freestyle.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Machine Learning for Finance Freestyle\n",
    "\n",
    "In this lab you'll be given the opportunity to apply everything you have learned to build a trading strategy for SP500 stocks. First, let's introduce the dataset you'll be using."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The Data\n",
    "\n",
    "Use BigQuery's magic function to pull data as follows:\n",
    "\n",
    "    Dataset Name: ml4f\n",
    "    Table Name: percent_change_sp500\n",
    "\n",
    "The following query will pull 10 rows of data from the table:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%bigquery df\n",
    "SELECT \n",
    "    *\n",
    "FROM\n",
    "    `cloud-training-prod-bucket.ml4f.percent_change_sp500`\n",
    "LIMIT\n",
    "    10000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>symbol</th>\n",
       "      <th>Date</th>\n",
       "      <th>Open</th>\n",
       "      <th>Close</th>\n",
       "      <th>tomorrow_close</th>\n",
       "      <th>tomo_close_m_close</th>\n",
       "      <th>close_MIN_prior_5_days</th>\n",
       "      <th>close_MIN_prior_20_days</th>\n",
       "      <th>close_MIN_prior_260_days</th>\n",
       "      <th>close_MAX_prior_5_days</th>\n",
       "      <th>...</th>\n",
       "      <th>close_STDDEV_prior_20_days</th>\n",
       "      <th>close_STDDEV_prior_260_days</th>\n",
       "      <th>close_values_prior_260</th>\n",
       "      <th>days_on_market</th>\n",
       "      <th>scaled_change</th>\n",
       "      <th>s_p_scaled_change</th>\n",
       "      <th>normalized_change</th>\n",
       "      <th>company</th>\n",
       "      <th>industry</th>\n",
       "      <th>direction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>AA</td>\n",
       "      <td>2004-09-13</td>\n",
       "      <td>30.76</td>\n",
       "      <td>30.79</td>\n",
       "      <td>30.70</td>\n",
       "      <td>-0.09</td>\n",
       "      <td>0.998701</td>\n",
       "      <td>0.962325</td>\n",
       "      <td>0.849627</td>\n",
       "      <td>1.081195</td>\n",
       "      <td>...</td>\n",
       "      <td>0.029852</td>\n",
       "      <td>0.092995</td>\n",
       "      <td>[28.56, 28.62, 28.29, 28.09, 27.99, 28.2, 28.5...</td>\n",
       "      <td>10748</td>\n",
       "      <td>-0.002923</td>\n",
       "      <td>0.002229</td>\n",
       "      <td>-0.005153</td>\n",
       "      <td>Alcoa Inc</td>\n",
       "      <td>Materials</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>CSC</td>\n",
       "      <td>1994-09-22</td>\n",
       "      <td>42.50</td>\n",
       "      <td>42.87</td>\n",
       "      <td>43.37</td>\n",
       "      <td>0.50</td>\n",
       "      <td>0.994168</td>\n",
       "      <td>0.982505</td>\n",
       "      <td>0.763937</td>\n",
       "      <td>1.029158</td>\n",
       "      <td>...</td>\n",
       "      <td>0.020679</td>\n",
       "      <td>0.605874</td>\n",
       "      <td>[90.37, 89.25, 90.87, 90.37, 90.12, 89.75, 88....</td>\n",
       "      <td>3219</td>\n",
       "      <td>0.011663</td>\n",
       "      <td>-0.003469</td>\n",
       "      <td>0.015132</td>\n",
       "      <td>Computer Sciences</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>UP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ALTR</td>\n",
       "      <td>1992-09-08</td>\n",
       "      <td>9.25</td>\n",
       "      <td>9.00</td>\n",
       "      <td>8.75</td>\n",
       "      <td>-0.25</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.972222</td>\n",
       "      <td>0.972222</td>\n",
       "      <td>1.042222</td>\n",
       "      <td>...</td>\n",
       "      <td>0.117427</td>\n",
       "      <td>0.805250</td>\n",
       "      <td>[23.5, 23.37, 23.37, 23.5, 23.5, 22.87, 22.5, ...</td>\n",
       "      <td>1122</td>\n",
       "      <td>-0.027778</td>\n",
       "      <td>0.004633</td>\n",
       "      <td>-0.032411</td>\n",
       "      <td>Altera Corp</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>DOWN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>APD</td>\n",
       "      <td>2010-02-17</td>\n",
       "      <td>70.48</td>\n",
       "      <td>70.18</td>\n",
       "      <td>69.95</td>\n",
       "      <td>-0.23</td>\n",
       "      <td>0.968225</td>\n",
       "      <td>0.968225</td>\n",
       "      <td>0.624109</td>\n",
       "      <td>1.003277</td>\n",
       "      <td>...</td>\n",
       "      <td>0.072852</td>\n",
       "      <td>0.153308</td>\n",
       "      <td>[52.55, 54.13, 55.4, 55.85, 52.8, 52.25, 54.53...</td>\n",
       "      <td>6778</td>\n",
       "      <td>-0.003277</td>\n",
       "      <td>0.006585</td>\n",
       "      <td>-0.009862</td>\n",
       "      <td>Air Products &amp; Chemicals Inc</td>\n",
       "      <td>Materials</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>CBE</td>\n",
       "      <td>1989-07-26</td>\n",
       "      <td>22.86</td>\n",
       "      <td>23.32</td>\n",
       "      <td>23.59</td>\n",
       "      <td>0.27</td>\n",
       "      <td>0.984134</td>\n",
       "      <td>0.952830</td>\n",
       "      <td>0.866638</td>\n",
       "      <td>1.051029</td>\n",
       "      <td>...</td>\n",
       "      <td>0.024737</td>\n",
       "      <td>0.320813</td>\n",
       "      <td>[38.76, 38.58, 38.41, 38.14, 37.61, 38.23, 38....</td>\n",
       "      <td>1913</td>\n",
       "      <td>0.011578</td>\n",
       "      <td>0.011655</td>\n",
       "      <td>-0.000077</td>\n",
       "      <td>Cooper Industries Plc</td>\n",
       "      <td>Industrials</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 26 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "  symbol        Date   Open  Close  tomorrow_close  tomo_close_m_close  \\\n",
       "0     AA  2004-09-13  30.76  30.79           30.70               -0.09   \n",
       "1    CSC  1994-09-22  42.50  42.87           43.37                0.50   \n",
       "2   ALTR  1992-09-08   9.25   9.00            8.75               -0.25   \n",
       "3    APD  2010-02-17  70.48  70.18           69.95               -0.23   \n",
       "4    CBE  1989-07-26  22.86  23.32           23.59                0.27   \n",
       "\n",
       "   close_MIN_prior_5_days  close_MIN_prior_20_days  close_MIN_prior_260_days  \\\n",
       "0                0.998701                 0.962325                  0.849627   \n",
       "1                0.994168                 0.982505                  0.763937   \n",
       "2                1.000000                 0.972222                  0.972222   \n",
       "3                0.968225                 0.968225                  0.624109   \n",
       "4                0.984134                 0.952830                  0.866638   \n",
       "\n",
       "   close_MAX_prior_5_days  ...  close_STDDEV_prior_20_days  \\\n",
       "0                1.081195  ...                    0.029852   \n",
       "1                1.029158  ...                    0.020679   \n",
       "2                1.042222  ...                    0.117427   \n",
