Lesson 7.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\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>A</th>\n",
       "      <th>AA</th>\n",
       "      <th>AAPL</th>\n",
       "      <th>ABC</th>\n",
       "      <th>ABT</th>\n",
       "      <th>ACS</th>\n",
       "      <th>ADBE</th>\n",
       "      <th>ADI</th>\n",
       "      <th>ADM</th>\n",
       "      <th>...</th>\n",
       "      <th>X</th>\n",
       "      <th>XEL</th>\n",
       "      <th>XL</th>\n",
       "      <th>XLNX</th>\n",
       "      <th>XOM</th>\n",
       "      <th>XRAY</th>\n",
       "      <th>XRX</th>\n",
       "      <th>YHOO</th>\n",
       "      <th>YUM</th>\n",
       "      <th>ZION</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2000-01-03</td>\n",
       "      <td>46.354670</td>\n",
       "      <td>30.172580</td>\n",
       "      <td>3.660058</td>\n",
       "      <td>3.326019</td>\n",
       "      <td>9.639804</td>\n",
       "      <td>0.019</td>\n",
       "      <td>16.274667</td>\n",
       "      <td>32.947692</td>\n",
       "      <td>7.912229</td>\n",
       "      <td>...</td>\n",
       "      <td>25.380447</td>\n",
       "      <td>8.810155</td>\n",
       "      <td>32.639550</td>\n",
       "      <td>35.933610</td>\n",
       "      <td>26.056526</td>\n",
       "      <td>7.151686</td>\n",
       "      <td>19.207609</td>\n",
       "      <td>118.750</td>\n",
       "      <td>7.452527</td>\n",
       "      <td>43.820247</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2000-01-04</td>\n",
       "      <td>42.813681</td>\n",
       "      <td>30.312376</td>\n",
       "      <td>3.351477</td>\n",
       "      <td>3.098946</td>\n",
       "      <td>9.364382</td>\n",
       "      <td>0.019</td>\n",
       "      <td>14.909395</td>\n",
       "      <td>31.280900</td>\n",
       "      <td>7.829816</td>\n",
       "      <td>...</td>\n",
       "      <td>24.433043</td>\n",
       "      <td>9.013020</td>\n",
       "      <td>31.170373</td>\n",
       "      <td>35.116937</td>\n",
       "      <td>25.557438</td>\n",
       "      <td>7.151686</td>\n",
       "      <td>18.314232</td>\n",
       "      <td>110.750</td>\n",
       "      <td>7.302727</td>\n",
       "      <td>41.698321</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2000-01-05</td>\n",
       "      <td>40.157947</td>\n",
       "      <td>32.059823</td>\n",
       "      <td>3.400523</td>\n",
       "      <td>3.339381</td>\n",
       "      <td>9.347165</td>\n",
       "      <td>0.019</td>\n",
       "      <td>15.204175</td>\n",
       "      <td>31.737555</td>\n",
       "      <td>7.706187</td>\n",
       "      <td>...</td>\n",
       "      <td>24.682360</td>\n",
       "      <td>9.360790</td>\n",
       "      <td>31.448326</td>\n",
       "      <td>34.348304</td>\n",
       "      <td>26.950724</td>\n",
       "      <td>7.275321</td>\n",
       "      <td>19.257241</td>\n",
       "      <td>102.625</td>\n",
       "      <td>7.340181</td>\n",
       "      <td>41.648973</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3 rows × 468 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         Date          A         AA      AAPL       ABC       ABT    ACS  \\\n",
       "0  2000-01-03  46.354670  30.172580  3.660058  3.326019  9.639804  0.019   \n",
       "1  2000-01-04  42.813681  30.312376  3.351477  3.098946  9.364382  0.019   \n",
       "2  2000-01-05  40.157947  32.059823  3.400523  3.339381  9.347165  0.019   \n",
       "\n",
       "        ADBE        ADI       ADM    ...              X       XEL         XL  \\\n",
       "0  16.274667  32.947692  7.912229    ...      25.380447  8.810155  32.639550   \n",
       "1  14.909395  31.280900  7.829816    ...      24.433043  9.013020  31.170373   \n",
       "2  15.204175  31.737555  7.706187    ...      24.682360  9.360790  31.448326   \n",
       "\n",
       "        XLNX        XOM      XRAY        XRX     YHOO       YUM       ZION  \n",
       "0  35.933610  26.056526  7.151686  19.207609  118.750  7.452527  43.820247  \n",
