n",
    "#建立训练模型过程\n",
    "model = Sequential()#建立层次模型\n",
    "model.add(LSTM(32, input_shape=(window, amount_of_features), return_sequences=False))#建立LSTM层\n",
    "#model.add(Dropout(d))\n",
    "model.add(Dense(label_dim*2,activation='relu'))   #建立全连接层  \n",
    "#model.add(Dropout(d))\n",
    "model.add(Dense(label_dim))\n",
    "model.compile(loss='mse',optimizer='adam',metrics=['accuracy'])\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 22110,
     "status": "ok",
     "timestamp": 1597492091563,
     "user": {
      "displayName": "Expss Xu",
      "photoUrl": "",
      "userId": "17480852382145563764"
     },
     "user_tz": -480
    },
    "id": "j1TvX44ootjs",
    "outputId": "9987a27f-6bcf-4bfb-d47f-b95c46501fe0",
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 111 samples, validate on 13 samples\n",
      "Epoch 1/200\n",
      "111/111 [==============================] - 1s 7ms/step - loss: 7.9568 - acc: 0.0901 - val_loss: 19.7063 - val_acc: 0.1538\n",
      "Epoch 2/200\n",
      "111/111 [==============================] - 0s 812us/step - loss: 7.1506 - acc: 0.0901 - val_loss: 18.1487 - val_acc: 0.1538\n",
      "Epoch 3/200\n",
      "111/111 [==============================] - 0s 871us/step - loss: 6.2971 - acc: 0.0901 - val_loss: 16.2256 - val_acc: 0.1538\n",
      "Epoch 4/200\n",
      "111/111 [==============================] - 0s 883us/step - loss: 5.2531 - acc: 0.0901 - val_loss: 13.7906 - val_acc: 0.1538\n",
      "Epoch 5/200\n",
      "111/111 [==============================] - 0s 870us/step - loss: 4.0172 - acc: 0.0901 - val_loss: 10.9641 - val_acc: 0.1538\n",
      "Epoch 6/200\n",
      "111/111 [==============================] - 0s 880us/step - loss: 2.8970 - acc: 0.0901 - val_loss: 8.1816 - val_acc: 0.1538\n",
      "Epoch 7/200\n",
      "111/111 [==============================] - 0s 821us/step - loss: 2.0284 - acc: 0.0901 - val_loss: 6.2961 - val_acc: 0.1538\n",
      "Epoch 8/200\n",
      "111/111 [==============================] - 0s 839us/step - loss: 1.4619 - acc: 0.0901 - val_loss: 5.0615 - val_acc: 0.1538\n",
      "Epoch 9/200\n",
      "111/111 [==============================] - 0s 988us/step - loss: 1.0562 - acc: 0.0901 - val_loss: 3.8703 - val_acc: 0.1538\n",
      "Epoch 10/200\n",
      "111/111 [==============================] - 0s 786us/step - loss: 0.7684 - acc: 0.0901 - val_loss: 2.9948 - val_acc: 0.1538\n",
      "Epoch 11/200\n",
      "111/111 [==============================] - 0s 861us/step - loss: 0.5301 - acc: 0.0991 - val_loss: 2.3896 - val_acc: 0.1538\n",
      "Epoch 12/200\n",
      "111/111 [==============================] - 0s 823us/step - loss: 0.3716 - acc: 0.0811 - val_loss: 1.9681 - val_acc: 0.1538\n",
      "Epoch 13/200\n",
      "111/111 [==============================] - 0s 792us/step - loss: 0.2758 - acc: 0.0901 - val_loss: 1.5245 - val_acc: 0.1538\n",
      "Epoch 14/200\n",
      "111/111 [==============================] - 0s 845us/step - loss: 0.2242 - acc: 0.0991 - val_loss: 1.2932 - val_acc: 0.1538\n",
      "Epoch 15/200\n",
      "111/111 [==============================] - 0s 816us/step - loss: 0.2037 - acc: 0.1171 - val_loss: 1.1250 - val_acc: 0.0769\n",
      "Epoch 16/200\n",
      "111/111 [==============================] - 0s 809us/step - loss: 0.1946 - acc: 0.1802 - val_loss: 1.0196 - val_acc: 0.0769\n",
      "Epoch 17/200\n",
      "111/111 [==============================] - 0s 786us/step - loss: 0.1834 - acc: 0.1622 - val_loss: 0.9212 - val_acc: 0.0769\n",
      "Epoch 18/200\n",
      "111/111 [==============================] - 0s 800us/step - loss: 0.1825 - acc: 0.1712 - val_loss: 0.9114 - val_acc: 0.0769\n",
      "Epoch 19/200\n",
      "111/111 [==============================] - 0s 795us/step - loss: 0.1791 - acc: 0.2162 - val_loss: 0.8947 - val_acc: 0.2308\n",
      "Epoch 20/200\n",
      "111/111 [==============================] - 0s 792us/step - loss: 0.1758 - acc: 0.2162 - val_loss: 0.9223 - val_acc: 0.2308\n",
      "Epoch 21/200\n",
      "111/111 [==============================] - 0s 780us/step - loss: 0.1731 - acc: 0.2162 - val_loss: 0.8695 - val_acc: 0.2308\n",
      "Epoch 22/200\n",
      "111/111 [==============================] - 0s 787us/step - loss: 0.1801 - acc: 0.2162 - val_loss: 0.9087 - val_acc: 0.2308\n",
      "Epoch 23/200\n",
