R&D.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 252,
   "id": "c2d31a50",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "       UTC timestamp          open          high           low         close  \\\n",
      "count   7.200000e+02    720.000000    720.000000    720.000000    720.000000   \n",
      "mean    1.632710e+12  47275.102917  47512.128750  47044.588611  47290.734583   \n",
      "std     7.487654e+08   4474.176712   4473.832542   4489.514015   4487.327318   \n",
      "min     1.631416e+12  40499.900000  41051.100000  39579.000000  40499.900000   \n",
      "25%     1.632063e+12  43393.375000  43687.750000  43149.150000  43392.100000   \n",
      "50%     1.632710e+12  47198.250000  47401.000000  47001.500000  47214.350000   \n",
      "75%     1.633357e+12  48773.600000  48977.300000  48565.050000  48838.600000   \n",
      "max     1.634004e+12  57501.600000  57837.700000  57358.200000  57499.900000   \n",
      "\n",
      "            volume  \n",
      "count   720.000000  \n",
      "mean    124.471403  \n",
      "std     179.623526  \n",
      "min       8.418664  \n",
      "25%      45.444360  \n",
      "50%      77.098601  \n",
      "75%     142.256810  \n",
      "max    2619.579706  \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>UTC timestamp</th>\n",
       "      <th>open</th>\n",
       "      <th>high</th>\n",
       "      <th>low</th>\n",
       "      <th>close</th>\n",
       "      <th>volume</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1631415600000</td>\n",
       "      <td>45299.0</td>\n",
       "      <td>45323.8</td>\n",
       "      <td>45226.3</td>\n",
       "      <td>45311.1</td>\n",
       "      <td>13.764573</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1631419200000</td>\n",
       "      <td>45311.1</td>\n",
       "      <td>45489.8</td>\n",
       "      <td>45239.3</td>\n",
       "      <td>45342.4</td>\n",
       "      <td>20.250714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1631422800000</td>\n",
       "      <td>45343.3</td>\n",
       "      <td>45490.0</td>\n",
       "      <td>45248.4</td>\n",
       "      <td>45395.8</td>\n",
       "      <td>10.472641</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1631426400000</td>\n",
       "      <td>45395.7</td>\n",
       "      <td>45415.0</td>\n",
       "      <td>45137.1</td>\n",
       "      <td>45190.3</td>\n",
       "      <td>11.760966</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1631430000000</td>\n",
       "      <td>45190.2</td>\n",
       "      <td>45368.7</td>\n",
       "      <td>45089.2</td>\n",
       "      <td>45363.4</td>\n",
       "      <td>22.654320</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   UTC timestamp     open     high      low    close     volume\n",
       "0  1631415600000  45299.0  45323.8  45226.3  45311.1  13.764573\n",
       "1  1631419200000  45311.1  45489.8  45239.3  45342.4  20.250714\n",
       "2  1631422800000  45343.3  45490.0  45248.4  45395.8  10.472641\n",
       "3  1631426400000  45395.7  45415.0  45137.1  45190.3  11.760966\n",
       "4  1631430000000  45190.2  45368.7  45089.2  45363.4  22.654320"
      ]
     },
     "execution_count": 252,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from get_data import get_ohlcv_series\n",
    "import pandas as pd\n",
    "from ta import add_all_ta_features\n",
    "from ta.utils import dropna\n",
    "\n",
    "from darts import TimeSeries\n",
    "from darts.utils.timeseries_generation import gaussian_timeseries, linear_timeseries, sine_timeseries\n",
    "from darts.models import RNNModel, TCNModel, TransformerModel, NBEATSModel, BlockRNNModel\n",
    "from darts.metrics import mape, smape, mase\n",
    "from darts.dataprocessing.transformers import Scaler, BoxCox\n",
    "from darts.utils.timeseries_generation import datetime_attribute_timeseries\n",
    "from darts.datasets import AirPassengersDataset, MonthlyMilkDataset\n",
    "from darts.models import KalmanFilter\n",
    "from darts.models.forecasting.gradient_boosted_model import LightGBMModel\n",
    "from darts.dataprocessing.transformers.missing_values_filler import MissingValuesFiller\n",
    "import json\n",
    "# if __name__ == '__main__':\n",
    "symbol = \"BTC-USD\"\n",
    "df = pd.read_csv(f\"{symbol}.csv\", index_col=0, parse_dates=True)\n",
    "print(df.describe())\n",
    "df.head()\n",
    "# todo kalman before features? Have to do df -> timeseries -> df -> apply feats -> timeseries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 253,
   "id": "ccac39be",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_ta(df):\n",
    "    # Add all ta features\n",
    "    df_ta = add_all_ta_features(df, open=\"open\", high=\"high\", low=\"low\", close=\"close\", volume=\"volume\")\n",
    "    # df = dropna(df)\n",
    "    df_ta = df_ta.drop('UTC timestamp','columns')\n",
    "\n",
    "    # Scale feats\n",
    "    ts_ta = TimeSeries.from_dataframe(df_ta,fill_missing_dates=True) #time_col=\"UTC timestamp\")\n",
    "    # scaler = Scaler()\n",
    "    # scaled_covars = scaler.fit_transform(ts_ta)\n",
    "    # scaled_covars.plot()\n",
    "    print(df_ta.columns)\n",
    "\n",
    "    return ts_ta\n",
    "\n",
    "# Scale covars\n",