       "3                1.003277  ...                    0.072852   \n",
       "4                1.051029  ...                    0.024737   \n",
       "\n",
       "   close_STDDEV_prior_260_days  \\\n",
       "0                     0.092995   \n",
       "1                     0.605874   \n",
       "2                     0.805250   \n",
       "3                     0.153308   \n",
       "4                     0.320813   \n",
       "\n",
       "                              close_values_prior_260  days_on_market  \\\n",
       "0  [28.56, 28.62, 28.29, 28.09, 27.99, 28.2, 28.5...           10748   \n",
       "1  [90.37, 89.25, 90.87, 90.37, 90.12, 89.75, 88....            3219   \n",
       "2  [23.5, 23.37, 23.37, 23.5, 23.5, 22.87, 22.5, ...            1122   \n",
       "3  [52.55, 54.13, 55.4, 55.85, 52.8, 52.25, 54.53...            6778   \n",
       "4  [38.76, 38.58, 38.41, 38.14, 37.61, 38.23, 38....            1913   \n",
       "\n",
       "   scaled_change  s_p_scaled_change  normalized_change  \\\n",
       "0      -0.002923           0.002229          -0.005153   \n",
       "1       0.011663          -0.003469           0.015132   \n",
       "2      -0.027778           0.004633          -0.032411   \n",
       "3      -0.003277           0.006585          -0.009862   \n",
       "4       0.011578           0.011655          -0.000077   \n",
       "\n",
       "                        company                industry  direction  \n",
       "0                     Alcoa Inc               Materials       STAY  \n",
       "1             Computer Sciences  Information Technology         UP  \n",
       "2                   Altera Corp  Information Technology       DOWN  \n",
       "3  Air Products & Chemicals Inc               Materials       STAY  \n",
       "4         Cooper Industries Plc             Industrials       STAY  \n",
       "\n",
       "[5 rows x 26 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you can see, the table contains daily open and close data for SP500 stocks. The table also contains some features that have been generated for you using [navigation functions](https://cloud.google.com/bigquery/docs/reference/standard-sql/navigation_functions) and [analytic functions](https://cloud.google.com/bigquery/docs/reference/standard-sql/analytic-function-concepts). Let's dig into the schema a bit more. "
   ]
  },
  {
   "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>table_catalog</th>\n",
       "      <th>table_schema</th>\n",
       "      <th>table_name</th>\n",
       "      <th>column_name</th>\n",
       "      <th>ordinal_position</th>\n",
       "      <th>is_nullable</th>\n",
       "      <th>data_type</th>\n",
       "      <th>is_hidden</th>\n",
       "      <th>is_system_defined</th>\n",
       "      <th>is_partitioning_column</th>\n",
       "      <th>clustering_ordinal_position</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>symbol</td>\n",
       "      <td>1</td>\n",
       "      <td>YES</td>\n",
       "      <td>STRING</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>Date</td>\n",
       "      <td>2</td>\n",
       "      <td>YES</td>\n",
       "      <td>DATE</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>Open</td>\n",
       "      <td>3</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>Close</td>\n",
       "      <td>4</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>tomorrow_close</td>\n",
       "      <td>5</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>tomo_close_m_close</td>\n",
       "      <td>6</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_MIN_prior_5_days</td>\n",
       "      <td>7</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_MIN_prior_20_days</td>\n",
       "      <td>8</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_MIN_prior_260_days</td>\n",
       "      <td>9</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_MAX_prior_5_days</td>\n",
       "      <td>10</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_MAX_prior_20_days</td>\n",
       "      <td>11</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_MAX_prior_260_days</td>\n",
       "      <td>12</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_AVG_prior_5_days</td>\n",
       "      <td>13</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_AVG_prior_20_days</td>\n",
       "      <td>14</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_AVG_prior_260_days</td>\n",
       "      <td>15</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_STDDEV_prior_5_days</td>\n",
       "      <td>16</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_STDDEV_prior_20_days</td>\n",
       "      <td>17</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_STDDEV_prior_260_days</td>\n",
       "      <td>18</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>close_values_prior_260</td>\n",
       "      <td>19</td>\n",
       "      <td>NO</td>\n",
       "      <td>ARRAY&lt;FLOAT64&gt;</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>days_on_market</td>\n",
       "      <td>20</td>\n",
       "      <td>YES</td>\n",
       "      <td>INT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>scaled_change</td>\n",
       "      <td>21</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>s_p_scaled_change</td>\n",
       "      <td>22</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>normalized_change</td>\n",
       "      <td>23</td>\n",
       "      <td>YES</td>\n",
       "      <td>FLOAT64</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>company</td>\n",
       "      <td>24</td>\n",
       "      <td>YES</td>\n",
       "      <td>STRING</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>industry</td>\n",
       "      <td>25</td>\n",
       "      <td>YES</td>\n",
       "      <td>STRING</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>cloud-training-prod-bucket</td>\n",
       "      <td>ml4f</td>\n",
       "      <td>percent_change_sp500</td>\n",
       "      <td>direction</td>\n",
       "      <td>26</td>\n",
       "      <td>YES</td>\n",
       "      <td>STRING</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NO</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 table_catalog table_schema            table_name  \\\n",
       "0   cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "1   cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "2   cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "3   cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "4   cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "5   cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "6   cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "7   cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "8   cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "9   cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "10  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "11  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "12  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "13  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "14  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "15  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "16  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "17  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "18  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "19  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "20  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "21  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "22  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "23  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "24  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "25  cloud-training-prod-bucket         ml4f  percent_change_sp500   \n",