      "111/111 [==============================] - 0s 776us/step - loss: 0.1714 - acc: 0.2162 - val_loss: 0.8951 - val_acc: 0.2308\n",
      "Epoch 24/200\n",
      "111/111 [==============================] - 0s 787us/step - loss: 0.1713 - acc: 0.2252 - val_loss: 0.8962 - val_acc: 0.2308\n",
      "Epoch 25/200\n",
      "111/111 [==============================] - 0s 780us/step - loss: 0.1694 - acc: 0.2432 - val_loss: 0.8639 - val_acc: 0.2308\n",
      "Epoch 26/200\n",
      "111/111 [==============================] - 0s 797us/step - loss: 0.1680 - acc: 0.2342 - val_loss: 0.8331 - val_acc: 0.2308\n",
      "Epoch 27/200\n",
      "111/111 [==============================] - 0s 796us/step - loss: 0.1658 - acc: 0.2432 - val_loss: 0.8034 - val_acc: 0.2308\n",
      "Epoch 28/200\n",
      "111/111 [==============================] - 0s 785us/step - loss: 0.1671 - acc: 0.1982 - val_loss: 0.8495 - val_acc: 0.1538\n",
      "Epoch 29/200\n",
      "111/111 [==============================] - 0s 805us/step - loss: 0.1658 - acc: 0.1622 - val_loss: 0.8810 - val_acc: 0.1538\n",
      "Epoch 30/200\n",
      "111/111 [==============================] - 0s 781us/step - loss: 0.1671 - acc: 0.2162 - val_loss: 0.7756 - val_acc: 0.2308\n",
      "Epoch 31/200\n",
      "111/111 [==============================] - 0s 795us/step - loss: 0.1761 - acc: 0.1892 - val_loss: 0.7926 - val_acc: 0.0769\n",
      "Epoch 32/200\n",
      "111/111 [==============================] - 0s 793us/step - loss: 0.1698 - acc: 0.1712 - val_loss: 0.8415 - val_acc: 0.2308\n",
      "Epoch 33/200\n",
      "111/111 [==============================] - 0s 779us/step - loss: 0.1747 - acc: 0.2072 - val_loss: 0.7545 - val_acc: 0.2308\n",
      "Epoch 34/200\n",
      "111/111 [==============================] - 0s 765us/step - loss: 0.1812 - acc: 0.1892 - val_loss: 0.7574 - val_acc: 0.1538\n",
      "Epoch 35/200\n",
      "111/111 [==============================] - 0s 789us/step - loss: 0.1718 - acc: 0.1532 - val_loss: 0.8277 - val_acc: 0.2308\n",
      "Epoch 36/200\n",
      "111/111 [==============================] - 0s 810us/step - loss: 0.1618 - acc: 0.1532 - val_loss: 0.7826 - val_acc: 0.3077\n",
      "Epoch 37/200\n",
      "111/111 [==============================] - 0s 813us/step - loss: 0.1607 - acc: 0.2162 - val_loss: 0.7751 - val_acc: 0.3077\n",
      "Epoch 38/200\n",
      "111/111 [==============================] - 0s 811us/step - loss: 0.1593 - acc: 0.2432 - val_loss: 0.8341 - val_acc: 0.2308\n",
      "Epoch 39/200\n",
      "111/111 [==============================] - 0s 797us/step - loss: 0.1573 - acc: 0.2252 - val_loss: 0.7566 - val_acc: 0.2308\n",
      "Epoch 40/200\n",
      "111/111 [==============================] - 0s 772us/step - loss: 0.1577 - acc: 0.1802 - val_loss: 0.7776 - val_acc: 0.2308\n",
      "Epoch 41/200\n",
      "111/111 [==============================] - 0s 781us/step - loss: 0.1562 - acc: 0.1802 - val_loss: 0.8059 - val_acc: 0.3077\n",
      "Epoch 42/200\n",
      "111/111 [==============================] - 0s 786us/step - loss: 0.1660 - acc: 0.2613 - val_loss: 0.7485 - val_acc: 0.2308\n",
      "Epoch 43/200\n",
      "111/111 [==============================] - 0s 786us/step - loss: 0.1603 - acc: 0.1712 - val_loss: 0.6855 - val_acc: 0.1538\n",
      "Epoch 44/200\n",
      "111/111 [==============================] - 0s 783us/step - loss: 0.1565 - acc: 0.1532 - val_loss: 0.7083 - val_acc: 0.2308\n",
      "Epoch 45/200\n",
      "111/111 [==============================] - 0s 794us/step - loss: 0.1538 - acc: 0.2162 - val_loss: 0.6876 - val_acc: 0.3077\n",
      "Epoch 46/200\n",
      "111/111 [==============================] - 0s 757us/step - loss: 0.1566 - acc: 0.1982 - val_loss: 0.7270 - val_acc: 0.2308\n",
      "Epoch 47/200\n",
      "111/111 [==============================] - 0s 748us/step - loss: 0.1516 - acc: 0.1712 - val_loss: 0.7475 - val_acc: 0.3077\n",
      "Epoch 48/200\n",
      "111/111 [==============================] - 0s 752us/step - loss: 0.1514 - acc: 0.2162 - val_loss: 0.7572 - val_acc: 0.3846\n",
      "Epoch 49/200\n",
      "111/111 [==============================] - 0s 765us/step - loss: 0.1549 - acc: 0.1802 - val_loss: 0.7257 - val_acc: 0.2308\n",
      "Epoch 50/200\n",
      "111/111 [==============================] - 0s 768us/step - loss: 0.1513 - acc: 0.1622 - val_loss: 0.6857 - val_acc: 0.2308\n",