    "def get_return_lags(df):\n",
    "    # Create the lagged percentage returns columns\n",
    "    lag_names = []\n",
    "    for i in [1,2,3,5,8,13,21,42]:\n",
    "        col_name = f\"lag_return_{i}\"\n",
    "        lag_names.append(col_name)\n",
    "        df[col_name] = df[\"close\"].pct_change(periods=i)\n",
    "        # df[col_name] = df[\"close\"].shift(periods=i)\n",
    "\n",
    "    \n",
    "    covars = TimeSeries.from_dataframe(df, value_cols=lag_names, fill_missing_dates=True)#,time_col=\"UTC timestamp\")\n",
    "    # scaler = Scaler()\n",
    "    # scaled_covars = scaler.fit_transform(covars)\n",
    "    # scaled_covars.plot()\n",
    "    return covars"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 254,
   "id": "6f80ba99",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['open', 'high', 'low', 'close', 'volume', 'lag_return_1',\n",
      "       'lag_return_2', 'lag_return_3', 'lag_return_5', 'lag_return_8',\n",
      "       'lag_return_13', 'lag_return_21', 'lag_return_42', 'volume_adi',\n",
      "       'volume_obv', 'volume_cmf', 'volume_fi', 'volume_mfi', 'volume_em',\n",
      "       'volume_sma_em', 'volume_vpt', 'volume_nvi', 'volume_vwap',\n",
      "       'volatility_atr', 'volatility_bbm', 'volatility_bbh', 'volatility_bbl',\n",
      "       'volatility_bbw', 'volatility_bbp', 'volatility_bbhi',\n",
      "       'volatility_bbli', 'volatility_kcc', 'volatility_kch', 'volatility_kcl',\n",
      "       'volatility_kcw', 'volatility_kcp', 'volatility_kchi',\n",
      "       'volatility_kcli', 'volatility_dcl', 'volatility_dch', 'volatility_dcm',\n",
      "       'volatility_dcw', 'volatility_dcp', 'volatility_ui', 'trend_macd',\n",
      "       'trend_macd_signal', 'trend_macd_diff', 'trend_sma_fast',\n",
      "       'trend_sma_slow', 'trend_ema_fast', 'trend_ema_slow', 'trend_adx',\n",
      "       'trend_adx_pos', 'trend_adx_neg', 'trend_vortex_ind_pos',\n",
      "       'trend_vortex_ind_neg', 'trend_vortex_ind_diff', 'trend_trix',\n",
      "       'trend_mass_index', 'trend_cci', 'trend_dpo', 'trend_kst',\n",
      "       'trend_kst_sig', 'trend_kst_diff', 'trend_ichimoku_conv',\n",
      "       'trend_ichimoku_base', 'trend_ichimoku_a', 'trend_ichimoku_b',\n",
      "       'trend_visual_ichimoku_a', 'trend_visual_ichimoku_b', 'trend_aroon_up',\n",
      "       'trend_aroon_down', 'trend_aroon_ind', 'trend_psar_up',\n",
      "       'trend_psar_down', 'trend_psar_up_indicator',\n",
      "       'trend_psar_down_indicator', 'trend_stc', 'momentum_rsi',\n",
      "       'momentum_stoch_rsi', 'momentum_stoch_rsi_k', 'momentum_stoch_rsi_d',\n",
      "       'momentum_tsi', 'momentum_uo', 'momentum_stoch',\n",
      "       'momentum_stoch_signal', 'momentum_wr', 'momentum_ao', 'momentum_kama',\n",
      "       'momentum_roc', 'momentum_ppo', 'momentum_ppo_signal',\n",
      "       'momentum_ppo_hist', 'others_dr', 'others_dlr', 'others_cr'],\n",
      "      dtype='object')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/michaelremington/opt/anaconda3/envs/stonk-pix-ai/lib/python3.8/site-packages/ta/trend.py:768: RuntimeWarning: invalid value encountered in double_scalars\n",
      "  dip[i] = 100 * (self._dip[i] / self._trs[i])\n",
      "/Users/michaelremington/opt/anaconda3/envs/stonk-pix-ai/lib/python3.8/site-packages/ta/trend.py:772: RuntimeWarning: invalid value encountered in double_scalars\n",
      "  din[i] = 100 * (self._din[i] / self._trs[i])\n"
     ]
    }
   ],
   "source": [
    "# Also adds features to df\n",
    "scaled_covars = get_return_lags(df)\n",
    "scaled_covars = get_ta(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 255,
   "id": "bd5aff55",
   "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>component</th>\n",
       "      <th>open</th>\n",
       "      <th>high</th>\n",
       "      <th>low</th>\n",
       "      <th>close</th>\n",
       "      <th>volume</th>\n",
       "      <th>lag_return_1</th>\n",
       "      <th>lag_return_2</th>\n",
       "      <th>lag_return_3</th>\n",
       "      <th>lag_return_5</th>\n",
       "      <th>lag_return_8</th>\n",
       "      <th>...</th>\n",
       "      <th>momentum_wr</th>\n",
       "      <th>momentum_ao</th>\n",
       "      <th>momentum_kama</th>\n",
       "      <th>momentum_roc</th>\n",
       "      <th>momentum_ppo</th>\n",
       "      <th>momentum_ppo_signal</th>\n",
       "      <th>momentum_ppo_hist</th>\n",
       "      <th>others_dr</th>\n",
       "      <th>others_dlr</th>\n",
       "      <th>others_cr</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>719.000000</td>\n",
       "      <td>718.000000</td>\n",
       "      <td>717.000000</td>\n",
       "      <td>715.000000</td>\n",
       "      <td>712.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>707.000000</td>\n",
       "      <td>687.000000</td>\n",
       "      <td>711.000000</td>\n",
       "      <td>708.000000</td>\n",
       "      <td>695.000000</td>\n",
       "      <td>687.000000</td>\n",
       "      <td>687.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>719.000000</td>\n",
       "      <td>720.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>47275.102917</td>\n",
       "      <td>47512.128750</td>\n",