       "\n",
       "                    column_name  ordinal_position is_nullable       data_type  \\\n",
       "0                        symbol                 1         YES          STRING   \n",
       "1                          Date                 2         YES            DATE   \n",
       "2                          Open                 3         YES         FLOAT64   \n",
       "3                         Close                 4         YES         FLOAT64   \n",
       "4                tomorrow_close                 5         YES         FLOAT64   \n",
       "5            tomo_close_m_close                 6         YES         FLOAT64   \n",
       "6        close_MIN_prior_5_days                 7         YES         FLOAT64   \n",
       "7       close_MIN_prior_20_days                 8         YES         FLOAT64   \n",
       "8      close_MIN_prior_260_days                 9         YES         FLOAT64   \n",
       "9        close_MAX_prior_5_days                10         YES         FLOAT64   \n",
       "10      close_MAX_prior_20_days                11         YES         FLOAT64   \n",
       "11     close_MAX_prior_260_days                12         YES         FLOAT64   \n",
       "12       close_AVG_prior_5_days                13         YES         FLOAT64   \n",
       "13      close_AVG_prior_20_days                14         YES         FLOAT64   \n",
       "14     close_AVG_prior_260_days                15         YES         FLOAT64   \n",
       "15    close_STDDEV_prior_5_days                16         YES         FLOAT64   \n",
       "16   close_STDDEV_prior_20_days                17         YES         FLOAT64   \n",
       "17  close_STDDEV_prior_260_days                18         YES         FLOAT64   \n",
       "18       close_values_prior_260                19          NO  ARRAY<FLOAT64>   \n",
       "19               days_on_market                20         YES           INT64   \n",
       "20                scaled_change                21         YES         FLOAT64   \n",
       "21            s_p_scaled_change                22         YES         FLOAT64   \n",
       "22            normalized_change                23         YES         FLOAT64   \n",
       "23                      company                24         YES          STRING   \n",
       "24                     industry                25         YES          STRING   \n",
       "25                    direction                26         YES          STRING   \n",
       "\n",
       "   is_hidden is_system_defined is_partitioning_column  \\\n",
       "0         NO                NO                     NO   \n",
       "1         NO                NO                     NO   \n",
       "2         NO                NO                     NO   \n",
       "3         NO                NO                     NO   \n",
       "4         NO                NO                     NO   \n",
       "5         NO                NO                     NO   \n",
       "6         NO                NO                     NO   \n",
       "7         NO                NO                     NO   \n",
       "8         NO                NO                     NO   \n",
       "9         NO                NO                     NO   \n",
       "10        NO                NO                     NO   \n",
       "11        NO                NO                     NO   \n",
       "12        NO                NO                     NO   \n",
       "13        NO                NO                     NO   \n",
       "14        NO                NO                     NO   \n",
       "15        NO                NO                     NO   \n",
       "16        NO                NO                     NO   \n",
       "17        NO                NO                     NO   \n",
       "18        NO                NO                     NO   \n",
       "19        NO                NO                     NO   \n",
       "20        NO                NO                     NO   \n",
       "21        NO                NO                     NO   \n",
       "22        NO                NO                     NO   \n",
       "23        NO                NO                     NO   \n",
       "24        NO                NO                     NO   \n",
       "25        NO                NO                     NO   \n",
       "\n",
       "    clustering_ordinal_position  \n",
       "0                           NaN  \n",
       "1                           NaN  \n",
       "2                           NaN  \n",
       "3                           NaN  \n",
       "4                           NaN  \n",
       "5                           NaN  \n",
       "6                           NaN  \n",
       "7                           NaN  \n",
       "8                           NaN  \n",
       "9                           NaN  \n",
       "10                          NaN  \n",
       "11                          NaN  \n",
       "12                          NaN  \n",
       "13                          NaN  \n",
       "14                          NaN  \n",
       "15                          NaN  \n",
       "16                          NaN  \n",
       "17                          NaN  \n",
       "18                          NaN  \n",
       "19                          NaN  \n",
       "20                          NaN  \n",
       "21                          NaN  \n",
       "22                          NaN  \n",
       "23                          NaN  \n",
       "24                          NaN  \n",
       "25                          NaN  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%bigquery \n",
    "SELECT\n",
    "    * EXCEPT(is_generated, generation_expression, is_stored, is_updatable)\n",
    "FROM\n",
    "    `cloud-training-prod-bucket.ml4f`.INFORMATION_SCHEMA.COLUMNS\n",
    "WHERE\n",
    "    table_name = \"percent_change_sp500\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Most of the features, like `open` and `close` are pretty straightforward. The features generated using analytic functions, such as `close_MIN_prior_5_days` are best described using an example. Let's take the 6 most recent rows of data for IBM and reproduce the `close_MIN_prior_5_days` column. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>symbol</th>\n",
       "      <th>Date</th>\n",
       "      <th>Open</th>\n",
       "      <th>Close</th>\n",
       "      <th>tomorrow_close</th>\n",