      "Epoch 51/200\n",
      "111/111 [==============================] - 0s 752us/step - loss: 0.1515 - acc: 0.2252 - val_loss: 0.6819 - val_acc: 0.3077\n",
      "Epoch 52/200\n",
      "111/111 [==============================] - 0s 838us/step - loss: 0.1500 - acc: 0.1982 - val_loss: 0.6827 - val_acc: 0.3077\n",
      "Epoch 53/200\n",
      "111/111 [==============================] - 0s 864us/step - loss: 0.1474 - acc: 0.1982 - val_loss: 0.6523 - val_acc: 0.3077\n",
      "Epoch 54/200\n",
      "111/111 [==============================] - 0s 777us/step - loss: 0.1454 - acc: 0.2523 - val_loss: 0.7440 - val_acc: 0.3077\n",
      "Epoch 55/200\n",
      "111/111 [==============================] - 0s 741us/step - loss: 0.1454 - acc: 0.2342 - val_loss: 0.7753 - val_acc: 0.3077\n",
      "Epoch 56/200\n",
      "111/111 [==============================] - 0s 736us/step - loss: 0.1440 - acc: 0.2162 - val_loss: 0.7453 - val_acc: 0.3846\n",
      "Epoch 57/200\n",
      "111/111 [==============================] - 0s 730us/step - loss: 0.1467 - acc: 0.2703 - val_loss: 0.7724 - val_acc: 0.2308\n",
      "Epoch 58/200\n",
      "111/111 [==============================] - 0s 735us/step - loss: 0.1478 - acc: 0.2523 - val_loss: 0.6743 - val_acc: 0.3077\n",
      "Epoch 59/200\n",
      "111/111 [==============================] - 0s 741us/step - loss: 0.1404 - acc: 0.2342 - val_loss: 0.7591 - val_acc: 0.2308\n",
      "Epoch 60/200\n",
      "111/111 [==============================] - 0s 736us/step - loss: 0.1389 - acc: 0.2613 - val_loss: 0.6787 - val_acc: 0.3846\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 61/200\n",
      "111/111 [==============================] - 0s 737us/step - loss: 0.1366 - acc: 0.2252 - val_loss: 0.6618 - val_acc: 0.3846\n",
      "Epoch 62/200\n",
      "111/111 [==============================] - 0s 839us/step - loss: 0.1403 - acc: 0.2883 - val_loss: 0.6716 - val_acc: 0.3846\n",
      "Epoch 63/200\n",
      "111/111 [==============================] - 0s 803us/step - loss: 0.1309 - acc: 0.2703 - val_loss: 0.7254 - val_acc: 0.2308\n",
      "Epoch 64/200\n",
      "111/111 [==============================] - 0s 868us/step - loss: 0.1292 - acc: 0.2883 - val_loss: 0.6759 - val_acc: 0.3077\n",
      "Epoch 65/200\n",
      "111/111 [==============================] - 0s 928us/step - loss: 0.1309 - acc: 0.2523 - val_loss: 0.6707 - val_acc: 0.3077\n",
      "Epoch 66/200\n",
      "111/111 [==============================] - 0s 915us/step - loss: 0.1295 - acc: 0.2432 - val_loss: 0.7325 - val_acc: 0.2308\n",
      "Epoch 67/200\n",
      "111/111 [==============================] - 0s 917us/step - loss: 0.1308 - acc: 0.2523 - val_loss: 0.6362 - val_acc: 0.3077\n",
      "Epoch 68/200\n",
      "111/111 [==============================] - 0s 861us/step - loss: 0.1311 - acc: 0.3063 - val_loss: 0.6510 - val_acc: 0.3846\n",
      "Epoch 69/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1214 - acc: 0.2973 - val_loss: 0.5902 - val_acc: 0.3077\n",
      "Epoch 70/200\n",
      "111/111 [==============================] - 0s 937us/step - loss: 0.1272 - acc: 0.2793 - val_loss: 0.6802 - val_acc: 0.2308\n",
      "Epoch 71/200\n",
      "111/111 [==============================] - 0s 876us/step - loss: 0.1201 - acc: 0.2793 - val_loss: 0.6620 - val_acc: 0.3846\n",
      "Epoch 72/200\n",
      "111/111 [==============================] - 0s 915us/step - loss: 0.1167 - acc: 0.2523 - val_loss: 0.8004 - val_acc: 0.0769\n",
      "Epoch 73/200\n",
      "111/111 [==============================] - 0s 862us/step - loss: 0.1185 - acc: 0.1892 - val_loss: 0.6558 - val_acc: 0.2308\n",
      "Epoch 74/200\n",
      "111/111 [==============================] - 0s 919us/step - loss: 0.1149 - acc: 0.3153 - val_loss: 0.6129 - val_acc: 0.3846\n",
      "Epoch 75/200\n",
      "111/111 [==============================] - 0s 965us/step - loss: 0.1147 - acc: 0.3153 - val_loss: 0.5744 - val_acc: 0.3846\n",
      "Epoch 76/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1116 - acc: 0.3063 - val_loss: 0.5835 - val_acc: 0.1538\n",
      "Epoch 77/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1072 - acc: 0.2703 - val_loss: 0.5114 - val_acc: 0.3077\n",
      "Epoch 78/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1121 - acc: 0.2703 - val_loss: 0.5618 - val_acc: 0.2308\n",