       "      <td>47044.588611</td>\n",
       "      <td>47290.734583</td>\n",
       "      <td>124.471403</td>\n",
       "      <td>0.000343</td>\n",
       "      <td>0.000690</td>\n",
       "      <td>0.001031</td>\n",
       "      <td>0.001747</td>\n",
       "      <td>0.002758</td>\n",
       "      <td>...</td>\n",
       "      <td>-42.866815</td>\n",
       "      <td>243.801273</td>\n",
       "      <td>47244.729631</td>\n",
       "      <td>0.412819</td>\n",
       "      <td>-2.622604</td>\n",
       "      <td>-2.782655</td>\n",
       "      <td>-0.119116</td>\n",
       "      <td>0.028478</td>\n",
       "      <td>0.031430</td>\n",
       "      <td>4.368984</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>4474.176712</td>\n",
       "      <td>4473.832542</td>\n",
       "      <td>4489.514015</td>\n",
       "      <td>4487.327318</td>\n",
       "      <td>179.623526</td>\n",
       "      <td>0.007627</td>\n",
       "      <td>0.010460</td>\n",
       "      <td>0.012550</td>\n",
       "      <td>0.016163</td>\n",
       "      <td>0.020250</td>\n",
       "      <td>...</td>\n",
       "      <td>27.766888</td>\n",
       "      <td>971.843779</td>\n",
       "      <td>4401.491256</td>\n",
       "      <td>2.462846</td>\n",
       "      <td>19.091320</td>\n",
       "      <td>15.267664</td>\n",
       "      <td>10.239921</td>\n",
       "      <td>0.778195</td>\n",
       "      <td>0.762658</td>\n",
       "      <td>9.903373</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>40499.900000</td>\n",
       "      <td>41051.100000</td>\n",
       "      <td>39579.000000</td>\n",
       "      <td>40499.900000</td>\n",
       "      <td>8.418664</td>\n",
       "      <td>-0.047894</td>\n",
       "      <td>-0.055061</td>\n",
       "      <td>-0.065385</td>\n",
       "      <td>-0.083475</td>\n",
       "      <td>-0.071475</td>\n",
       "      <td>...</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-2752.405294</td>\n",
       "      <td>41488.792562</td>\n",
       "      <td>-7.953141</td>\n",
       "      <td>-48.910730</td>\n",
       "      <td>-37.351297</td>\n",
       "      <td>-18.449559</td>\n",
       "      <td>-4.789444</td>\n",
       "      <td>-4.907937</td>\n",
       "      <td>-10.618149</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>43393.375000</td>\n",
       "      <td>43687.750000</td>\n",
       "      <td>43149.150000</td>\n",
       "      <td>43392.100000</td>\n",
       "      <td>45.444360</td>\n",
       "      <td>-0.003226</td>\n",
       "      <td>-0.004397</td>\n",
       "      <td>-0.005353</td>\n",
       "      <td>-0.006672</td>\n",
       "      <td>-0.008354</td>\n",
       "      <td>...</td>\n",
       "      <td>-67.847337</td>\n",
       "      <td>-300.015441</td>\n",
       "      <td>43297.892759</td>\n",
       "      <td>-0.925998</td>\n",
       "      <td>-15.762489</td>\n",
       "      <td>-13.432125</td>\n",
       "      <td>-7.105500</td>\n",
       "      <td>-0.324201</td>\n",
       "      <td>-0.323094</td>\n",
       "      <td>-4.235165</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>47198.250000</td>\n",
       "      <td>47401.000000</td>\n",
       "      <td>47001.500000</td>\n",
       "      <td>47214.350000</td>\n",
       "      <td>77.098601</td>\n",
       "      <td>0.000141</td>\n",
       "      <td>0.000243</td>\n",
       "      <td>0.000160</td>\n",
       "      <td>0.000673</td>\n",
       "      <td>0.001119</td>\n",
       "      <td>...</td>\n",
       "      <td>-37.148615</td>\n",
       "      <td>216.794706</td>\n",
       "      <td>47341.903554</td>\n",
       "      <td>0.385368</td>\n",
       "      <td>-3.023265</td>\n",
       "      <td>-2.204748</td>\n",
       "      <td>-1.545698</td>\n",
       "      <td>0.014110</td>\n",
       "      <td>0.014111</td>\n",
       "      <td>4.200406</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>48773.600000</td>\n",
       "      <td>48977.300000</td>\n",
       "      <td>48565.050000</td>\n",
       "      <td>48838.600000</td>\n",
       "      <td>142.256810</td>\n",
       "      <td>0.003761</td>\n",
       "      <td>0.005094</td>\n",
       "      <td>0.006037</td>\n",
       "      <td>0.008948</td>\n",
       "      <td>0.013271</td>\n",
       "      <td>...</td>\n",
       "      <td>-18.903990</td>\n",
       "      <td>837.621765</td>\n",
       "      <td>48433.629544</td>\n",
       "      <td>1.773342</td>\n",
       "      <td>8.954388</td>\n",
       "      <td>6.565045</td>\n",
       "      <td>4.861717</td>\n",
       "      <td>0.375166</td>\n",
       "      <td>0.375347</td>\n",
       "      <td>7.785068</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>57501.600000</td>\n",
       "      <td>57837.700000</td>\n",
       "      <td>57358.200000</td>\n",
       "      <td>57499.900000</td>\n",
       "      <td>2619.579706</td>\n",
       "      <td>0.048991</td>\n",
       "      <td>0.057902</td>\n",
       "      <td>0.061269</td>\n",
       "      <td>0.079445</td>\n",
       "      <td>0.088551</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>3103.303824</td>\n",
       "      <td>56977.924738</td>\n",
       "      <td>9.117276</td>\n",