       "      <th>tomo_close_m_close</th>\n",
       "      <th>close_MIN_prior_5_days</th>\n",
       "      <th>close_MIN_prior_20_days</th>\n",
       "      <th>close_MIN_prior_260_days</th>\n",
       "      <th>close_MAX_prior_5_days</th>\n",
       "      <th>...</th>\n",
       "      <th>close_STDDEV_prior_20_days</th>\n",
       "      <th>close_STDDEV_prior_260_days</th>\n",
       "      <th>close_values_prior_260</th>\n",
       "      <th>days_on_market</th>\n",
       "      <th>scaled_change</th>\n",
       "      <th>s_p_scaled_change</th>\n",
       "      <th>normalized_change</th>\n",
       "      <th>company</th>\n",
       "      <th>industry</th>\n",
       "      <th>direction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>IBM</td>\n",
       "      <td>2013-02-01</td>\n",
       "      <td>204.65</td>\n",
       "      <td>205.18</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.989716</td>\n",
       "      <td>0.937323</td>\n",
       "      <td>0.879813</td>\n",
       "      <td>0.998977</td>\n",
       "      <td>...</td>\n",
       "      <td>0.025808</td>\n",
       "      <td>0.031267</td>\n",
       "      <td>[180.52, 188.52, 189.98, 191.93, 191.73, 190.9...</td>\n",
       "      <td>12860</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.011539</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Intl Business Machines Corp</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>IBM</td>\n",
       "      <td>2013-01-31</td>\n",
       "      <td>203.32</td>\n",
       "      <td>203.07</td>\n",
       "      <td>205.18</td>\n",
       "      <td>2.11</td>\n",
       "      <td>1.002216</td>\n",
       "      <td>0.947063</td>\n",
       "      <td>0.888955</td>\n",
       "      <td>1.009356</td>\n",
       "      <td>...</td>\n",
       "      <td>0.025220</td>\n",
       "      <td>0.031925</td>\n",
       "      <td>[181.07, 180.52, 188.52, 189.98, 191.93, 191.7...</td>\n",
       "      <td>12859</td>\n",
       "      <td>0.010391</td>\n",
       "      <td>0.010053</td>\n",
       "      <td>0.000338</td>\n",
       "      <td>Intl Business Machines Corp</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>IBM</td>\n",
       "      <td>2013-01-30</td>\n",
       "      <td>203.69</td>\n",
       "      <td>203.52</td>\n",
       "      <td>203.07</td>\n",
       "      <td>-0.45</td>\n",
       "      <td>1.001867</td>\n",
       "      <td>0.941185</td>\n",
       "      <td>0.884434</td>\n",
       "      <td>1.007125</td>\n",
       "      <td>...</td>\n",
       "      <td>0.024643</td>\n",
       "      <td>0.032221</td>\n",
       "      <td>[180.0, 181.07, 180.52, 188.52, 189.98, 191.93...</td>\n",
       "      <td>12858</td>\n",
       "      <td>-0.002211</td>\n",
       "      <td>-0.002563</td>\n",
       "      <td>0.000352</td>\n",
       "      <td>Intl Business Machines Corp</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>IBM</td>\n",
       "      <td>2013-01-29</td>\n",
       "      <td>204.34</td>\n",
       "      <td>203.90</td>\n",
       "      <td>203.52</td>\n",
       "      <td>-0.38</td>\n",
       "      <td>0.961648</td>\n",
       "      <td>0.930996</td>\n",
       "      <td>0.878666</td>\n",
       "      <td>1.005248</td>\n",
       "      <td>...</td>\n",
       "      <td>0.023987</td>\n",
       "      <td>0.032552</td>\n",
       "      <td>[179.16, 180.0, 181.07, 180.52, 188.52, 189.98...</td>\n",
       "      <td>12857</td>\n",
       "      <td>-0.001864</td>\n",
       "      <td>-0.003900</td>\n",
       "      <td>0.002036</td>\n",
       "      <td>Intl Business Machines Corp</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>IBM</td>\n",
       "      <td>2013-01-28</td>\n",
       "      <td>204.85</td>\n",
       "      <td>204.93</td>\n",
       "      <td>203.90</td>\n",
       "      <td>-1.03</td>\n",
       "      <td>0.948958</td>\n",
       "      <td>0.926316</td>\n",
       "      <td>0.874250</td>\n",
       "      <td>1.000195</td>\n",
       "      <td>...</td>\n",
       "      <td>0.021542</td>\n",
       "      <td>0.032677</td>\n",
       "      <td>[180.55, 179.16, 180.0, 181.07, 180.52, 188.52...</td>\n",
       "      <td>12856</td>\n",
       "      <td>-0.005026</td>\n",
       "      <td>0.005106</td>\n",
       "      <td>-0.010132</td>\n",
       "      <td>Intl Business Machines Corp</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>DOWN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>IBM</td>\n",
       "      <td>2013-01-25</td>\n",
       "      <td>204.45</td>\n",
       "      <td>204.97</td>\n",
       "      <td>204.93</td>\n",
       "      <td>-0.04</td>\n",
       "      <td>0.944772</td>\n",
       "      <td>0.926136</td>\n",
       "      <td>0.874079</td>\n",
       "      <td>0.998780</td>\n",
       "      <td>...</td>\n",
       "      <td>0.018510</td>\n",
       "      <td>0.032874</td>\n",
       "      <td>[182.32, 180.55, 179.16, 180.0, 181.07, 180.52...</td>\n",
       "      <td>12855</td>\n",
       "      <td>-0.000195</td>\n",
       "      <td>-0.001850</td>\n",
       "      <td>0.001655</td>\n",
       "      <td>Intl Business Machines Corp</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6 rows × 26 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "  symbol        Date    Open   Close  tomorrow_close  tomo_close_m_close  \\\n",
       "0    IBM  2013-02-01  204.65  205.18             NaN                 NaN   \n",
       "1    IBM  2013-01-31  203.32  203.07          205.18                2.11   \n",
       "2    IBM  2013-01-30  203.69  203.52          203.07               -0.45   \n",
       "3    IBM  2013-01-29  204.34  203.90          203.52               -0.38   \n",
       "4    IBM  2013-01-28  204.85  204.93          203.90               -1.03   \n",
       "5    IBM  2013-01-25  204.45  204.97          204.93               -0.04   \n",
       "\n",
       "   close_MIN_prior_5_days  close_MIN_prior_20_days  close_MIN_prior_260_days  \\\n",
       "0                0.989716                 0.937323                  0.879813   \n",
       "1                1.002216                 0.947063                  0.888955   \n",
       "2                1.001867                 0.941185                  0.884434   \n",
       "3                0.961648                 0.930996                  0.878666   \n",
       "4                0.948958                 0.926316                  0.874250   \n",
       "5                0.944772                 0.926136                  0.874079   \n",
       "\n",
       "   close_MAX_prior_5_days  ...  close_STDDEV_prior_20_days  \\\n",
       "0                0.998977  ...                    0.025808   \n",
       "1                1.009356  ...                    0.025220   \n",
       "2                1.007125  ...                    0.024643   \n",
       "3                1.005248  ...                    0.023987   \n",
       "4                1.000195  ...                    0.021542   \n",
       "5                0.998780  ...                    0.018510   \n",
       "\n",