      "Epoch 79/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1085 - acc: 0.2432 - val_loss: 0.6993 - val_acc: 0.1538\n",
      "Epoch 80/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1180 - acc: 0.2523 - val_loss: 0.7533 - val_acc: 0.2308\n",
      "Epoch 81/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1138 - acc: 0.2703 - val_loss: 0.6389 - val_acc: 0.3077\n",
      "Epoch 82/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1045 - acc: 0.2793 - val_loss: 0.5646 - val_acc: 0.0769\n",
      "Epoch 83/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1052 - acc: 0.2703 - val_loss: 0.5082 - val_acc: 0.2308\n",
      "Epoch 84/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1017 - acc: 0.2523 - val_loss: 0.5198 - val_acc: 0.2308\n",
      "Epoch 85/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1066 - acc: 0.2162 - val_loss: 0.5518 - val_acc: 0.2308\n",
      "Epoch 86/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.1093 - acc: 0.3243 - val_loss: 0.5060 - val_acc: 0.3077\n",
      "Epoch 87/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.0992 - acc: 0.2613 - val_loss: 0.5591 - val_acc: 0.2308\n",
      "Epoch 88/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.0979 - acc: 0.2793 - val_loss: 0.5237 - val_acc: 0.2308\n",
      "Epoch 89/200\n",
      "111/111 [==============================] - 0s 982us/step - loss: 0.0930 - acc: 0.2973 - val_loss: 0.4781 - val_acc: 0.3846\n",
      "Epoch 90/200\n",
      "111/111 [==============================] - 0s 1ms/step - loss: 0.0934 - acc: 0.3423 - val_loss: 0.4611 - val_acc: 0.3077\n",
      "Epoch 91/200\n",
      "111/111 [==============================] - 0s 902us/step - loss: 0.0944 - acc: 0.2973 - val_loss: 0.4754 - val_acc: 0.3846\n",
      "Epoch 92/200\n",
      "111/111 [==============================] - 0s 775us/step - loss: 0.0928 - acc: 0.3063 - val_loss: 0.4495 - val_acc: 0.3077\n",
      "Epoch 93/200\n",
      "111/111 [==============================] - 0s 858us/step - loss: 0.0954 - acc: 0.2883 - val_loss: 0.4964 - val_acc: 0.3846\n",
      "Epoch 94/200\n",
      "111/111 [==============================] - 0s 973us/step - loss: 0.0908 - acc: 0.2973 - val_loss: 0.4629 - val_acc: 0.3077\n",
      "Epoch 95/200\n",
      "111/111 [==============================] - 0s 953us/step - loss: 0.0879 - acc: 0.2793 - val_loss: 0.4767 - val_acc: 0.3846\n",
      "Epoch 96/200\n",
      "111/111 [==============================] - 0s 856us/step - loss: 0.0897 - acc: 0.2973 - val_loss: 0.4404 - val_acc: 0.3846\n",
      "Epoch 97/200\n",
      "111/111 [==============================] - 0s 773us/step - loss: 0.0883 - acc: 0.3514 - val_loss: 0.4971 - val_acc: 0.5385\n",
      "Epoch 98/200\n",
      "111/111 [==============================] - 0s 790us/step - loss: 0.0860 - acc: 0.2342 - val_loss: 0.4721 - val_acc: 0.4615\n",
      "Epoch 99/200\n",
      "111/111 [==============================] - 0s 795us/step - loss: 0.0889 - acc: 0.3423 - val_loss: 0.5240 - val_acc: 0.3846\n",
      "Epoch 100/200\n",
      "111/111 [==============================] - 0s 807us/step - loss: 0.0857 - acc: 0.3604 - val_loss: 0.4308 - val_acc: 0.4615\n",
      "Epoch 101/200\n",
      "111/111 [==============================] - 0s 818us/step - loss: 0.0897 - acc: 0.2973 - val_loss: 0.4963 - val_acc: 0.5385\n",
      "Epoch 102/200\n",
      "111/111 [==============================] - 0s 816us/step - loss: 0.0894 - acc: 0.3604 - val_loss: 0.5137 - val_acc: 0.4615\n",
      "Epoch 103/200\n",
      "111/111 [==============================] - 0s 808us/step - loss: 0.1046 - acc: 0.3063 - val_loss: 0.5676 - val_acc: 0.3846\n",
      "Epoch 104/200\n",
      "111/111 [==============================] - 0s 814us/step - loss: 0.0905 - acc: 0.3333 - val_loss: 0.6175 - val_acc: 0.3077\n",
      "Epoch 105/200\n",
      "111/111 [==============================] - 0s 808us/step - loss: 0.1014 - acc: 0.3423 - val_loss: 0.4978 - val_acc: 0.4615\n",
      "Epoch 106/200\n",
      "111/111 [==============================] - 0s 807us/step - loss: 0.0995 - acc: 0.3153 - val_loss: 0.3539 - val_acc: 0.4615\n",
      "Epoch 107/200\n",
      "111/111 [==============================] - 0s 796us/step - loss: 0.1001 - acc: 0.3153 - val_loss: 0.3443 - val_acc: 0.3846\n",
      "Epoch 108/200\n",