       "      <td>64.434878</td>\n",
       "      <td>37.840395</td>\n",
       "      <td>44.782317</td>\n",
       "      <td>4.899109</td>\n",
       "      <td>4.782884</td>\n",
       "      <td>26.900252</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8 rows × 96 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "component          open          high           low         close  \\\n",
       "count        720.000000    720.000000    720.000000    720.000000   \n",
       "mean       47275.102917  47512.128750  47044.588611  47290.734583   \n",
       "std         4474.176712   4473.832542   4489.514015   4487.327318   \n",
       "min        40499.900000  41051.100000  39579.000000  40499.900000   \n",
       "25%        43393.375000  43687.750000  43149.150000  43392.100000   \n",
       "50%        47198.250000  47401.000000  47001.500000  47214.350000   \n",
       "75%        48773.600000  48977.300000  48565.050000  48838.600000   \n",
       "max        57501.600000  57837.700000  57358.200000  57499.900000   \n",
       "\n",
       "component       volume  lag_return_1  lag_return_2  lag_return_3  \\\n",
       "count       720.000000    719.000000    718.000000    717.000000   \n",
       "mean        124.471403      0.000343      0.000690      0.001031   \n",
       "std         179.623526      0.007627      0.010460      0.012550   \n",
       "min           8.418664     -0.047894     -0.055061     -0.065385   \n",
       "25%          45.444360     -0.003226     -0.004397     -0.005353   \n",
       "50%          77.098601      0.000141      0.000243      0.000160   \n",
       "75%         142.256810      0.003761      0.005094      0.006037   \n",
       "max        2619.579706      0.048991      0.057902      0.061269   \n",
       "\n",
       "component  lag_return_5  lag_return_8  ...  momentum_wr  momentum_ao  \\\n",
       "count        715.000000    712.000000  ...   707.000000   687.000000   \n",
       "mean           0.001747      0.002758  ...   -42.866815   243.801273   \n",
       "std            0.016163      0.020250  ...    27.766888   971.843779   \n",
       "min           -0.083475     -0.071475  ...  -100.000000 -2752.405294   \n",
       "25%           -0.006672     -0.008354  ...   -67.847337  -300.015441   \n",
       "50%            0.000673      0.001119  ...   -37.148615   216.794706   \n",
       "75%            0.008948      0.013271  ...   -18.903990   837.621765   \n",
       "max            0.079445      0.088551  ...    -0.000000  3103.303824   \n",
       "\n",
       "component  momentum_kama  momentum_roc  momentum_ppo  momentum_ppo_signal  \\\n",
       "count         711.000000    708.000000    695.000000           687.000000   \n",
       "mean        47244.729631      0.412819     -2.622604            -2.782655   \n",
       "std          4401.491256      2.462846     19.091320            15.267664   \n",
       "min         41488.792562     -7.953141    -48.910730           -37.351297   \n",
       "25%         43297.892759     -0.925998    -15.762489           -13.432125   \n",
       "50%         47341.903554      0.385368     -3.023265            -2.204748   \n",
       "75%         48433.629544      1.773342      8.954388             6.565045   \n",
       "max         56977.924738      9.117276     64.434878            37.840395   \n",
       "\n",
       "component  momentum_ppo_hist   others_dr  others_dlr   others_cr  \n",
       "count             687.000000  720.000000  719.000000  720.000000  \n",
       "mean               -0.119116    0.028478    0.031430    4.368984  \n",
       "std                10.239921    0.778195    0.762658    9.903373  \n",
       "min               -18.449559   -4.789444   -4.907937  -10.618149  \n",
       "25%                -7.105500   -0.324201   -0.323094   -4.235165  \n",
       "50%                -1.545698    0.014110    0.014111    4.200406  \n",
       "75%                 4.861717    0.375166    0.375347    7.785068  \n",
       "max                44.782317    4.899109    4.782884   26.900252  \n",
       "\n",
       "[8 rows x 96 columns]"
      ]
     },
     "execution_count": 255,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scaled_covars.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "id": "ba0817e4",
   "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>component</th>\n",
       "      <th>open</th>\n",
       "      <th>high</th>\n",
       "      <th>low</th>\n",
       "      <th>close</th>\n",
       "      <th>volume</th>\n",
       "      <th>lag_return_1</th>\n",
       "      <th>lag_return_2</th>\n",
       "      <th>lag_return_3</th>\n",
       "      <th>lag_return_5</th>\n",
       "      <th>lag_return_8</th>\n",
       "      <th>...</th>\n",
       "      <th>momentum_wr</th>\n",
       "      <th>momentum_ao</th>\n",
       "      <th>momentum_kama</th>\n",
       "      <th>momentum_roc</th>\n",
       "      <th>momentum_ppo</th>\n",
       "      <th>momentum_ppo_signal</th>\n",
       "      <th>momentum_ppo_hist</th>\n",
       "      <th>others_dr</th>\n",
       "      <th>others_dlr</th>\n",
       "      <th>others_cr</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "      <td>720.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.398501</td>\n",