       "   close_STDDEV_prior_260_days  \\\n",
       "0                     0.031267   \n",
       "1                     0.031925   \n",
       "2                     0.032221   \n",
       "3                     0.032552   \n",
       "4                     0.032677   \n",
       "5                     0.032874   \n",
       "\n",
       "                              close_values_prior_260  days_on_market  \\\n",
       "0  [180.52, 188.52, 189.98, 191.93, 191.73, 190.9...           12860   \n",
       "1  [181.07, 180.52, 188.52, 189.98, 191.93, 191.7...           12859   \n",
       "2  [180.0, 181.07, 180.52, 188.52, 189.98, 191.93...           12858   \n",
       "3  [179.16, 180.0, 181.07, 180.52, 188.52, 189.98...           12857   \n",
       "4  [180.55, 179.16, 180.0, 181.07, 180.52, 188.52...           12856   \n",
       "5  [182.32, 180.55, 179.16, 180.0, 181.07, 180.52...           12855   \n",
       "\n",
       "   scaled_change  s_p_scaled_change  normalized_change  \\\n",
       "0            NaN          -0.011539                NaN   \n",
       "1       0.010391           0.010053           0.000338   \n",
       "2      -0.002211          -0.002563           0.000352   \n",
       "3      -0.001864          -0.003900           0.002036   \n",
       "4      -0.005026           0.005106          -0.010132   \n",
       "5      -0.000195          -0.001850           0.001655   \n",
       "\n",
       "                       company                industry  direction  \n",
       "0  Intl Business Machines Corp  Information Technology       STAY  \n",
       "1  Intl Business Machines Corp  Information Technology       STAY  \n",
       "2  Intl Business Machines Corp  Information Technology       STAY  \n",
       "3  Intl Business Machines Corp  Information Technology       STAY  \n",
       "4  Intl Business Machines Corp  Information Technology       DOWN  \n",
       "5  Intl Business Machines Corp  Information Technology       STAY  \n",
       "\n",
       "[6 rows x 26 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%bigquery\n",
    "SELECT \n",
    "    *\n",
    "FROM\n",
    "    `cloud-training-prod-bucket.ml4f.percent_change_sp500`\n",
    "WHERE\n",
    "    symbol = 'IBM'\n",
    "ORDER BY \n",
    "    Date DESC\n",
    "LIMIT 6"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For `Date = 2013-02-01` how did we arrive at `close_MIN_prior_5_days = 0.989716`? The minimum close over the past five days was `203.07`. This is normalized by the current day's close of `205.18` to get `close_MIN_prior_5_days = 203.07 / 205.18 = 0.989716`. The other features utilizing analytic functions were generated in a similar way. Here are explanations for some of the other features:\n",
    "\n",
    "* __scaled_change__: `tomo_close_m_close / close`\n",
    "* __s_p_scaled_change__: This value is calculated the same way as `scaled_change` but for the S&P 500 index. \n",
    "* __normalized_change__: `scaled_change - s_p_scaled_change` The normalization using the S&P index fund helps ensure that the future price of a stock is not due to larger market effects. Normalization helps us isolate the factors contributing to the performance of a stock_market.\n",
    "* __direction__: This is the target variable we're trying to predict. The logic for this variable is as follows: \n",
    "\n",
    "    ```sql\n",
    "    CASE \n",
    "        WHEN normalized_change < -0.01 THEN 'DOWN'\n",
    "        WHEN normalized_change > 0.01 THEN 'UP'\n",
    "        ELSE 'STAY'\n",
    "    END AS direction\n",
    "    ```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create classification model for `direction`\n",
    "\n",
    "In this example, your job is to create a classification model to predict the `direction` of each stock. Be creative! You can do this in any number of ways. For example, you can use BigQuery, Scikit-Learn, or AutoML. Feel free to add additional features, or use time series models.   \n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Establish a Simple Benchmark\n",
    "\n",
    "One way to assess the performance of a model is to compare it to a simple benchmark. We can do this by seeing what kind of accuracy we would get using the naive strategy of just predicting the majority class. Across the entire dataset, the majority class is 'STAY'. Using the following query we can see how this naive strategy would perform."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>Direction</th>\n",
       "      <th>percentage</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>STAY</td>\n",
       "      <td>53.766049</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>UP</td>\n",
       "      <td>23.240681</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>DOWN</td>\n",
       "      <td>22.993271</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Direction  percentage\n",
       "0      STAY   53.766049\n",
       "1        UP   23.240681\n",
       "2      DOWN   22.993271"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%bigquery\n",
    "WITH subset as (\n",
    "    SELECT \n",
    "        Direction\n",
    "    FROM\n",
    "        `cloud-training-prod-bucket.ml4f.percent_change_sp500`\n",
    "    WHERE\n",
    "        tomorrow_close IS NOT NULL\n",
    ")\n",
    "SELECT \n",
    "    Direction,\n",
    "    100.0 * COUNT(*) / (SELECT COUNT(*) FROM subset) as percentage\n",
    "FROM\n",
    "    subset\n",
    "GROUP BY\n",
    "    Direction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So, the naive strategy of just guessing the majority class would have accuracy of around 54% across the entire dataset. See if you can improve on this. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train Your Own Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "from sklearn import linear_model\n",
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.metrics import r2_score\n",
    "\n",
    "plt.rc('figure', figsize=(12, 8.0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "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>symbol</th>\n",
       "      <th>Date</th>\n",
       "      <th>Open</th>\n",
       "      <th>Close</th>\n",
       "      <th>tomorrow_close</th>\n",
       "      <th>tomo_close_m_close</th>\n",
       "      <th>close_MIN_prior_5_days</th>\n",
       "      <th>close_MIN_prior_20_days</th>\n",
       "      <th>close_MIN_prior_260_days</th>\n",
       "      <th>close_MAX_prior_5_days</th>\n",
       "      <th>...</th>\n",
       "      <th>close_STDDEV_prior_20_days</th>\n",
       "      <th>close_STDDEV_prior_260_days</th>\n",
       "      <th>close_values_prior_260</th>\n",
       "      <th>days_on_market</th>\n",
       "      <th>scaled_change</th>\n",
       "      <th>s_p_scaled_change</th>\n",
       "      <th>normalized_change</th>\n",
       "      <th>company</th>\n",
       "      <th>industry</th>\n",
       "      <th>direction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>CSC</td>\n",
       "      <td>1994-09-22</td>\n",
       "      <td>42.50</td>\n",
       "      <td>42.87</td>\n",
       "      <td>43.37</td>\n",