      "111/111 [==============================] - 0s 826us/step - loss: 0.0982 - acc: 0.3694 - val_loss: 0.3855 - val_acc: 0.4615\n",
      "Epoch 109/200\n",
      "111/111 [==============================] - 0s 836us/step - loss: 0.0985 - acc: 0.3423 - val_loss: 0.4306 - val_acc: 0.5385\n",
      "Epoch 110/200\n",
      "111/111 [==============================] - 0s 898us/step - loss: 0.0968 - acc: 0.3333 - val_loss: 0.4606 - val_acc: 0.3846\n",
      "Epoch 111/200\n",
      "111/111 [==============================] - 0s 865us/step - loss: 0.0845 - acc: 0.4144 - val_loss: 0.4007 - val_acc: 0.5385\n",
      "Epoch 112/200\n",
      "111/111 [==============================] - 0s 851us/step - loss: 0.0776 - acc: 0.3784 - val_loss: 0.3902 - val_acc: 0.4615\n",
      "Epoch 113/200\n",
      "111/111 [==============================] - 0s 846us/step - loss: 0.0773 - acc: 0.3514 - val_loss: 0.4312 - val_acc: 0.4615\n",
      "Epoch 114/200\n",
      "111/111 [==============================] - 0s 880us/step - loss: 0.0765 - acc: 0.4234 - val_loss: 0.3996 - val_acc: 0.6154\n",
      "Epoch 115/200\n",
      "111/111 [==============================] - 0s 829us/step - loss: 0.0765 - acc: 0.3964 - val_loss: 0.3796 - val_acc: 0.5385\n",
      "Epoch 116/200\n",
      "111/111 [==============================] - 0s 802us/step - loss: 0.0770 - acc: 0.3604 - val_loss: 0.4157 - val_acc: 0.5385\n",
      "Epoch 117/200\n",
      "111/111 [==============================] - 0s 836us/step - loss: 0.0766 - acc: 0.3784 - val_loss: 0.3799 - val_acc: 0.5385\n",
      "Epoch 118/200\n",
      "111/111 [==============================] - 0s 817us/step - loss: 0.0815 - acc: 0.4054 - val_loss: 0.4398 - val_acc: 0.4615\n",
      "Epoch 119/200\n",
      "111/111 [==============================] - 0s 825us/step - loss: 0.0798 - acc: 0.3604 - val_loss: 0.4246 - val_acc: 0.5385\n",
      "Epoch 120/200\n",
      "111/111 [==============================] - 0s 974us/step - loss: 0.0802 - acc: 0.3874 - val_loss: 0.3815 - val_acc: 0.5385\n",
      "Epoch 121/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "111/111 [==============================] - 0s 874us/step - loss: 0.0734 - acc: 0.3784 - val_loss: 0.3602 - val_acc: 0.6154\n",
      "Epoch 122/200\n",
      "111/111 [==============================] - 0s 834us/step - loss: 0.0748 - acc: 0.3964 - val_loss: 0.3918 - val_acc: 0.3846\n",
      "Epoch 123/200\n",
      "111/111 [==============================] - 0s 821us/step - loss: 0.0725 - acc: 0.3874 - val_loss: 0.3860 - val_acc: 0.5385\n",
      "Epoch 124/200\n",
      "111/111 [==============================] - 0s 834us/step - loss: 0.0720 - acc: 0.3694 - val_loss: 0.3636 - val_acc: 0.5385\n",
      "Epoch 125/200\n",
      "111/111 [==============================] - 0s 833us/step - loss: 0.0692 - acc: 0.4144 - val_loss: 0.4061 - val_acc: 0.5385\n",
      "Epoch 126/200\n",
      "111/111 [==============================] - 0s 819us/step - loss: 0.0686 - acc: 0.3964 - val_loss: 0.3578 - val_acc: 0.4615\n",
      "Epoch 127/200\n",
      "111/111 [==============================] - 0s 804us/step - loss: 0.0687 - acc: 0.4234 - val_loss: 0.3722 - val_acc: 0.4615\n",
      "Epoch 128/200\n",
      "111/111 [==============================] - 0s 828us/step - loss: 0.0662 - acc: 0.4414 - val_loss: 0.3440 - val_acc: 0.5385\n",
      "Epoch 129/200\n",
      "111/111 [==============================] - 0s 844us/step - loss: 0.0681 - acc: 0.4144 - val_loss: 0.3264 - val_acc: 0.6154\n",
      "Epoch 130/200\n",
      "111/111 [==============================] - 0s 861us/step - loss: 0.0706 - acc: 0.4324 - val_loss: 0.3935 - val_acc: 0.4615\n",
      "Epoch 131/200\n",
      "111/111 [==============================] - 0s 818us/step - loss: 0.0728 - acc: 0.3784 - val_loss: 0.4689 - val_acc: 0.5385\n",
      "Epoch 132/200\n",
      "111/111 [==============================] - 0s 834us/step - loss: 0.0760 - acc: 0.4054 - val_loss: 0.4340 - val_acc: 0.4615\n",
      "Epoch 133/200\n",
      "111/111 [==============================] - 0s 840us/step - loss: 0.0724 - acc: 0.4144 - val_loss: 0.3225 - val_acc: 0.5385\n",
      "Epoch 134/200\n",
      "111/111 [==============================] - 0s 786us/step - loss: 0.0652 - acc: 0.4505 - val_loss: 0.3540 - val_acc: 0.5385\n",
      "Epoch 135/200\n",