       "      <td>0.384892</td>\n",
       "      <td>0.419906</td>\n",
       "      <td>0.399461</td>\n",
       "      <td>0.044445</td>\n",
       "      <td>0.497890</td>\n",
       "      <td>0.493560</td>\n",
       "      <td>0.524266</td>\n",
       "      <td>0.523764</td>\n",
       "      <td>0.464850</td>\n",
       "      <td>...</td>\n",
       "      <td>0.573124</td>\n",
       "      <td>0.503421</td>\n",
       "      <td>0.370616</td>\n",
       "      <td>0.491019</td>\n",
       "      <td>0.418958</td>\n",
       "      <td>0.472825</td>\n",
       "      <td>0.285970</td>\n",
       "      <td>0.497280</td>\n",
       "      <td>0.509701</td>\n",
       "      <td>0.399461</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.263161</td>\n",
       "      <td>0.266512</td>\n",
       "      <td>0.252515</td>\n",
       "      <td>0.263960</td>\n",
       "      <td>0.068791</td>\n",
       "      <td>0.078663</td>\n",
       "      <td>0.092467</td>\n",
       "      <td>0.098901</td>\n",
       "      <td>0.099197</td>\n",
       "      <td>0.126173</td>\n",
       "      <td>...</td>\n",
       "      <td>0.275465</td>\n",
       "      <td>0.166432</td>\n",
       "      <td>0.282521</td>\n",
       "      <td>0.143247</td>\n",
       "      <td>0.174638</td>\n",
       "      <td>0.207139</td>\n",
       "      <td>0.159193</td>\n",
       "      <td>0.080321</td>\n",
       "      <td>0.078644</td>\n",
       "      <td>0.263960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.170187</td>\n",
       "      <td>0.157069</td>\n",
       "      <td>0.200805</td>\n",
       "      <td>0.170129</td>\n",
       "      <td>0.014180</td>\n",
       "      <td>0.461108</td>\n",
       "      <td>0.448599</td>\n",
       "      <td>0.474619</td>\n",
       "      <td>0.471692</td>\n",
       "      <td>0.396294</td>\n",
       "      <td>...</td>\n",
       "      <td>0.324558</td>\n",
       "      <td>0.402698</td>\n",
       "      <td>0.117980</td>\n",
       "      <td>0.414751</td>\n",
       "      <td>0.293829</td>\n",
       "      <td>0.329548</td>\n",
       "      <td>0.183445</td>\n",
       "      <td>0.460878</td>\n",
       "      <td>0.473173</td>\n",
       "      <td>0.170129</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.393981</td>\n",
       "      <td>0.378272</td>\n",
       "      <td>0.417482</td>\n",
       "      <td>0.394968</td>\n",
       "      <td>0.026302</td>\n",
       "      <td>0.495924</td>\n",
       "      <td>0.489954</td>\n",
       "      <td>0.517317</td>\n",
       "      <td>0.517104</td>\n",
       "      <td>0.454370</td>\n",
       "      <td>...</td>\n",
       "      <td>0.638251</td>\n",
       "      <td>0.499443</td>\n",
       "      <td>0.359369</td>\n",
       "      <td>0.491876</td>\n",
       "      <td>0.410870</td>\n",
       "      <td>0.476105</td>\n",
       "      <td>0.255458</td>\n",
       "      <td>0.495797</td>\n",
       "      <td>0.508035</td>\n",
       "      <td>0.394968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.486640</td>\n",
       "      <td>0.472174</td>\n",
       "      <td>0.505425</td>\n",
       "      <td>0.490512</td>\n",
       "      <td>0.051256</td>\n",
       "      <td>0.533063</td>\n",
       "      <td>0.532402</td>\n",
       "      <td>0.563890</td>\n",
       "      <td>0.568813</td>\n",
       "      <td>0.531657</td>\n",
       "      <td>...</td>\n",
       "      <td>0.809265</td>\n",
       "      <td>0.608281</td>\n",
       "      <td>0.448064</td>\n",
       "      <td>0.567985</td>\n",
       "      <td>0.523935</td>\n",
       "      <td>0.604519</td>\n",
       "      <td>0.363859</td>\n",
       "      <td>0.533063</td>\n",
       "      <td>0.545093</td>\n",
       "      <td>0.490512</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8 rows × 96 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "component        open        high         low       close      volume  \\\n",
       "count      720.000000  720.000000  720.000000  720.000000  720.000000   \n",
       "mean         0.398501    0.384892    0.419906    0.399461    0.044445   \n",
       "std          0.263161    0.266512    0.252515    0.263960    0.068791   \n",
       "min          0.000000    0.000000    0.000000    0.000000    0.000000   \n",
       "25%          0.170187    0.157069    0.200805    0.170129    0.014180   \n",
       "50%          0.393981    0.378272    0.417482    0.394968    0.026302   \n",
       "75%          0.486640    0.472174    0.505425    0.490512    0.051256   \n",
       "max          1.000000    1.000000    1.000000    1.000000    1.000000   \n",
       "\n",
       "component  lag_return_1  lag_return_2  lag_return_3  lag_return_5  \\\n",
       "count        720.000000    720.000000    720.000000    720.000000   \n",
       "mean           0.497890      0.493560      0.524266      0.523764   \n",
       "std            0.078663      0.092467      0.098901      0.099197   \n",
       "min            0.000000      0.000000      0.000000      0.000000   \n",
       "25%            0.461108      0.448599      0.474619      0.471692   \n",
       "50%            0.495924      0.489954      0.517317      0.517104   \n",