       "1  35.116937  25.557438  7.151686  18.314232  110.750  7.302727  41.698321  \n",
       "2  34.348304  26.950724  7.275321  19.257241  102.625  7.340181  41.648973  \n",
       "\n",
       "[3 rows x 468 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('AdjustedClosingPrices.csv')\n",
    "df.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4173, 468)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df['Date'] = pd.to_datetime(df.Date)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Date    datetime64[ns]\n",
       "A              float64\n",
       "AA             float64\n",
       "AAPL           float64\n",
       "ABC            float64\n",
       "ABT            float64\n",
       "ACS            float64\n",
       "ADBE           float64\n",
       "ADI            float64\n",
       "ADM            float64\n",
       "ADP            float64\n",
       "ADSK           float64\n",
       "AEE            float64\n",
       "AEP            float64\n",
       "AES            float64\n",
       "AET            float64\n",
       "AFL            float64\n",
       "AGN            float64\n",
       "AIG            float64\n",
       "AIV            float64\n",
       "AIZ            float64\n",
       "AKAM           float64\n",
       "AKS            float64\n",
       "ALL            float64\n",
       "AMAT           float64\n",
       "AMD            float64\n",
       "AMGN           float64\n",
       "AMP            float64\n",
       "AMT            float64\n",
       "AMZN           float64\n",
       "             ...      \n",
       "VMC            float64\n",
       "VNO            float64\n",
       "VRSN           float64\n",
       "VTR            float64\n",
       "VZ             float64\n",
       "WAT            float64\n",
       "WDC            float64\n",
       "WEC            float64\n",
       "WFC            float64\n",
       "WFR            float64\n",
       "WHR            float64\n",
       "WIN            float64\n",
       "WM             float64\n",
       "WMB            float64\n",
       "WMT            float64\n",
       "WPI            float64\n",
       "WU             float64\n",
       "WY             float64\n",
       "WYN            float64\n",
       "WYNN           float64\n",
       "X              float64\n",
       "XEL            float64\n",
       "XL             float64\n",
       "XLNX           float64\n",
       "XOM            float64\n",
       "XRAY           float64\n",
       "XRX            float64\n",
       "YHOO           float64\n",
       "YUM            float64\n",
       "ZION           float64\n",
       "Length: 468, dtype: object"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>A</th>\n",
       "      <th>AA</th>\n",
       "      <th>AAPL</th>\n",
       "      <th>ABC</th>\n",
       "      <th>ABT</th>\n",
       "      <th>ACS</th>\n",
       "      <th>ADBE</th>\n",
       "      <th>ADI</th>\n",
       "      <th>ADM</th>\n",
       "      <th>ADP</th>\n",
       "      <th>...</th>\n",
       "      <th>X</th>\n",
       "      <th>XEL</th>\n",
       "      <th>XL</th>\n",
       "      <th>XLNX</th>\n",
       "      <th>XOM</th>\n",
       "      <th>XRAY</th>\n",
       "      <th>XRX</th>\n",
       "      <th>YHOO</th>\n",
       "      <th>YUM</th>\n",
       "      <th>ZION</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2000-01-03</th>\n",
       "      <td>46.354670</td>\n",
       "      <td>30.172580</td>\n",
       "      <td>3.660058</td>\n",
       "      <td>3.326019</td>\n",
       "      <td>9.639804</td>\n",
       "      <td>0.019</td>\n",
       "      <td>16.274667</td>\n",
       "      <td>32.947692</td>\n",
       "      <td>7.912229</td>\n",
       "      <td>29.184633</td>\n",
       "      <td>...</td>\n",
       "      <td>25.380447</td>\n",
       "      <td>8.810155</td>\n",
       "      <td>32.639550</td>\n",
       "      <td>35.933610</td>\n",
       "      <td>26.056526</td>\n",
       "      <td>7.151686</td>\n",
       "      <td>19.207609</td>\n",
       "      <td>118.750000</td>\n",