       "      <td>0.50</td>\n",
       "      <td>0.994168</td>\n",
       "      <td>0.982505</td>\n",
       "      <td>0.763937</td>\n",
       "      <td>1.029158</td>\n",
       "      <td>...</td>\n",
       "      <td>0.020679</td>\n",
       "      <td>0.605874</td>\n",
       "      <td>[90.37, 89.25, 90.87, 90.37, 90.12, 89.75, 88....</td>\n",
       "      <td>3219</td>\n",
       "      <td>0.011663</td>\n",
       "      <td>-0.003469</td>\n",
       "      <td>0.015132</td>\n",
       "      <td>Computer Sciences</td>\n",
       "      <td>Information Technology</td>\n",
       "      <td>UP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>A</td>\n",
       "      <td>2009-05-21</td>\n",
       "      <td>18.22</td>\n",
       "      <td>17.68</td>\n",
       "      <td>17.69</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.998303</td>\n",
       "      <td>0.927602</td>\n",
       "      <td>0.703620</td>\n",
       "      <td>1.054299</td>\n",
       "      <td>...</td>\n",
       "      <td>0.049348</td>\n",
       "      <td>0.489701</td>\n",
       "      <td>[31.9, 32.37, 32.23, 32.53, 35.34, 35.09, 36.2...</td>\n",
       "      <td>2390</td>\n",
       "      <td>0.000566</td>\n",
       "      <td>-0.001497</td>\n",
       "      <td>0.002063</td>\n",
       "      <td>Agilent Technologies Inc</td>\n",
       "      <td>Health Care</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ANF</td>\n",
       "      <td>1999-12-30</td>\n",
       "      <td>28.50</td>\n",
       "      <td>28.12</td>\n",
       "      <td>26.69</td>\n",
       "      <td>-1.43</td>\n",
       "      <td>0.871266</td>\n",
       "      <td>0.862376</td>\n",
       "      <td>0.775605</td>\n",
       "      <td>0.977952</td>\n",
       "      <td>...</td>\n",
       "      <td>0.084508</td>\n",
       "      <td>0.879128</td>\n",
       "      <td>[60.0, 60.75, 63.0, 64.94, 61.5, 63.56, 63.75,...</td>\n",
       "      <td>823</td>\n",
       "      <td>-0.050853</td>\n",
       "      <td>0.003264</td>\n",
       "      <td>-0.054117</td>\n",
       "      <td>Abercrombie &amp; Fitch Company A</td>\n",
       "      <td>Consumer Discretionary</td>\n",
       "      <td>DOWN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>AMZN</td>\n",
       "      <td>2004-12-14</td>\n",
       "      <td>40.15</td>\n",
       "      <td>40.44</td>\n",
       "      <td>41.00</td>\n",
       "      <td>0.56</td>\n",
       "      <td>0.944609</td>\n",
       "      <td>0.940900</td>\n",
       "      <td>0.836548</td>\n",
       "      <td>0.984916</td>\n",
       "      <td>...</td>\n",
       "      <td>0.019816</td>\n",
       "      <td>0.139052</td>\n",
       "      <td>[53.93, 51.51, 51.8, 51.56, 51.08, 49.34, 49.2...</td>\n",
       "      <td>1899</td>\n",
       "      <td>0.013848</td>\n",
       "      <td>0.001945</td>\n",
       "      <td>0.011903</td>\n",
       "      <td>Amazon.com Inc</td>\n",
       "      <td>Consumer Discretionary</td>\n",
       "      <td>UP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>DNB</td>\n",
       "      <td>1985-07-10</td>\n",
       "      <td>17.47</td>\n",
       "      <td>17.29</td>\n",
       "      <td>17.08</td>\n",
       "      <td>-0.21</td>\n",
       "      <td>0.976865</td>\n",
       "      <td>0.926547</td>\n",
       "      <td>0.690573</td>\n",
       "      <td>1.013881</td>\n",
       "      <td>...</td>\n",
       "      <td>0.023376</td>\n",
       "      <td>0.082239</td>\n",
       "      <td>[12.32, 12.14, 12.27, 12.24, 12.24, 12.09, 12....</td>\n",
       "      <td>527</td>\n",
       "      <td>-0.012146</td>\n",
       "      <td>0.002963</td>\n",
       "      <td>-0.015109</td>\n",
       "      <td>Dun &amp; Bradstreet Corp</td>\n",
       "      <td>Industrials</td>\n",
       "      <td>DOWN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 26 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "  symbol        Date   Open  Close  tomorrow_close  tomo_close_m_close  \\\n",
       "0    CSC  1994-09-22  42.50  42.87           43.37                0.50   \n",
       "1      A  2009-05-21  18.22  17.68           17.69                0.01   \n",
       "2    ANF  1999-12-30  28.50  28.12           26.69               -1.43   \n",
       "3   AMZN  2004-12-14  40.15  40.44           41.00                0.56   \n",
       "4    DNB  1985-07-10  17.47  17.29           17.08               -0.21   \n",
       "\n",
       "   close_MIN_prior_5_days  close_MIN_prior_20_days  close_MIN_prior_260_days  \\\n",
       "0                0.994168                 0.982505                  0.763937   \n",
       "1                0.998303                 0.927602                  0.703620   \n",
       "2                0.871266                 0.862376                  0.775605   \n",
       "3                0.944609                 0.940900                  0.836548   \n",
       "4                0.976865                 0.926547                  0.690573   \n",
       "\n",
       "   close_MAX_prior_5_days  ...  close_STDDEV_prior_20_days  \\\n",
       "0                1.029158  ...                    0.020679   \n",
       "1                1.054299  ...                    0.049348   \n",
       "2                0.977952  ...                    0.084508   \n",
       "3                0.984916  ...                    0.019816   \n",
       "4                1.013881  ...                    0.023376   \n",
       "\n",
       "   close_STDDEV_prior_260_days  \\\n",
       "0                     0.605874   \n",
       "1                     0.489701   \n",
       "2                     0.879128   \n",
       "3                     0.139052   \n",
       "4                     0.082239   \n",
       "\n",
       "                              close_values_prior_260  days_on_market  \\\n",
       "0  [90.37, 89.25, 90.87, 90.37, 90.12, 89.75, 88....            3219   \n",
       "1  [31.9, 32.37, 32.23, 32.53, 35.34, 35.09, 36.2...            2390   \n",
       "2  [60.0, 60.75, 63.0, 64.94, 61.5, 63.56, 63.75,...             823   \n",
       "3  [53.93, 51.51, 51.8, 51.56, 51.08, 49.34, 49.2...            1899   \n",
       "4  [12.32, 12.14, 12.27, 12.24, 12.24, 12.09, 12....             527   \n",
       "\n",
       "   scaled_change  s_p_scaled_change  normalized_change  \\\n",
       "0       0.011663          -0.003469           0.015132   \n",
       "1       0.000566          -0.001497           0.002063   \n",
       "2      -0.050853           0.003264          -0.054117   \n",
       "3       0.013848           0.001945           0.011903   \n",
       "4      -0.012146           0.002963          -0.015109   \n",
       "\n",
       "                         company                industry  direction  \n",
       "0              Computer Sciences  Information Technology         UP  \n",
       "1       Agilent Technologies Inc             Health Care       STAY  \n",
       "2  Abercrombie & Fitch Company A  Consumer Discretionary       DOWN  \n",
       "3                 Amazon.com Inc  Consumer Discretionary         UP  \n",