       "75%            0.533063      0.532402      0.563890      0.568813   \n",
       "max            1.000000      1.000000      1.000000      1.000000   \n",
       "\n",
       "component  lag_return_8  ...  momentum_wr  momentum_ao  momentum_kama  \\\n",
       "count        720.000000  ...   720.000000   720.000000     720.000000   \n",
       "mean           0.464850  ...     0.573124     0.503421       0.370616   \n",
       "std            0.126173  ...     0.275465     0.166432       0.282521   \n",
       "min            0.000000  ...     0.000000     0.000000       0.000000   \n",
       "25%            0.396294  ...     0.324558     0.402698       0.117980   \n",
       "50%            0.454370  ...     0.638251     0.499443       0.359369   \n",
       "75%            0.531657  ...     0.809265     0.608281       0.448064   \n",
       "max            1.000000  ...     1.000000     1.000000       1.000000   \n",
       "\n",
       "component  momentum_roc  momentum_ppo  momentum_ppo_signal  momentum_ppo_hist  \\\n",
       "count        720.000000    720.000000           720.000000         720.000000   \n",
       "mean           0.491019      0.418958             0.472825           0.285970   \n",
       "std            0.143247      0.174638             0.207139           0.159193   \n",
       "min            0.000000      0.000000             0.000000           0.000000   \n",
       "25%            0.414751      0.293829             0.329548           0.183445   \n",
       "50%            0.491876      0.410870             0.476105           0.255458   \n",
       "75%            0.567985      0.523935             0.604519           0.363859   \n",
       "max            1.000000      1.000000             1.000000           1.000000   \n",
       "\n",
       "component   others_dr  others_dlr   others_cr  \n",
       "count      720.000000  720.000000  720.000000  \n",
       "mean         0.497280    0.509701    0.399461  \n",
       "std          0.080321    0.078644    0.263960  \n",
       "min          0.000000    0.000000    0.000000  \n",
       "25%          0.460878    0.473173    0.170129  \n",
       "50%          0.495797    0.508035    0.394968  \n",
       "75%          0.533063    0.545093    0.490512  \n",
       "max          1.000000    1.000000    1.000000  \n",
       "\n",
       "[8 rows x 96 columns]"
      ]
     },
     "execution_count": 256,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Scale target\n",
    "# ts = TimeSeries.from_dataframe(df, time_col=\"UTC timestamp\", \n",
    "#                                # value_cols=[\"close\",\"open\",\"high\",\"low\",\"volume\",\"trend_ema_fast\",\"trend_ema_slow\",\"trend_macd_diff\",\"momentum_rsi\",\"volume_vwap\",\"momentum_kama\"],\n",
    "#                                # value_cols=['close'],\n",
    "#                                fill_missing_dates=True)\n",
    "ts = scaled_covars\n",
    "mvf = MissingValuesFiller()\n",
    "ts = mvf.transform(ts)\n",
    "scaler = Scaler()\n",
    "scaled = scaler.fit_transform(ts)\n",
    "scaled.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 257,
   "id": "f5000a80",
   "metadata": {},
   "outputs": [],
   "source": [
    "target_var = 'close'\n",
    "target_var_idx = scaled.columns.get_loc(target_var)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 258,
   "id": "fcd5ba1f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "split = int(len(scaled) * .1) # todo split param.\n",
    "train, val = scaled[:-split], scaled[-split:]\n",
    "train_covar, val_covar = scaled_covars[:-split], scaled_covars[-split:]\n",
    "train[target_var].plot(label='training')\n",
    "val[target_var].plot(label='validation')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 259,
   "id": "49a313fd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Kalman Smoothing\n",
    "filtered = KalmanFilter(P=1000., R=50, Q=1).filter(train[target_var])\n",
    "train[target_var].plot(label='actual')\n",
    "filtered.univariate_component(0).plot(label=\"Filtered\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "id": "66cc88d4",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[2021-10-11 20:23:21,775] INFO | darts.models.forecasting.torch_forecasting_model | Train dataset contains 577 samples.\n",
      "[2021-10-11 20:23:21,775] INFO | darts.models.forecasting.torch_forecasting_model | Train dataset contains 577 samples.\n",
      "[2021-10-11 20:23:21,777] INFO | darts.models.forecasting.tcn_model | Number of layers chosen: 4\n",
      "[2021-10-11 20:23:21,777] INFO | darts.models.forecasting.tcn_model | Number of layers chosen: 4\n",
      "[2021-10-11 20:23:21,780] INFO | darts.models.forecasting.torch_forecasting_model | Time series values are 64-bits; casting model to float64. If training is too slow you can try casting your data to 32-bits.\n",
      "[2021-10-11 20:23:21,780] INFO | darts.models.forecasting.torch_forecasting_model | Time series values are 64-bits; casting model to float64. If training is too slow you can try casting your data to 32-bits.\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e532c9507a044fcabf8d2202d9227749",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/30 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training loss: 0.0471, validation loss: 0.0688, best val loss: 0.0688\r"