      "111/111 [==============================] - 0s 798us/step - loss: 0.0646 - acc: 0.4685 - val_loss: 0.3619 - val_acc: 0.5385\n",
      "Epoch 136/200\n",
      "111/111 [==============================] - 0s 791us/step - loss: 0.0692 - acc: 0.4054 - val_loss: 0.3607 - val_acc: 0.5385\n",
      "Epoch 137/200\n",
      "111/111 [==============================] - 0s 855us/step - loss: 0.0734 - acc: 0.4144 - val_loss: 0.2822 - val_acc: 0.6154\n",
      "Epoch 138/200\n",
      "111/111 [==============================] - 0s 828us/step - loss: 0.0661 - acc: 0.4324 - val_loss: 0.3096 - val_acc: 0.6154\n",
      "Epoch 139/200\n",
      "111/111 [==============================] - 0s 799us/step - loss: 0.0636 - acc: 0.4324 - val_loss: 0.3411 - val_acc: 0.5385\n",
      "Epoch 140/200\n",
      "111/111 [==============================] - 0s 798us/step - loss: 0.0625 - acc: 0.4054 - val_loss: 0.3804 - val_acc: 0.5385\n",
      "Epoch 141/200\n",
      "111/111 [==============================] - 0s 777us/step - loss: 0.0646 - acc: 0.4595 - val_loss: 0.3385 - val_acc: 0.6154\n",
      "Epoch 142/200\n",
      "111/111 [==============================] - 0s 760us/step - loss: 0.0646 - acc: 0.4505 - val_loss: 0.2967 - val_acc: 0.5385\n",
      "Epoch 143/200\n",
      "111/111 [==============================] - 0s 779us/step - loss: 0.0592 - acc: 0.4324 - val_loss: 0.3199 - val_acc: 0.5385\n",
      "Epoch 144/200\n",
      "111/111 [==============================] - 0s 786us/step - loss: 0.0590 - acc: 0.4595 - val_loss: 0.2874 - val_acc: 0.5385\n",
      "Epoch 145/200\n",
      "111/111 [==============================] - 0s 805us/step - loss: 0.0626 - acc: 0.4505 - val_loss: 0.3234 - val_acc: 0.6154\n",
      "Epoch 146/200\n",
      "111/111 [==============================] - 0s 820us/step - loss: 0.0631 - acc: 0.4144 - val_loss: 0.3031 - val_acc: 0.5385\n",
      "Epoch 147/200\n",
      "111/111 [==============================] - 0s 820us/step - loss: 0.0604 - acc: 0.4414 - val_loss: 0.2986 - val_acc: 0.5385\n",
      "Epoch 148/200\n",
      "111/111 [==============================] - 0s 802us/step - loss: 0.0615 - acc: 0.4775 - val_loss: 0.3718 - val_acc: 0.5385\n",
      "Epoch 149/200\n",
      "111/111 [==============================] - 0s 811us/step - loss: 0.0615 - acc: 0.4505 - val_loss: 0.4109 - val_acc: 0.5385\n",
      "Epoch 150/200\n",
      "111/111 [==============================] - 0s 802us/step - loss: 0.0608 - acc: 0.4955 - val_loss: 0.3483 - val_acc: 0.6154\n",
      "Epoch 151/200\n",
      "111/111 [==============================] - 0s 808us/step - loss: 0.0635 - acc: 0.4595 - val_loss: 0.3065 - val_acc: 0.4615\n",
      "Epoch 152/200\n",
      "111/111 [==============================] - 0s 825us/step - loss: 0.0540 - acc: 0.4955 - val_loss: 0.2866 - val_acc: 0.4615\n",
      "Epoch 153/200\n",
      "111/111 [==============================] - 0s 966us/step - loss: 0.0537 - acc: 0.5045 - val_loss: 0.3537 - val_acc: 0.5385\n",
      "Epoch 154/200\n",
      "111/111 [==============================] - 0s 839us/step - loss: 0.0523 - acc: 0.5495 - val_loss: 0.3152 - val_acc: 0.5385\n",
      "Epoch 155/200\n",
      "111/111 [==============================] - 0s 756us/step - loss: 0.0500 - acc: 0.5135 - val_loss: 0.3285 - val_acc: 0.6154\n",
      "Epoch 156/200\n",
      "111/111 [==============================] - 0s 760us/step - loss: 0.0510 - acc: 0.5045 - val_loss: 0.3210 - val_acc: 0.5385\n",
      "Epoch 157/200\n",
      "111/111 [==============================] - 0s 765us/step - loss: 0.0542 - acc: 0.4595 - val_loss: 0.3720 - val_acc: 0.5385\n",
      "Epoch 158/200\n",
      "111/111 [==============================] - 0s 794us/step - loss: 0.0549 - acc: 0.5225 - val_loss: 0.3536 - val_acc: 0.6154\n",
      "Epoch 159/200\n",
      "111/111 [==============================] - 0s 811us/step - loss: 0.0583 - acc: 0.4505 - val_loss: 0.3452 - val_acc: 0.5385\n",
      "Epoch 160/200\n",
      "111/111 [==============================] - 0s 812us/step - loss: 0.0508 - acc: 0.5045 - val_loss: 0.3235 - val_acc: 0.5385\n",
      "Epoch 161/200\n",
      "111/111 [==============================] - 0s 800us/step - loss: 0.0494 - acc: 0.5315 - val_loss: 0.3318 - val_acc: 0.6154\n",
      "Epoch 162/200\n",