       "4          Dun & Bradstreet Corp             Industrials       DOWN  \n",
       "\n",
       "[5 rows x 26 columns]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['Date'] =  pd.to_datetime(df['Date'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TODO: Write code to build a model to predict Direction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 10000 entries, 0 to 9999\n",
      "Data columns (total 26 columns):\n",
      " #   Column                       Non-Null Count  Dtype         \n",
      "---  ------                       --------------  -----         \n",
      " 0   symbol                       10000 non-null  object        \n",
      " 1   Date                         10000 non-null  datetime64[ns]\n",
      " 2   Open                         10000 non-null  float64       \n",
      " 3   Close                        10000 non-null  float64       \n",
      " 4   tomorrow_close               9997 non-null   float64       \n",
      " 5   tomo_close_m_close           9997 non-null   float64       \n",
      " 6   close_MIN_prior_5_days       10000 non-null  float64       \n",
      " 7   close_MIN_prior_20_days      10000 non-null  float64       \n",
      " 8   close_MIN_prior_260_days     10000 non-null  float64       \n",
      " 9   close_MAX_prior_5_days       10000 non-null  float64       \n",
      " 10  close_MAX_prior_20_days      10000 non-null  float64       \n",
      " 11  close_MAX_prior_260_days     10000 non-null  float64       \n",
      " 12  close_AVG_prior_5_days       10000 non-null  float64       \n",
      " 13  close_AVG_prior_20_days      10000 non-null  float64       \n",
      " 14  close_AVG_prior_260_days     10000 non-null  float64       \n",
      " 15  close_STDDEV_prior_5_days    10000 non-null  float64       \n",
      " 16  close_STDDEV_prior_20_days   10000 non-null  float64       \n",
      " 17  close_STDDEV_prior_260_days  10000 non-null  float64       \n",
      " 18  close_values_prior_260       10000 non-null  object        \n",
      " 19  days_on_market               10000 non-null  int64         \n",
      " 20  scaled_change                9997 non-null   float64       \n",
      " 21  s_p_scaled_change            10000 non-null  float64       \n",
      " 22  normalized_change            9997 non-null   float64       \n",
      " 23  company                      10000 non-null  object        \n",
      " 24  industry                     10000 non-null  object        \n",
      " 25  direction                    10000 non-null  object        \n",
      "dtypes: datetime64[ns](1), float64(19), int64(1), object(5)\n",
      "memory usage: 2.0+ MB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n"
     ]
    },
    {
     "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>symbol</th>\n",
       "      <th>Date</th>\n",
       "      <th>Open</th>\n",
       "      <th>Close</th>\n",
       "      <th>tomorrow_close</th>\n",
       "      <th>tomo_close_m_close</th>\n",
       "      <th>close_MIN_prior_5_days</th>\n",
       "      <th>close_MIN_prior_20_days</th>\n",
       "      <th>close_MIN_prior_260_days</th>\n",
       "      <th>close_MAX_prior_5_days</th>\n",
       "      <th>...</th>\n",
       "      <th>close_STDDEV_prior_20_days</th>\n",
       "      <th>close_STDDEV_prior_260_days</th>\n",
       "      <th>close_values_prior_260</th>\n",
       "      <th>days_on_market</th>\n",
       "      <th>scaled_change</th>\n",
       "      <th>s_p_scaled_change</th>\n",
       "      <th>normalized_change</th>\n",
       "      <th>company</th>\n",
       "      <th>industry</th>\n",
       "      <th>direction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3031</th>\n",
       "      <td>DIS</td>\n",
       "      <td>1963-03-19</td>\n",
       "      <td>34.00</td>\n",
       "      <td>33.38</td>\n",
       "      <td>33.88</td>\n",
       "      <td>0.50</td>\n",
       "      <td>1.018574</td>\n",
       "      <td>0.981126</td>\n",
       "      <td>0.644098</td>\n",
       "      <td>1.048532</td>\n",
       "      <td>...</td>\n",
       "      <td>0.017209</td>\n",
       "      <td>0.114285</td>\n",
       "      <td>[37.5, 38.13, 38.0, 38.75, 39.38, 39.0, 39.25,...</td>\n",
       "      <td>306</td>\n",
       "      <td>0.014979</td>\n",
       "      <td>0.007332</td>\n",
       "      <td>0.007647</td>\n",
       "      <td>Walt Disney Co</td>\n",
       "      <td>Consumer Discretionary</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4411</th>\n",
       "      <td>DD</td>\n",
       "      <td>1963-04-01</td>\n",
       "      <td>241.38</td>\n",
       "      <td>241.38</td>\n",
       "      <td>241.00</td>\n",
       "      <td>-0.38</td>\n",
       "      <td>0.981854</td>\n",
       "      <td>0.980819</td>\n",
       "      <td>0.687712</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.005568</td>\n",
       "      <td>0.095126</td>\n",
       "      <td>[253.0, 249.0, 250.0, 248.13, 247.0, 244.0, 24...</td>\n",
       "      <td>315</td>\n",
       "      <td>-0.001574</td>\n",
       "      <td>-0.000150</td>\n",
       "      <td>-0.001425</td>\n",
       "      <td>DuPont, E.I. de Nemours</td>\n",
       "      <td>Materials</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4479</th>\n",
       "      <td>DD</td>\n",
       "      <td>1963-04-02</td>\n",
       "      <td>241.38</td>\n",
       "      <td>241.00</td>\n",
       "      <td>241.75</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.991701</td>\n",
       "      <td>0.982365</td>\n",
       "      <td>0.688797</td>\n",
       "      <td>1.001577</td>\n",
       "      <td>...</td>\n",
       "      <td>0.005937</td>\n",
       "      <td>0.095040</td>\n",
       "      <td>[249.0, 250.0, 248.13, 247.0, 244.0, 243.5, 24...</td>\n",
       "      <td>316</td>\n",
       "      <td>0.003112</td>\n",
       "      <td>0.007780</td>\n",
       "      <td>-0.004668</td>\n",
       "      <td>DuPont, E.I. de Nemours</td>\n",
       "      <td>Materials</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8183</th>\n",
       "      <td>CAT</td>\n",
       "      <td>1963-07-05</td>\n",
       "      <td>44.50</td>\n",
       "      <td>44.88</td>\n",
       "      <td>44.88</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.946970</td>\n",
       "      <td>0.946970</td>\n",
       "      <td>0.659982</td>\n",
       "      <td>0.991533</td>\n",
       "      <td>...</td>\n",
       "      <td>0.018280</td>\n",
       "      <td>0.082238</td>\n",
       "      <td>[30.87, 29.62, 30.12, 30.0, 31.62, 31.87, 32.0...</td>\n",
       "      <td>381</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.006836</td>\n",
       "      <td>0.006836</td>\n",
       "      <td>Caterpillar Inc</td>\n",
       "      <td>Industrials</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3764</th>\n",
       "      <td>AA</td>\n",
       "      <td>1963-08-02</td>\n",
       "      <td>62.25</td>\n",
       "      <td>62.00</td>\n",
       "      <td>61.88</td>\n",
       "      <td>-0.12</td>\n",
       "      <td>0.987903</td>\n",
       "      <td>0.973871</td>\n",