     ]
    }
   ],
   "source": [
    "# Train\n",
    "model_cov = TCNModel(input_chunk_length=48, output_chunk_length=24, n_epochs=30, random_state=0,optimizer_kwargs={'lr':.001})\n",
    "# model_cov = TransformerModel(input_chunk_length=48, output_chunk_length=24, n_epochs=21, random_state=0,optimizer_kwargs={'lr':.001})\n",
    "# model_cov = RNNModel(model=\"GRU\",input_chunk_length=48, output_chunk_length=24, n_epochs=21, random_state=0,optimizer_kwargs={'lr':.001})\n",
    "# model_cov = LightGBMModel(lags=[-1,-2,-21,-34,-68,-3,-5,-8,-13])\n",
    "\n",
    "\n",
    "model_cov.fit(series=train, \n",
    "              val_series=val,\n",
    "              # past_covariates=train_covar, val_past_covariates=val_covar,\n",
    "              verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "id": "1a0359a1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAPE = 16.24%\n",
      "sMAPE = 18.40%\n",
      "MASE = 10.70%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pred_cov = model_cov.predict(n=len(val), \n",
    "                             series=train,\n",
    "                             # past_covariates=scaled_covars\n",
    "                             )\n",
    "scaled[target_var].plot(label='actual')\n",
    "pred_cov[target_var].plot(label='forecast',low_quantile=0.05, high_quantile=0.95)\n",
    "# plt.legend();\n",
    "# todo MAPE\n",
    "print('MAPE = {:.2f}%'.format(mape(scaled[target_var], pred_cov[target_var])))\n",
    "print('sMAPE = {:.2f}%'.format(smape(scaled[target_var], pred_cov[target_var])))\n",
    "print('MASE = {:.2f}%'.format(mase(scaled[target_var], pred_cov[target_var], insample=train[target_var])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "id": "073a7e65",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pred_cov = model_cov.predict(n=len(val), series=train,\n",
    "                             # past_covariates=scaled_covars\n",
    "                             )\n",
    "\n",
    "val[target_var].plot(label='actual')\n",
    "plt = pred_cov[target_var].plot(label='forecast')\n",
    "# plt.legend();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 263,
   "id": "cb356c0d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b7fbee0d6189468c8086ef54826de82a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/61 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAPE (using covariates) = 6.05%\n",
      "sMAPE = 6.32%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# historical_forecasts\n",
    "backtest_cov = model_cov.historical_forecasts(scaled,\n",
    "                                              # past_covariates=scaled_covars,\n",
    "                                              start=len(train),\n",
    "                                              forecast_horizon=12,\n",
    "                                              stride=1,\n",
    "                                              retrain=False,\n",
    "                                              verbose=True)\n",
    "\n",
    "scaled[target_var].plot(label='actual')\n",
    "# backtest_cov.plot(label='forecast')\n",
    "backtest_cov.univariate_component(target_var_idx).plot(label='forecast')\n",
    "# plt.legend();\n",
    "backtest_mape = mape(scaled[target_var], backtest_cov.univariate_component(target_var_idx))\n",
    "print('MAPE (using covariates) = {:.2f}%'.format(backtest_mape))\n",
    "print('sMAPE = {:.2f}%'.format(smape(scaled[target_var], backtest_cov.univariate_component(target_var_idx))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "id": "7be6199a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a945f505c90f46248b00dbd65f478d38",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/61 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.19370463646451114\n"
     ]
    }
   ],
   "source": [
    "from darts.metrics import coefficient_of_variation, mae, mse, r2_score, rmse\n",
    "# Backtest\n",
    "backtest_error = model_cov.backtest(scaled,\n",
    "                                              metric=mae,\n",
    "                                              start=len(train),\n",
    "                                              forecast_horizon=12,\n",
    "                                              stride=1,\n",
    "                                              retrain=False,\n",
    "                                              verbose=True)\n",
    "\n",
    "print(backtest_error)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 265,
   "id": "302a5a51",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[2021-10-11 20:23:59,377] INFO | darts.models.forecasting.torch_forecasting_model | Train dataset contains 649 samples.\n",
      "[2021-10-11 20:23:59,377] INFO | darts.models.forecasting.torch_forecasting_model | Train dataset contains 649 samples.\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "07854944be8e4d54a6edeb3c7fe516cf",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/30 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training loss: 0.0345\r"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Retrain on full series\n",
    "model_cov.fit(series=scaled, \n",
