hw3/hw3.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Stock Prediction\n",
    "\n",
    "\n",
    "\n",
    "**利用三種不同的model預測股票的漲跌**\n",
    "\n",
    "1.logistic regression\n",
    "\n",
    "2.support vector machine\n",
    "\n",
    "3.neural network\n",
    "\n",
    "**What We Care**\n",
    "\n",
    "我們在乎的是close price的高低\n",
    "是我們要train的目標"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from sklearn import datasets\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "%matplotlib inline\n",
    "\n",
    "from sklearn.metrics import accuracy_score\n",
    "from keras import models, layers, optimizers, regularizers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "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>Date</th>\n",
       "      <th>Open Price</th>\n",
       "      <th>Close Price</th>\n",
       "      <th>High Price</th>\n",
       "      <th>Low Price</th>\n",
       "      <th>Volume</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>02-Jan-2009</td>\n",
       "      <td>902.99</td>\n",
       "      <td>931.80</td>\n",
       "      <td>934.73</td>\n",
       "      <td>899.35</td>\n",
       "      <td>4048270080</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>05-Jan-2009</td>\n",
       "      <td>929.17</td>\n",
       "      <td>927.45</td>\n",
       "      <td>936.63</td>\n",
       "      <td>919.53</td>\n",
       "      <td>5413910016</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>06-Jan-2009</td>\n",
       "      <td>931.17</td>\n",
       "      <td>934.70</td>\n",
       "      <td>943.85</td>\n",
       "      <td>927.28</td>\n",
       "      <td>5392620032</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>07-Jan-2009</td>\n",
       "      <td>927.45</td>\n",
       "      <td>906.65</td>\n",
       "      <td>927.45</td>\n",
       "      <td>902.37</td>\n",
       "      <td>4704940032</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>08-Jan-2009</td>\n",
       "      <td>905.73</td>\n",
       "      <td>909.73</td>\n",
       "      <td>910.00</td>\n",
       "      <td>896.81</td>\n",
       "      <td>4991549952</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          Date  Open Price  Close Price  High Price  Low Price      Volume\n",
       "0  02-Jan-2009      902.99       931.80      934.73     899.35  4048270080\n",
       "1  05-Jan-2009      929.17       927.45      936.63     919.53  5413910016\n",
       "2  06-Jan-2009      931.17       934.70      943.85     927.28  5392620032\n",
       "3  07-Jan-2009      927.45       906.65      927.45     902.37  4704940032\n",
       "4  08-Jan-2009      905.73       909.73      910.00     896.81  4991549952"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df=pd.read_csv('./train_stock.csv')\n",
    "df_test=pd.read_csv('./test_stock.csv')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Date:當天日期\n",
    "\n",
    "Open Price:開盤價\n",
    "\n",
    "Close Price:收盤價\n",
    "\n",
    "High Price:當日最高價\n",
    "\n",
    "Low Price:當日最低價\n",
    "\n",
    "Volume:當日的總交易量|"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 利用圖表顯示歷年來的漲跌情形\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#將日期中的年份分出來\n",
    "df['Date'] = pd.to_datetime(df['Date'])\n",
    "df['year'] = df['Date'].dt.year\n",
    "\n",
    "\n",
    "plt.figure(figsize=(15,7))\n",
    "plt.plot_date(df['Date'],df['Open Price'],'-')\n",
    "plt.xlabel('year')\n",
    "plt.ylabel('Open Price')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "如果當天的Close Price比昨天高\n",
    "\n",
    "視為漲，設為1\n",
    "\n",
    "如果當天的Close Price比昨天低\n",
    "\n",
    "視為跌，設為0\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['up/down']=np.zeros(len(df))\n",
    "df_test['up/down']=np.zeros(len(df_test))\n",
    "\n",
    "updown=np.zeros(len(df))\n",
    "for index in range(len(df)-1):\n",
    "    today=df.iloc[index,2]\n",
    "    yesterday=df.iloc[index-1,2]\n",
    "    if today>yesterday:\n",
    "        updown[index-1]=1\n",
    "    else:\n",
    "        updown[index-1]=0\n",
    "df['up/down']=updown\n",
    "\n",
    "updown_test=np.zeros(len(df_test))\n",
    "for index in range(len(df_test)-1):\n",
    "    today=df_test.iloc[index,2]\n",
    "    yesterday=df_test.iloc[index-1,2]\n",
    "    if today>yesterday:\n",
    "        updown_test[index-1]=1\n",
    "    else:\n",
    "        updown_test[index-1]=0\n",
    "df_test['up/down']=updown_test   \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "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>Date</th>\n",
       "      <th>Open Price</th>\n",
       "      <th>Close Price</th>\n",
       "      <th>High Price</th>\n",
       "      <th>Low Price</th>\n",
       "      <th>Volume</th>\n",
       "      <th>year</th>\n",
       "      <th>up/down</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2009-01-02</td>\n",
       "      <td>902.99</td>\n",
       "      <td>931.80</td>\n",
       "      <td>934.73</td>\n",
       "      <td>899.35</td>\n",
       "      <td>4048270080</td>\n",
       "      <td>2009</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2009-01-05</td>\n",
       "      <td>929.17</td>\n",
       "      <td>927.45</td>\n",
       "      <td>936.63</td>\n",
       "      <td>919.53</td>\n",
       "      <td>5413910016</td>\n",
       "      <td>2009</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2009-01-06</td>\n",
       "      <td>931.17</td>\n",
       "      <td>934.70</td>\n",
       "      <td>943.85</td>\n",