      "111/111 [==============================] - 0s 787us/step - loss: 0.0535 - acc: 0.4865 - val_loss: 0.2825 - val_acc: 0.6154\n",
      "Epoch 163/200\n",
      "111/111 [==============================] - 0s 787us/step - loss: 0.0542 - acc: 0.4865 - val_loss: 0.3837 - val_acc: 0.5385\n",
      "Epoch 164/200\n",
      "111/111 [==============================] - 0s 888us/step - loss: 0.0549 - acc: 0.4865 - val_loss: 0.2726 - val_acc: 0.5385\n",
      "Epoch 165/200\n",
      "111/111 [==============================] - 0s 806us/step - loss: 0.0545 - acc: 0.5135 - val_loss: 0.2755 - val_acc: 0.6154\n",
      "Epoch 166/200\n",
      "111/111 [==============================] - 0s 805us/step - loss: 0.0486 - acc: 0.4775 - val_loss: 0.3293 - val_acc: 0.6923\n",
      "Epoch 167/200\n",
      "111/111 [==============================] - 0s 790us/step - loss: 0.0477 - acc: 0.5135 - val_loss: 0.3350 - val_acc: 0.5385\n",
      "Epoch 168/200\n",
      "111/111 [==============================] - 0s 766us/step - loss: 0.0480 - acc: 0.4955 - val_loss: 0.3165 - val_acc: 0.5385\n",
      "Epoch 169/200\n",
      "111/111 [==============================] - 0s 769us/step - loss: 0.0489 - acc: 0.4955 - val_loss: 0.3186 - val_acc: 0.6154\n",
      "Epoch 170/200\n",
      "111/111 [==============================] - 0s 883us/step - loss: 0.0467 - acc: 0.5045 - val_loss: 0.2969 - val_acc: 0.4615\n",
      "Epoch 171/200\n",
      "111/111 [==============================] - 0s 882us/step - loss: 0.0444 - acc: 0.5495 - val_loss: 0.3077 - val_acc: 0.5385\n",
      "Epoch 172/200\n",
      "111/111 [==============================] - 0s 852us/step - loss: 0.0509 - acc: 0.5045 - val_loss: 0.3080 - val_acc: 0.5385\n",
      "Epoch 173/200\n",
      "111/111 [==============================] - 0s 800us/step - loss: 0.0490 - acc: 0.4505 - val_loss: 0.3243 - val_acc: 0.4615\n",
      "Epoch 174/200\n",
      "111/111 [==============================] - 0s 759us/step - loss: 0.0468 - acc: 0.4685 - val_loss: 0.3141 - val_acc: 0.5385\n",
      "Epoch 175/200\n",
      "111/111 [==============================] - 0s 812us/step - loss: 0.0450 - acc: 0.4865 - val_loss: 0.3443 - val_acc: 0.3846\n",
      "Epoch 176/200\n",
      "111/111 [==============================] - 0s 812us/step - loss: 0.0479 - acc: 0.4775 - val_loss: 0.2935 - val_acc: 0.6154\n",
      "Epoch 177/200\n",
      "111/111 [==============================] - 0s 764us/step - loss: 0.0469 - acc: 0.5315 - val_loss: 0.3037 - val_acc: 0.4615\n",
      "Epoch 178/200\n",
      "111/111 [==============================] - 0s 760us/step - loss: 0.0489 - acc: 0.4955 - val_loss: 0.2786 - val_acc: 0.7692\n",
      "Epoch 179/200\n",
      "111/111 [==============================] - 0s 753us/step - loss: 0.0466 - acc: 0.5405 - val_loss: 0.2950 - val_acc: 0.6154\n",
      "Epoch 180/200\n",
      "111/111 [==============================] - 0s 767us/step - loss: 0.0448 - acc: 0.5045 - val_loss: 0.2836 - val_acc: 0.5385\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 181/200\n",
      "111/111 [==============================] - 0s 764us/step - loss: 0.0443 - acc: 0.5225 - val_loss: 0.3033 - val_acc: 0.4615\n",
      "Epoch 182/200\n",
      "111/111 [==============================] - 0s 784us/step - loss: 0.0431 - acc: 0.4865 - val_loss: 0.3183 - val_acc: 0.5385\n",
      "Epoch 183/200\n",
      "111/111 [==============================] - 0s 786us/step - loss: 0.0446 - acc: 0.5405 - val_loss: 0.2966 - val_acc: 0.5385\n",
      "Epoch 184/200\n",
      "111/111 [==============================] - 0s 769us/step - loss: 0.0452 - acc: 0.5315 - val_loss: 0.2879 - val_acc: 0.6154\n",
      "Epoch 185/200\n",
      "111/111 [==============================] - 0s 771us/step - loss: 0.0446 - acc: 0.5495 - val_loss: 0.2980 - val_acc: 0.6154\n",
      "Epoch 186/200\n",
      "111/111 [==============================] - 0s 764us/step - loss: 0.0450 - acc: 0.4775 - val_loss: 0.3224 - val_acc: 0.5385\n",
      "Epoch 187/200\n",
      "111/111 [==============================] - 0s 758us/step - loss: 0.0433 - acc: 0.5045 - val_loss: 0.2804 - val_acc: 0.6154\n",
      "Epoch 188/200\n",
      "111/111 [==============================] - 0s 766us/step - loss: 0.0442 - acc: 0.5225 - val_loss: 0.3025 - val_acc: 0.6154\n",
      "Epoch 189/200\n",
      "111/111 [==============================] - 0s 775us/step - loss: 0.0427 - acc: 0.5405 - val_loss: 0.2977 - val_acc: 0.6154\n",