       "      <td>0.798387</td>\n",
       "      <td>1.026290</td>\n",
       "      <td>...</td>\n",
       "      <td>0.042494</td>\n",
       "      <td>0.073914</td>\n",
       "      <td>[54.25, 53.75, 53.25, 53.75, 56.0, 56.88, 58.2...</td>\n",
       "      <td>401</td>\n",
       "      <td>-0.001935</td>\n",
       "      <td>0.005916</td>\n",
       "      <td>-0.007852</td>\n",
       "      <td>Alcoa Inc</td>\n",
       "      <td>Materials</td>\n",
       "      <td>STAY</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 26 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     symbol       Date    Open   Close  tomorrow_close  tomo_close_m_close  \\\n",
       "3031    DIS 1963-03-19   34.00   33.38           33.88                0.50   \n",
       "4411     DD 1963-04-01  241.38  241.38          241.00               -0.38   \n",
       "4479     DD 1963-04-02  241.38  241.00          241.75                0.75   \n",
       "8183    CAT 1963-07-05   44.50   44.88           44.88                0.00   \n",
       "3764     AA 1963-08-02   62.25   62.00           61.88               -0.12   \n",
       "\n",
       "      close_MIN_prior_5_days  close_MIN_prior_20_days  \\\n",
       "3031                1.018574                 0.981126   \n",
       "4411                0.981854                 0.980819   \n",
       "4479                0.991701                 0.982365   \n",
       "8183                0.946970                 0.946970   \n",
       "3764                0.987903                 0.973871   \n",
       "\n",
       "      close_MIN_prior_260_days  close_MAX_prior_5_days  ...  \\\n",
       "3031                  0.644098                1.048532  ...   \n",
       "4411                  0.687712                1.000000  ...   \n",
       "4479                  0.688797                1.001577  ...   \n",
       "8183                  0.659982                0.991533  ...   \n",
       "3764                  0.798387                1.026290  ...   \n",
       "\n",
       "      close_STDDEV_prior_20_days  close_STDDEV_prior_260_days  \\\n",
       "3031                    0.017209                     0.114285   \n",
       "4411                    0.005568                     0.095126   \n",
       "4479                    0.005937                     0.095040   \n",
       "8183                    0.018280                     0.082238   \n",
       "3764                    0.042494                     0.073914   \n",
       "\n",
       "                                 close_values_prior_260  days_on_market  \\\n",
       "3031  [37.5, 38.13, 38.0, 38.75, 39.38, 39.0, 39.25,...             306   \n",
       "4411  [253.0, 249.0, 250.0, 248.13, 247.0, 244.0, 24...             315   \n",
       "4479  [249.0, 250.0, 248.13, 247.0, 244.0, 243.5, 24...             316   \n",
       "8183  [30.87, 29.62, 30.12, 30.0, 31.62, 31.87, 32.0...             381   \n",
       "3764  [54.25, 53.75, 53.25, 53.75, 56.0, 56.88, 58.2...             401   \n",
       "\n",
       "      scaled_change  s_p_scaled_change  normalized_change  \\\n",
       "3031       0.014979           0.007332           0.007647   \n",
       "4411      -0.001574          -0.000150          -0.001425   \n",
       "4479       0.003112           0.007780          -0.004668   \n",
       "8183       0.000000          -0.006836           0.006836   \n",
       "3764      -0.001935           0.005916          -0.007852   \n",
       "\n",
       "                      company                industry  direction  \n",
       "3031           Walt Disney Co  Consumer Discretionary       STAY  \n",
       "4411  DuPont, E.I. de Nemours               Materials       STAY  \n",
       "4479  DuPont, E.I. de Nemours               Materials       STAY  \n",
       "8183          Caterpillar Inc             Industrials       STAY  \n",
       "3764                Alcoa Inc               Materials       STAY  \n",
       "\n",
       "[5 rows x 26 columns]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(type(df))\n",
    "df.dropna(inplace=True)\n",
    "df.sort_values(by=['Date'], inplace=True, ascending=True)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['symbol', 'Date', 'Open', 'Close', 'tomorrow_close',\n",
       "       'tomo_close_m_close', 'close_MIN_prior_5_days',\n",
       "       'close_MIN_prior_20_days', 'close_MIN_prior_260_days',\n",
       "       'close_MAX_prior_5_days', 'close_MAX_prior_20_days',\n",
       "       'close_MAX_prior_260_days', 'close_AVG_prior_5_days',\n",
       "       'close_AVG_prior_20_days', 'close_AVG_prior_260_days',\n",
       "       'close_STDDEV_prior_5_days', 'close_STDDEV_prior_20_days',\n",
       "       'close_STDDEV_prior_260_days', 'close_values_prior_260',\n",
       "       'days_on_market', 'scaled_change', 's_p_scaled_change',\n",
       "       'normalized_change', 'company', 'industry', 'direction'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.plot(x='Date', y='Close');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9997, 26)"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "def encode(arg):\n",
    "    switcher = {\n",
    "        'STAY': 0,\n",
    "        'UP': 1,\n",
    "        'DOWN': -1\n",
    "    }\n",
    "    return switcher.get(arg, -5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "encode('UP')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['direction_encoded'] = df.apply(lambda x: encode(x['direction']), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSE: 0.40\n"
     ]
    }
   ],
   "source": [
    "features = [\n",
    "    'close_MIN_prior_5_days',\n",
    "    'close_MIN_prior_20_days',\n",
    "    'close_MIN_prior_260_days',\n",
    "    'close_MAX_prior_5_days',\n",
    "    'close_MAX_prior_20_days',\n",
    "    'close_MAX_prior_260_days',\n",
    "    'close_AVG_prior_5_days',\n",
    "    'close_AVG_prior_20_days',\n",
    "    'close_AVG_prior_260_days',\n",
    "#     'symbol',\n",
    "    'days_on_market',\n",
    "    'scaled_change',\n",
    "    's_p_scaled_change',\n",
    "    'normalized_change'\n",
    "]\n",
    "target = 'direction_encoded'\n",
    "\n",
    "X_train, X_test = df.loc[:8000, features], df.loc[8000:, features]\n",
    "Y_train, Y_test = df.loc[:8000, target], df.loc[8000:, target]\n",
    "# clf = OneVsRestClassifier(SVC()).fit(X_train, Y_train)\n",
    "regr = linear_model.LinearRegression(fit_intercept=False)\n",
    "regr.fit(X_train, Y_train)\n",
    "y_pred = regr.predict(X_test)\n",
    "# y_pred = clf.predict(X_test)\n",
    "print('RMSE: {0:.2f}'.format(np.sqrt(mean_squared_error(Y_test, y_pred))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variance Score: 0.57\n"
     ]
    }
   ],
   "source": [
    "print('Variance Score: {0:.2f}'.format(r2_score(Y_test, y_pred)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "No handles with labels found to put in legend.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(Y_test, y_pred)\n",
    "plt.xlabel('Actual')\n",
    "plt.ylabel('Predicted')\n",
    "plt.legend();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}