       "      <td>927.28</td>\n",
       "      <td>5392620032</td>\n",
       "      <td>2009</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2009-01-07</td>\n",
       "      <td>927.45</td>\n",
       "      <td>906.65</td>\n",
       "      <td>927.45</td>\n",
       "      <td>902.37</td>\n",
       "      <td>4704940032</td>\n",
       "      <td>2009</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2009-01-08</td>\n",
       "      <td>905.73</td>\n",
       "      <td>909.73</td>\n",
       "      <td>910.00</td>\n",
       "      <td>896.81</td>\n",
       "      <td>4991549952</td>\n",
       "      <td>2009</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Date  Open Price  Close Price  High Price  Low Price      Volume  \\\n",
       "0 2009-01-02      902.99       931.80      934.73     899.35  4048270080   \n",
       "1 2009-01-05      929.17       927.45      936.63     919.53  5413910016   \n",
       "2 2009-01-06      931.17       934.70      943.85     927.28  5392620032   \n",
       "3 2009-01-07      927.45       906.65      927.45     902.37  4704940032   \n",
       "4 2009-01-08      905.73       909.73      910.00     896.81  4991549952   \n",
       "\n",
       "   year  up/down  \n",
       "0  2009      0.0  \n",
       "1  2009      1.0  \n",
       "2  2009      0.0  \n",
       "3  2009      1.0  \n",
       "4  2009      0.0  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_x=df[['Open Price','Close Price','High Price','Low Price','Volume']]\n",
    "train_y=df[['up/down']]\n",
    "test_x=df_test[['Open Price','Close Price','High Price','Low Price','Volume']]\n",
    "test_y=df_test[['up/down']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## logistic regression"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "準確率只有0.519"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.5198412698412699\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "logic_model= LogisticRegression()\n",
    "logic_result= logic_model.fit(train_x, train_y)\n",
    "\n",
    "logic_predict = logic_model.predict(test_x)\n",
    "print(accuracy_score(test_y,logic_predict))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 試著減少feature的數量\n",
    "\n",
    "只用 high price 與 low price 準確率變成0.508"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.5079365079365079\n"
     ]
    }
   ],
   "source": [
    "\n",
    "logic_result_less= logic_model.fit(df[['High Price','Low Price']], train_y)\n",
    "logic_predict_less = logic_model.predict(df_test[['High Price','Low Price']])\n",
    "print(accuracy_score(test_y,logic_predict_less))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "\n",
    "利用high price , open price和close price 準確率剩下0.46"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.4603174603174603\n"
     ]
    }
   ],
   "source": [
    "logic_result_price= logic_model.fit(df[['High Price','Open Price','Close Price']], train_y)\n",
    "logic_predict_price = logic_model.predict(df_test[['High Price','Open Price','Close Price']])\n",
    "print(accuracy_score(test_y,logic_predict_price))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "留下除了Volume以外的feature，準確率為0.48"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.48412698412698413\n"
     ]
    }
   ],
   "source": [
    "logic_result_price= logic_model.fit(df[['Open Price','Close Price','High Price','Low Price']], train_y)\n",
    "logic_predict_price = logic_model.predict(df_test[['Open Price','Close Price','High Price','Low Price']])\n",
    "print(accuracy_score(test_y,logic_predict_price))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1., 0., 1., 0., 1., 1., 1., 0., 0., 1., 0., 1., 1., 0., 1., 1., 1.,\n",
       "       0., 1., 1., 1., 1., 1., 1., 0., 1., 1., 0., 0., 0., 0., 0., 1., 1.,\n",
       "       1., 1., 0., 0., 1., 1., 1., 0., 0., 1., 0., 0., 0., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 0., 1., 1., 0., 1., 0., 0., 1., 1., 1., 0.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 0., 1., 1., 1., 0., 1., 0.,\n",
       "       0., 1., 1., 0., 0., 1., 1., 1., 1., 1., 1., 1., 1., 0., 0., 1., 1.,\n",
       "       0., 1., 0., 1., 1., 0., 1., 0., 1., 1., 1., 1., 0., 0., 0., 1., 1.,\n",
       "       1., 1., 1., 1., 0., 1., 0., 1., 0., 0., 0., 1., 1., 0., 1., 1., 0.,\n",
       "       1., 1., 1., 1., 1., 0., 1., 1., 1., 1., 1., 0., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 0., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 0., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 0., 1., 1., 1., 0., 1., 0., 0., 1., 1., 1.,\n",
       "       0., 1., 0., 1., 1., 0., 1., 0., 1., 0., 0., 0., 1., 1., 1., 1., 1.,\n",
       "       0., 0., 1., 1., 1., 1., 0., 0., 0., 1., 0., 1., 1., 1., 0., 1., 0.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 0., 0., 1., 0.])"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "logic_predict_price"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## support vector machine\n",
    "\n",
    "與logistic regression一開始的結果一樣都是0.519，因為svm預設的kernel是linear的，\n",
    "\n",
    "所以做出來的結果與linear regression一樣是可預見的"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.5198412698412699\n"
     ]
    }
   ],
   "source": [
    "from sklearn import svm\n",
    "\n",
    "svm_model = svm.SVC()\n",
    "svm_result= svm_model.fit(train_x, train_y)\n",
    "\n",
    "svm_predict=svm_model.predict(test_x)\n",
    "print(accuracy_score(test_y,svm_predict))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "將kernel改成polynomial，準確度卻沒有改變"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.5198412698412699\n"