    "              # past_covariates=train_covar, val_past_covariates=val_covar,\n",
    "              verbose=True)\n",
    "\n",
    "\n",
    "# Predict future\n",
    "pred_cov = model_cov.predict(n=len(scaled)//10,\n",
    "                             # series=scaled,\n",
    "                             \n",
    "                             )\n",
    "scaled[target_var].plot(label='Past')\n",
    "pred_cov[target_var].plot(label='Forecast')\n",
    "# plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 266,
   "id": "5ae2f1c6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# JSON out\n",
    "scaler.inverse_transform(scaled)[target_var].plot()\n",
    "scaler.inverse_transform(pred_cov)[target_var].plot()\n",
    "\n",
    "past_close = scaler.inverse_transform(scaled)[target_var].pd_dataframe()\n",
    "prediction = scaler.inverse_transform(pred_cov)[target_var].pd_dataframe()\n",
    "period = \"1h\"\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 268,
   "id": "e9aef89b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set index to Unix timestamps\n",
    "past_close.set_index(df['UTC timestamp'][:len(past_close)],inplace=True)\n",
    "\n",
    "# Write future timestamps to predictions\n",
    "timestamp_len_ms = past_close.index[1] - past_close.index[0]\n",
    "future_timestamps = []\n",
    "for i in range(1,len(prediction)+1):\n",
    "    timestamp = past_close.index[-1]+timestamp_len_ms*i\n",
    "    future_timestamps.append(timestamp)\n",
    "prediction.set_index(pd.Series(future_timestamps),inplace=True)\n",
    "prediction = prediction.round(2)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "id": "81cf7fb9",
   "metadata": {},
   "outputs": [],
   "source": [
    "return_dict = {\n",
    "    \"symbol\":symbol,\n",
    "    \"past\":json.loads(past_close.to_json()),\n",
    "    \"prediction\":json.loads(prediction.to_json()),\n",
    "    \"period\":period,\n",
    "    \"MAPE\":backtest_mape\n",
    "}\n",
    "\n",
    "json.dump(return_dict, open(f\"{symbol}.json\",'w'), indent=4)\n",
    "# print(json.dumps(return_dict, indent=4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "d0408692",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#tinydb\n",
    "from tinydb import TinyDB, Query\n",
    "db = TinyDB('db.json')\n",
    "table = db.table('forecasts')\n",
    "table.insert(return_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f8b4561a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# read json\n",
    "import json\n",
    "forecasts = []\n",
    "for s in ['BTC-USD','ETH-USD']:\n",
    "    read_json = json.load(open(f\"{s}.json\",'r'))\n",
    "    forecasts.append(read_json)\n",
    "    \n",
    "# print(json.dumps(read_json, indent=4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cd8c0fad",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Start mongo if not running\n",
    "!docker run -p 27017:27017 -v $(pwd):/out --rm -d mongo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "b5e2a2bc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "BTC-USD\n",
      "ETH-USD\n",
      "dict_keys(['_id', 'symbol', 'past', 'prediction', 'period', 'MAPE'])\n",
      "dict_keys(['_id', 'symbol', 'past', 'prediction', 'period', 'MAPE'])\n"
     ]
    }
   ],
   "source": [
    "# Write forecasts to mongo\n",
    "import pymongo\n",
    "client = pymongo.MongoClient(\"localhost\", 27017)\n",
    "db = client.db_name\n",
    "\n",
    "for f in forecasts:\n",
    "    db.forecasts.delete_many({'symbol':f['symbol'],'period':f['period']})\n",
    "    db.forecasts.insert_one(f)\n",
    "\n",
    "cursor = db.forecasts.find({})\n",
    "for c in cursor:\n",
    "    print(c['symbol'])\n",
    "    \n",
    "cursor = db.forecasts.find()\n",
    "import pprint\n",
    "# pprint.pprint(cursor)\n",
    "for c in cursor:\n",
    "    pprint.pprint(c.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "63002805",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Ensemble\n",
    "from darts.models.forecasting.regression_ensemble_model import RegressionEnsembleModel\n",
    "from darts.models.forecasting.baselines import NaiveEnsembleModel\n",
    "\n",
    "TCN = TCNModel(input_chunk_length=48, output_chunk_length=24, n_epochs=21, random_state=0,optimizer_kwargs={'lr':.001})\n",
    "RNN = RNNModel(model=\"GRU\",input_chunk_length=48, output_chunk_length=24, n_epochs=21, random_state=0,optimizer_kwargs={'lr':.001})\n",
    "# GBM = LightGBMModel(lags=[-1,-2,-21,-34,-68,-3,-5,-8,-13])\n",
    "\n",
    "\n",
    "ensemble = RegressionEnsembleModel([TCN,RNN],regression_train_n_points=4)\n",
    "# ensemble = NaiveEnsembleModel([TCN,RNN])\n",
    "\n",
    "ensemble.fit(train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0043e828",
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_cov = ensemble.predict(n=len(train)//4, series=train,\n",
    "                             # past_covariates=scaled_covars\n",
    "                             )\n",
    "target_var = 'close'\n",
    "scaled[target_var].plot(label='actual')\n",
    "pred_cov[target_var].plot(label='forecast',low_quantile=0.05, high_quantile=0.95)\n",
    "# plt.legend();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1ca7c310",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}