      "Epoch 190/200\n",
      "111/111 [==============================] - 0s 764us/step - loss: 0.0427 - acc: 0.5225 - val_loss: 0.3403 - val_acc: 0.4615\n",
      "Epoch 191/200\n",
      "111/111 [==============================] - 0s 772us/step - loss: 0.0426 - acc: 0.5586 - val_loss: 0.3030 - val_acc: 0.6923\n",
      "Epoch 192/200\n",
      "111/111 [==============================] - 0s 760us/step - loss: 0.0495 - acc: 0.4865 - val_loss: 0.2705 - val_acc: 0.5385\n",
      "Epoch 193/200\n",
      "111/111 [==============================] - 0s 774us/step - loss: 0.0461 - acc: 0.5225 - val_loss: 0.2729 - val_acc: 0.6154\n",
      "Epoch 194/200\n",
      "111/111 [==============================] - 0s 759us/step - loss: 0.0437 - acc: 0.5315 - val_loss: 0.3216 - val_acc: 0.5385\n",
      "Epoch 195/200\n",
      "111/111 [==============================] - 0s 800us/step - loss: 0.0431 - acc: 0.5495 - val_loss: 0.2974 - val_acc: 0.5385\n",
      "Epoch 196/200\n",
      "111/111 [==============================] - 0s 799us/step - loss: 0.0416 - acc: 0.5225 - val_loss: 0.2803 - val_acc: 0.6154\n",
      "Epoch 197/200\n",
      "111/111 [==============================] - 0s 851us/step - loss: 0.0461 - acc: 0.5766 - val_loss: 0.2625 - val_acc: 0.5385\n",
      "Epoch 198/200\n",
      "111/111 [==============================] - 0s 825us/step - loss: 0.0435 - acc: 0.5315 - val_loss: 0.2773 - val_acc: 0.4615\n",
      "Epoch 199/200\n",
      "111/111 [==============================] - 0s 810us/step - loss: 0.0429 - acc: 0.5495 - val_loss: 0.2958 - val_acc: 0.5385\n",
      "Epoch 200/200\n",
      "111/111 [==============================] - 0s 824us/step - loss: 0.0417 - acc: 0.5135 - val_loss: 0.2615 - val_acc: 0.6154\n"
     ]
    }
   ],
   "source": [
    "#神经网络训练及结果\n",
    "history = model.fit(X, y, epochs =epoch, batch_size = batch_size,verbose=1,validation_split=0.1) #训练模型nb_epoch次"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1, 12, 1)\n"
     ]
    }
   ],
   "source": [
    "#筛选测试集\n",
    "test=pd.read_csv(test_file,header=0) \n",
    "test=test['passengers']/scale#选数据集的哪一列\n",
    "test=np.reshape(test.values, (1, window,amount_of_features))\n",
    "print(test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[380.64984 352.39444 424.23895 395.47702 460.3201  533.6566  535.4953\n",
      " 525.6112 ]\n"
     ]
    }
   ],
   "source": [
    "#实际预测\n",
    "predict=model.predict(test)[0,:]*scale\n",
    "#储存文件\n",
    "(pd.DataFrame(predict)).to_csv(save_file)\n",
    "print(predict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 25217,
     "status": "ok",
     "timestamp": 1597492094690,
     "user": {
      "displayName": "Expss Xu",
      "photoUrl": "",
      "userId": "17480852382145563764"
     },
     "user_tz": -480
    },
    "id": "Dx57DzoPotj7"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#迭代图像\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "epochs_range = range(epoch)\n",
    "plt.plot(epochs_range, loss, label='Train Loss')\n",
    "plt.plot(epochs_range, val_loss, label='Validation Loss')\n",
    "plt.legend(loc='upper right')\n",
    "plt.title('Train and Val Loss')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "data_times=np.arange(len(data))\n",
    "all_data=np.hstack((data,predict)) \n",
    "predicted_times=np.arange(len(all_data))\n",
    "plt.figure(figsize=(15, 5))\n",
    "plt.axvline(len(data), linestyle=\"dotted\", linewidth=5, color='r')\n",
    "predicted_lines = plt.plot(predicted_times, all_data, label=\"prediction\", color=\"b\")\n",
    "data_lines = plt.plot(data_times, data, label=\"data\", color=\"k\")\n",
    "plt.legend(handles=[data_lines[0], predicted_lines[0]],loc=\"upper left\")\n",
    "#为训练，验证，预测数据分别创建图例\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 25211,
     "status": "ok",
     "timestamp": 1597492094691,
     "user": {
      "displayName": "Expss Xu",
      "photoUrl": "",
      "userId": "17480852382145563764"
     },
     "user_tz": -480
    },
    "id": "7kbUZXklotkA"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "colab": {
   "collapsed_sections": [],
   "name": "lstm.ipynb",
   "provenance": []
  },
  "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": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}