       "      <td>7.452527</td>\n",
       "      <td>43.820247</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2000-01-04</th>\n",
       "      <td>42.813681</td>\n",
       "      <td>30.312376</td>\n",
       "      <td>3.351477</td>\n",
       "      <td>3.098946</td>\n",
       "      <td>9.364382</td>\n",
       "      <td>0.019</td>\n",
       "      <td>14.909395</td>\n",
       "      <td>31.280900</td>\n",
       "      <td>7.829816</td>\n",
       "      <td>29.184633</td>\n",
       "      <td>...</td>\n",
       "      <td>24.433043</td>\n",
       "      <td>9.013020</td>\n",
       "      <td>31.170373</td>\n",
       "      <td>35.116937</td>\n",
       "      <td>25.557438</td>\n",
       "      <td>7.151686</td>\n",
       "      <td>18.314232</td>\n",
       "      <td>110.750000</td>\n",
       "      <td>7.302727</td>\n",
       "      <td>41.698321</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2000-01-05</th>\n",
       "      <td>40.157947</td>\n",
       "      <td>32.059823</td>\n",
       "      <td>3.400523</td>\n",
       "      <td>3.339381</td>\n",
       "      <td>9.347165</td>\n",
       "      <td>0.019</td>\n",
       "      <td>15.204175</td>\n",
       "      <td>31.737555</td>\n",
       "      <td>7.706187</td>\n",
       "      <td>28.904013</td>\n",
       "      <td>...</td>\n",
       "      <td>24.682360</td>\n",
       "      <td>9.360790</td>\n",
       "      <td>31.448326</td>\n",
       "      <td>34.348304</td>\n",
       "      <td>26.950724</td>\n",
       "      <td>7.275321</td>\n",
       "      <td>19.257241</td>\n",
       "      <td>102.625000</td>\n",
       "      <td>7.340181</td>\n",
       "      <td>41.648973</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2000-01-06</th>\n",
       "      <td>38.628883</td>\n",
       "      <td>31.640436</td>\n",
       "      <td>3.106247</td>\n",
       "      <td>3.593170</td>\n",
       "      <td>9.674231</td>\n",
       "      <td>0.019</td>\n",
       "      <td>15.328291</td>\n",
       "      <td>30.869910</td>\n",
       "      <td>7.747397</td>\n",
       "      <td>29.289867</td>\n",
       "      <td>...</td>\n",
       "      <td>24.831950</td>\n",
       "      <td>9.273847</td>\n",
       "      <td>31.686571</td>\n",
       "      <td>31.369849</td>\n",
       "      <td>28.344010</td>\n",
       "      <td>7.284835</td>\n",
       "      <td>18.860185</td>\n",
       "      <td>92.046875</td>\n",
       "      <td>7.277768</td>\n",
       "      <td>42.241139</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4 rows × 467 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                    A         AA      AAPL       ABC       ABT    ACS  \\\n",
       "Date                                                                    \n",
       "2000-01-03  46.354670  30.172580  3.660058  3.326019  9.639804  0.019   \n",
       "2000-01-04  42.813681  30.312376  3.351477  3.098946  9.364382  0.019   \n",
       "2000-01-05  40.157947  32.059823  3.400523  3.339381  9.347165  0.019   \n",
       "2000-01-06  38.628883  31.640436  3.106247  3.593170  9.674231  0.019   \n",
       "\n",
       "                 ADBE        ADI       ADM        ADP    ...              X  \\\n",
       "Date                                                     ...                  \n",
       "2000-01-03  16.274667  32.947692  7.912229  29.184633    ...      25.380447   \n",
       "2000-01-04  14.909395  31.280900  7.829816  29.184633    ...      24.433043   \n",
       "2000-01-05  15.204175  31.737555  7.706187  28.904013    ...      24.682360   \n",
       "2000-01-06  15.328291  30.869910  7.747397  29.289867    ...      24.831950   \n",
       "\n",
       "                 XEL         XL       XLNX        XOM      XRAY        XRX  \\\n",
       "Date                                                                         \n",
       "2000-01-03  8.810155  32.639550  35.933610  26.056526  7.151686  19.207609   \n",
       "2000-01-04  9.013020  31.170373  35.116937  25.557438  7.151686  18.314232   \n",