     ]
    }
   ],
   "source": [
    "svm_model_poly = svm.SVC(kernel='poly')\n",
    "svm_result_poly= svm_model_poly.fit(train_x, train_y)\n",
    "\n",
    "svm_predict_poly = svm_model_poly.predict(test_x)\n",
    "print(accuracy_score(test_y, svm_predict_poly))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "試著調整penalty，準確度仍然沒有改變"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.5198412698412699\n"
     ]
    }
   ],
   "source": [
    "svm_model_penalty = svm.SVC(kernel='poly',C=3.0)\n",
    "svm_result_lpenalty= svm_model_penalty.fit(train_x, train_y)\n",
    "\n",
    "svm_predict_penalty = svm_model_penalty.predict(test_x)\n",
    "print(accuracy_score(test_y, svm_predict_penalty))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "只用open price 與 close price與volume\n",
    "\n",
    "依然維持0.519"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.5198412698412699\n"
     ]
    }
   ],
   "source": [
    "svm_model_less = svm.SVC(kernel='poly',C=3.0)\n",
    "svm_result_less= svm_model_less.fit(df[['Open Price','Close Price','Volume']], train_y)\n",
    "\n",
    "svm_predict_less = svm_model_less.predict(df_test[['Open Price','Close Price','Volume']])\n",
    "print(accuracy_score(test_y, svm_predict_less))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "令人意外的是\n",
    "\n",
    "SVM做出來的prediction結果全部為1\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,\n",
       "       1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.])"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svm_predict_less"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## neural network\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "hidden_units = 40    # how many neurons in the hidden layer\n",
    "activation = 'sigmoid'  # activation function for hidden layer\n",
    "l2 = 0.4              # regularization - how much we penalize large parameter values\n",
    "learning_rate = 0.1  # how big our steps are in gradient descent\n",
    "epochs = 100          # how many epochs to train for\n",
    "batch_size = 32      # how many samples to use for each gradient descent update\n",
    "class_weight = {     # define how each class weight\n",
    "    0: 1.,\n",
    "    1: 160 / 302\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\user\\Anaconda3\\lib\\site-packages\\tensorflow\\python\\framework\\op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 40)                240       \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 41        \n",
      "=================================================================\n",
      "Total params: 281\n",
      "Trainable params: 281\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "WARNING:tensorflow:From C:\\Users\\user\\Anaconda3\\lib\\site-packages\\tensorflow\\python\\ops\\math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n"
     ]
    }
   ],
   "source": [
    "\n",
    "\n",
    "model = models.Sequential()\n",
    "\n",
    "model.add(layers.Dense(input_dim=train_x.shape[1],\n",
    "                       units=40, \n",
    "                       activation='relu'))\n",
    "\n",
    "model.add(layers.Dense(units=1,\n",
    "                       activation='sigmoid'))\n",
    "\n",
    "model.compile(loss='binary_crossentropy',\n",
    "              optimizer='adam',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "model.summary()\n",
    "history = model.fit(train_x, \n",
    "                    train_y, \n",
    "                    epochs=50, \n",
    "                    validation_data=(test_x, test_y),\n",
    "                    verbose=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "2264/2264 [==============================] - 0s 41us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 2/100\n",
      "2264/2264 [==============================] - 0s 39us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 3/100\n",
      "2264/2264 [==============================] - 0s 38us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 4/100\n",
      "2264/2264 [==============================] - 0s 38us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 5/100\n",
      "2264/2264 [==============================] - 0s 36us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 6/100\n",
      "2264/2264 [==============================] - 0s 39us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 7/100\n",
      "2264/2264 [==============================] - 0s 40us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 8/100\n",
      "2264/2264 [==============================] - 0s 38us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 9/100\n",
      "2264/2264 [==============================] - 0s 37us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 10/100\n",
      "2264/2264 [==============================] - 0s 41us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 11/100\n",
      "2264/2264 [==============================] - 0s 43us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 12/100\n",
      "2264/2264 [==============================] - 0s 40us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 13/100\n",
      "2264/2264 [==============================] - 0s 40us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 14/100\n",
      "2264/2264 [==============================] - 0s 37us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 15/100\n",
      "2264/2264 [==============================] - 0s 33us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 16/100\n",
