stock_validation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from ESRNN import ESRNN\n",
    "plt.style.use('ggplot')\n",
    "pd.options.display.max_rows = 999\n",
    "np.set_printoptions(threshold=np.inf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Plot\n",
    "def plot_prediction(y, y_hat):\n",
    "    n_y = len(y)\n",
    "    n_yhat = len(y_hat)\n",
    "    ds_y = np.array(range(n_y))\n",
    "    ds_yhat = np.array(range(n_y-n_yhat, n_y))\n",
    "\n",
    "    plt.plot(ds_y, y, label = 'y')\n",
    "    plt.plot(ds_yhat, y_hat, label='y_hat')\n",
    "    plt.legend(loc='upper left')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def ffill_missing_dates_particular_serie(serie, min_date, max_date, freq):\n",
    "    date_range = pd.date_range(start=min_date, end=max_date, freq=freq)\n",
    "    unique_id = serie['unique_id'].unique()\n",
    "    df_balanced = pd.DataFrame({'ds':date_range, 'key':[1]*len(date_range), 'unique_id': unique_id[0]})\n",
    "\n",
    "    # Check balance\n",
    "    check_balance = df_balanced.groupby(['unique_id']).size().reset_index(name='count')\n",
    "    assert len(set(check_balance['count'].values)) <= 1\n",
    "    df_balanced = df_balanced.merge(serie, how=\"left\", on=['unique_id', 'ds'])\n",
    "\n",
    "    df_balanced['y'] = df_balanced['y'].fillna(method='ffill')\n",
    "\n",
    "    return df_balanced\n",
    "\n",
    "def ffill_missing_dates_per_serie(df, freq, fixed_max_date=None):\n",
    "    \"\"\"Receives a DataFrame with a date column and forward fills the missing gaps in dates, not filling dates before\n",
    "    the first appearance of a unique key\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    df: DataFrame\n",
    "        Input DataFrame\n",
    "    key: str or list\n",
    "        Name(s) of the column(s) which make a unique time series\n",
    "    date_col: str\n",
    "        Name of the column that contains the time column\n",
    "    freq: str\n",
    "        Pandas time frequency standard strings, like \"W-THU\" or \"D\" or \"M\"\n",
    "    numeric_to_fill: str or list\n",
    "        Name(s) of the columns with numeric values to fill \"fill_value\" with\n",
    "    \"\"\"\n",
    "    if fixed_max_date is None:\n",
    "        df_max_min_dates = df[['unique_id', 'ds']].groupby('unique_id').agg(['min', 'max']).reset_index()\n",
    "    else:\n",
    "        df_max_min_dates = df[['unique_id', 'ds']].groupby('unique_id').agg(['min']).reset_index()\n",
    "        df_max_min_dates['max'] = fixed_max_date\n",
    "\n",
    "    df_max_min_dates.columns = df_max_min_dates.columns.droplevel()\n",
    "    df_max_min_dates.columns = ['unique_id', 'min_date', 'max_date']\n",
    "\n",
    "    df_list = []\n",
    "    for index, row in df_max_min_dates.iterrows():\n",
    "        df_id = df[df['unique_id'] == row['unique_id']]\n",
    "        df_id = ffill_missing_dates_particular_serie(df_id, row['min_date'], row['max_date'], freq)\n",
    "        df_list.append(df_id)\n",
    "\n",
    "    df_dates = pd.concat(df_list).reset_index(drop=True).drop('key', axis=1)[['unique_id', 'ds', 'y']]\n",
    "\n",
    "    return df_dates"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv('data/stock/train.csv')\n",
    "data['Date'] = pd.to_datetime(data['Date'])\n",
    "data['unique_id'] = data['Company']\n",
    "data = data.rename(columns={'Date':'ds', 'Close':'y'})\n",
    "\n",
    "#Series must be complete in the frequency\n",
    "data = ffill_missing_dates_per_serie(data,'D')\n",
    "data = data.drop_duplicates(['unique_id','ds'])\n",
    "\n",
    "X_train = data[['unique_id','ds']]\n",
    "X_train['x'] = '1'\n",
    "y_train = data[['unique_id','ds','y']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_test = pd.read_csv('data/stock/test.csv')\n",
    "data_test['ds'] = pd.to_datetime(data_test['Date'])\n",
    "data_test['unique_id'] = data_test['Company']\n",
    "X_test = data_test[['unique_id','ds','Close']]\n",
    "X_test.columns = ['unique_id', 'ds', 'y']\n",
    "uniques = X_test['unique_id'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "=============== Training ESRNN  ===============\n",
      "\n",
      "========= Epoch 0 finished =========\n",
      "Training time: 3.18633\n",
      "Training loss (75 prc): 0.04249\n",
      "========= Epoch 1 finished =========\n",
      "Training time: 2.99311\n",
      "Training loss (75 prc): 0.02695\n",
      "========= Epoch 2 finished =========\n",
      "Training time: 3.16133\n",
      "Training loss (75 prc): 0.02352\n",
      "========= Epoch 3 finished =========\n",
      "Training time: 3.22648\n",
      "Training loss (75 prc): 0.02316\n",
      "========= Epoch 4 finished =========\n",
      "Training time: 3.25065\n",
      "Training loss (75 prc): 0.02302\n",
      "Train finished! \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Train model\n",
    "esrnn = ESRNN(max_epochs=5, batch_size=4, learning_rate=3e-4, gradient_clipping_threshold=30,\n",
    "              dilations=[[1, 7]], add_nl_layer=True, per_series_lr_multip=0.8, \n",
    "              seasonality=[], input_size=7, output_size=34, max_periods=12, level_variability_penalty=10,\n",
    "              training_percentile=75, data_augmentation=True)\n",
    "esrnn.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "========= Predicting ESRNN =========\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/cchallu/anaconda3/envs/esrnn_torch/lib/python3.6/site-packages/pandas/core/frame.py:7138: FutureWarning: Sorting because non-concatenation axis is not aligned. A future version\n",
      "of pandas will change to not sort by default.\n",
      "\n",
      "To accept the future behavior, pass 'sort=False'.\n",
      "\n",
      "To retain the current behavior and silence the warning, pass 'sort=True'.\n",
      "\n",
      "  sort=sort,\n"
     ]
    }
   ],
   "source": [
    "y_hat = esrnn.predict(X_test)\n",
    "X_plot = y_train.append(y_hat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_id = 0\n",
    "y_test_plot = X_plot.loc[X_plot['unique_id']==uniques[plot_id]]\n",
    "plot_prediction(y_test_plot['y'], y_test_plot['y_hat'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0317271506388637"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.abs(y_hat['y_hat']-y_hat['y']).mean()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "esrnn_torch",
   "language": "python",
   "name": "esrnn_torch"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}