       "2000-01-05  9.360790  31.448326  34.348304  26.950724  7.275321  19.257241   \n",
       "2000-01-06  9.273847  31.686571  31.369849  28.344010  7.284835  18.860185   \n",
       "\n",
       "                  YHOO       YUM       ZION  \n",
       "Date                                         \n",
       "2000-01-03  118.750000  7.452527  43.820247  \n",
       "2000-01-04  110.750000  7.302727  41.698321  \n",
       "2000-01-05  102.625000  7.340181  41.648973  \n",
       "2000-01-06   92.046875  7.277768  42.241139  \n",
       "\n",
       "[4 rows x 467 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = df.set_index('Date')\n",
    "df.head(4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x181fd51dcf8>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEACAYAAACpoOGTAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhGEYhhGW/wcf1Mg6QELeDQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x181fc1f28d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.IBM.plot()\n",
    "plt.ylabel('IBM')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "184024"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.isnull().sum().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df1 = df.replace(0, np.nan)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# fill the missing value\n",
    "# axis = 0 is add by column 1 means filling by rows\n",
    "df2 = df1.fillna(method='ffill',axis=0).fillna(method='bfill',axis=0)\n",
    "df2.isnull().sum().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# If the price is 0, change that to \n",
    "df_log = np.log(df2)\n",
    "df_log.isnull().sum().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df_r = df_log - df_log.shift()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "467"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_r.isnull().sum().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# drop the first row\n",
    "df_r1 = df_r.drop(df_r.index[0])\n",
    "df_r1.isnull().sum().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "today = df_r1.IBM.shift(periods=1)[2:]\n",
    "yesterday = df_r1.IBM.shift(periods=2)[2:]\n",
    "tomorrow = df_r1.IBM[2:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "#start using tensorflow\n",
    "today_s = today.values\n",
    "yesterday_s = yesterday.values\n",
    "tomorrow_s = tomorrow.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "w1 = tf.Variable([1.0], dtype='float32')\n",
    "w2 = tf.Variable([1.0], dtype='float32')\n",
    "b = tf.Variable([1.0], dtype='float32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "x1 = tf.placeholder(tf.float32)\n",
    "x2 = tf.placeholder(tf.float32)\n",
    "y = tf.placeholder(tf.float32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "loss = tf.reduce_mean(tf.square(y - (x1 * w2 + x2 * w2 + b)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "STEP_SIZE = 0.5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "optimizer = tf.train.GradientDescentOptimizer(STEP_SIZE)\n",
    "train = optimizer.minimize(loss)\n",
    "init = tf.global_variables_initializer()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sess = tf.Session()\n",
    "MAXSTEPS = 100000\n",
    "MOD = 10000\n",
    "#initialize all variables\n",
    "sess.run(init)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 1.00104\n",
      "10000 0.000272223\n",
      "20000 0.000272208\n",
      "30000 0.000272208\n",
      "40000 0.000272208\n",
      "50000 0.000272208\n",
      "60000 0.000272208\n",
      "70000 0.000272208\n",
      "80000 0.000272208\n",
      "90000 0.000272208\n",
      "100000 0.000272208\n"
     ]
    }
   ],
   "source": [
    "for step in range(MAXSTEPS+1):\n",
    "    (_, w1opt, w2opt, bopt, loss_opt) = sess.run([train,w1,w2,b,loss],feed_dict={x1:yesterday_s, x2:today_s, y:tomorrow_s})\n",
    "    if (step % MOD) == 0:\n",
    "        print(step, loss_opt)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.00027268764526033797"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tomorrow_s.var()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}