      "2264/2264 [==============================] - 0s 41us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 17/100\n",
      "2264/2264 [==============================] - 0s 35us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 18/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 19/100\n",
      "2264/2264 [==============================] - 0s 32us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 20/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 21/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 22/100\n",
      "2264/2264 [==============================] - 0s 32us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 23/100\n",
      "2264/2264 [==============================] - 0s 32us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 24/100\n",
      "2264/2264 [==============================] - 0s 39us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 25/100\n",
      "2264/2264 [==============================] - 0s 33us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 26/100\n",
      "2264/2264 [==============================] - 0s 32us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 27/100\n",
      "2264/2264 [==============================] - 0s 33us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 28/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 29/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 30/100\n",
      "2264/2264 [==============================] - 0s 46us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 31/100\n",
      "2264/2264 [==============================] - 0s 50us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 32/100\n",
      "2264/2264 [==============================] - 0s 43us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 33/100\n",
      "2264/2264 [==============================] - 0s 44us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 34/100\n",
      "2264/2264 [==============================] - 0s 46us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 35/100\n",
      "2264/2264 [==============================] - 0s 48us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 36/100\n",
      "2264/2264 [==============================] - 0s 47us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 37/100\n",
      "2264/2264 [==============================] - 0s 44us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 38/100\n",
      "2264/2264 [==============================] - 0s 41us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 39/100\n",
      "2264/2264 [==============================] - 0s 39us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 40/100\n",
      "2264/2264 [==============================] - 0s 43us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 41/100\n",
      "2264/2264 [==============================] - 0s 45us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 42/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 43/100\n",
      "2264/2264 [==============================] - 0s 35us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 44/100\n",
      "2264/2264 [==============================] - 0s 32us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 45/100\n",
      "2264/2264 [==============================] - 0s 32us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 46/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 47/100\n",
      "2264/2264 [==============================] - 0s 35us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 48/100\n",
      "2264/2264 [==============================] - 0s 33us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 49/100\n",
      "2264/2264 [==============================] - 0s 35us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 50/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 51/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 52/100\n",
      "2264/2264 [==============================] - 0s 33us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 53/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 54/100\n",
      "2264/2264 [==============================] - 0s 35us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 55/100\n",
      "2264/2264 [==============================] - 0s 44us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 56/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 57/100\n",
      "2264/2264 [==============================] - 0s 38us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 58/100\n",
      "2264/2264 [==============================] - 0s 33us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 59/100\n",
      "2264/2264 [==============================] - 0s 33us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 60/100\n",
      "2264/2264 [==============================] - 0s 36us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 61/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 62/100\n",
      "2264/2264 [==============================] - 0s 36us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 63/100\n",
      "2264/2264 [==============================] - 0s 38us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 64/100\n",
      "2264/2264 [==============================] - 0s 41us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 65/100\n",
      "2264/2264 [==============================] - 0s 40us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 66/100\n",
      "2264/2264 [==============================] - 0s 36us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 67/100\n",
      "2264/2264 [==============================] - 0s 37us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 68/100\n",
      "2264/2264 [==============================] - 0s 40us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 69/100\n",
      "2264/2264 [==============================] - 0s 37us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 70/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 71/100\n",
      "2264/2264 [==============================] - 0s 35us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 72/100\n",
      "2264/2264 [==============================] - 0s 35us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 73/100\n",
      "2264/2264 [==============================] - 0s 40us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 74/100\n",
      "2264/2264 [==============================] - 0s 42us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 75/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 76/100\n",
      "2264/2264 [==============================] - 0s 36us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 77/100\n",
      "2264/2264 [==============================] - 0s 37us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 78/100\n",
      "2264/2264 [==============================] - 0s 39us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 79/100\n",
      "2264/2264 [==============================] - 0s 35us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 80/100\n",
      "2264/2264 [==============================] - 0s 41us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 81/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 82/100\n",
      "2264/2264 [==============================] - 0s 38us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 83/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2264/2264 [==============================] - 0s 44us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 84/100\n",
      "2264/2264 [==============================] - 0s 66us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 85/100\n",
      "2264/2264 [==============================] - 0s 41us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 86/100\n",
      "2264/2264 [==============================] - 0s 42us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 87/100\n",
      "2264/2264 [==============================] - 0s 39us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 88/100\n",
      "2264/2264 [==============================] - 0s 36us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 89/100\n",
      "2264/2264 [==============================] - 0s 35us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 90/100\n",
      "2264/2264 [==============================] - 0s 37us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 91/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 92/100\n",
      "2264/2264 [==============================] - 0s 41us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 93/100\n",
      "2264/2264 [==============================] - 0s 49us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 94/100\n",
      "2264/2264 [==============================] - 0s 41us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 95/100\n",
      "2264/2264 [==============================] - 0s 45us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 96/100\n",
      "2264/2264 [==============================] - 0s 41us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 97/100\n",
      "2264/2264 [==============================] - 0s 39us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 98/100\n",
      "2264/2264 [==============================] - 0s 43us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 99/100\n",
      "2264/2264 [==============================] - 0s 40us/step - loss: 4.6620 - acc: 0.4541\n",
      "Epoch 100/100\n",
      "2264/2264 [==============================] - 0s 34us/step - loss: 4.6620 - acc: 0.4541\n",
      "2264/2264 [==============================] - 0s 14us/step\n",
      "252/252 [==============================] - 0s 32us/step\n"
     ]
    }
   ],
   "source": [
    "history=model.fit(train_x, train_y, epochs=epochs, batch_size=batch_size, class_weight=class_weight)\n",
    "train_acc=model.evaluate(train_x, train_y, batch_size=batch_size)[1]\n",
    "test_acc=model.evaluate(test_x, test_y, batch_size=batch_size)[1]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "利用NN做出來的結果甚至比logistic regression還差\n",
    "\n",
    "只有0.456的準確度\n",
    "\n",
    "而且似乎愈train愈糟\n",
    "\n",
    "沒有趨於穩定的感覺\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training accuracy: 0.4540636044508998\n",
      "Testing accuracy: 0.4801587329970466\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Training accuracy: %s\" % train_acc)\n",
    "print(\"Testing accuracy: %s\" % test_acc)\n",
    "\n",
    "losses=history.history['loss']\n",
    "plt.plot(range(len(losses)), losses)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Discussion\n",
    "\n",
    "令人意外的是感覺應該功力最高的Neural Network做出來的結果竟然是**最差的**\n",
    "\n",
    "我猜可能是因為選擇的model不適合，無法fitting出更好的準確度。\n",
    "\n",
    "另外如果把Volume放入特徵訓練，似乎都會導向相同的結果，\n",
    "\n",
    "都會把**全部的預測結果猜成1**，不知道是不是因為Volume這欄位的數值\n",
    "\n",
    "比其他欄位**大非常多**，才導致這樣的結果。\n",
    "\n",
    "\n",
    "### Accuracy\n",
    "\n",
    "這次所分析的data總共兩千多筆，而且feature的項目也不多\n",
    "\n",
    "我認為要能精準的預測股票市場的漲跌還是太困難了。\n",
    "\n",
    "或者應該說，現在也還沒有人能夠準確的分析以及預測股票，\n",
    "\n",
    "股票漲幅**牽涉的原因**太多了，沒人說得準。\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n"
   ]
  }
 ],
 "metadata": {
  "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}