research_amp/cc/notebooks/Master_machine_learning.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "54da1cc1",
   "metadata": {},
   "source": [
    "# Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "44249f9a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:32:39.872102Z",
     "start_time": "2022-05-31T19:32:35.779728Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/app/helpers/hparquet.py:17: TqdmExperimentalWarning: Using `tqdm.autonotebook.tqdm` in notebook mode. Use `tqdm.tqdm` instead to force console mode (e.g. in jupyter console)\n",
      "  from tqdm.autonotebook import tqdm\n"
     ]
    }
   ],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "import logging\n",
    "import os\n",
    "import warnings\n",
    "from datetime import timedelta\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import sklearn.metrics as metrics\n",
    "from scipy import stats\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.model_selection import (\n",
    "    GridSearchCV,\n",
    "    TimeSeriesSplit,\n",
    "    cross_val_score,\n",
    "    train_test_split,\n",
    ")\n",
    "\n",
    "import core.config.config_ as cconconf\n",
    "import core.config.config_utils as ccocouti\n",
    "import core.features as cofeatur\n",
    "import helpers.hdbg as hdbg\n",
    "import helpers.hpandas as hpandas\n",
    "import helpers.hparquet as hparque\n",
    "import helpers.hprint as hprint\n",
    "import im_v2.crypto_chassis.data.client.crypto_chassis_clients as imvccdcccc\n",
    "import research_amp.cc.crypto_chassis_api as raccchap\n",
    "import research_amp.transform as ramptran"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "296c560f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:32:39.916921Z",
     "start_time": "2022-05-31T19:32:39.874616Z"
    }
   },
   "outputs": [],
   "source": [
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "6153685d",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:32:39.974558Z",
     "start_time": "2022-05-31T19:32:39.918968Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[0m\u001b[36mINFO\u001b[0m: > cmd='/venv/lib/python3.8/site-packages/ipykernel_launcher.py -f /home/.local/share/jupyter/runtime/kernel-ae9aac72-9204-4f1a-850d-5311b3824397.json'\n",
      "WARNING: Running in Jupyter\n",
      "\u001b[33mWARNING\u001b[0m: Disabling annoying warnings\n"
     ]
    }
   ],
   "source": [
    "hdbg.init_logger(verbosity=logging.INFO)\n",
    "\n",
    "_LOG = logging.getLogger(__name__)\n",
    "\n",
    "hprint.config_notebook()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "03cde933",
   "metadata": {},
   "source": [
    "# Config"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "542fec59",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:32:40.027181Z",
     "start_time": "2022-05-31T19:32:39.978097Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "data:\n",
      "  im_client_ohlcv:\n",
      "    universe_version: v1\n",
      "    resample_1min: True\n",
      "    root_dir: s3://cryptokaizen-data/reorg/historical.manual.pq\n",
      "    partition_mode: by_year_month\n",
      "    data_snapshot: latest\n",
      "    aws_profile: ck\n",
      "  im_client_bid_ask:\n",
      "    base_path: s3://cryptokaizen-data/reorg/historical.manual.pq/20220520/bid_ask/crypto_chassis/binance/currency_pair=BTC_USDT/year=2022/\n",
      "    aws_profile: ck\n",
      "    resample_bid_ask: 1T\n",
      "  read_data:\n",
      "    full_symbols: ['binance::BTC_USDT']\n",
      "    start_ts: 2022-01-01 00:00:00+00:00\n",
      "    end_ts: 2022-03-31 23:59:00+00:00\n",
      "    columns: ['close', 'full_symbol', 'volume']\n",
      "    filter_data_mode: assert\n",
      "model:\n",
      "  resampling_rule: 5T\n",
      "  target_value: volume\n",
      "  delay_lag: 1\n",
      "  num_lags: 4\n",
      "  test_size: 0.2\n",
      "  shuffle: False\n",
      "  n_splits: 5\n"
     ]
    }
   ],
   "source": [
    "def get_cmtask1953_config() -> cconconf.Config:\n",
    "    config = cconconf.Config()\n",
    "    param_dict = {\n",
    "        \"data\": {\n",
    "            # Parameters for client initialization.\n",
    "            \"im_client_ohlcv\": {\n",
    "                \"universe_version\": \"v1\",\n",
    "                \"resample_1min\": True,\n",
    "                \"root_dir\": os.path.join(\n",
    "                    # hs3.get_s3_bucket_path(\"ck\"),\n",
    "                    \"s3://cryptokaizen-data\",\n",
    "                    \"reorg\",\n",
    "                    \"historical.manual.pq\",\n",
    "                ),\n",
    "                \"partition_mode\": \"by_year_month\",\n",
    "                \"dataset\": \"ohlcv\",\n",
    "                \"contract_type\": \"spot\",\n",
    "                \"data_snapshot\": \"latest\",\n",
    "                \"aws_profile\": \"ck\",\n",
    "            },\n",
    "            \"im_client_bid_ask\": {\n",
    "                \"base_path\": \"s3://cryptokaizen-data/reorg/historical.manual.pq/20220520/bid_ask/crypto_chassis/binance/currency_pair=BTC_USDT/year=2022/\",\n",
    "                \"aws_profile\": \"ck\",\n",
    "                \"resample_bid_ask\": \"1T\",\n",
    "            },\n",
    "            # Parameters for data query.\n",
    "            \"read_data\": {\n",
    "                \"full_symbols\": [\"binance::BTC_USDT\"],\n",
    "                \"start_ts\": pd.Timestamp(\"2022-01-01 00:00\", tz=\"UTC\"),\n",
    "                \"end_ts\": pd.Timestamp(\"2022-03-31 23:59\", tz=\"UTC\"),\n",
    "                \"columns\": [\"close\", \"full_symbol\", \"volume\"],\n",
    "                \"filter_data_mode\": \"assert\",\n",
    "            },\n",
    "        },\n",
    "        \"model\": {\n",
    "            \"resampling_rule\": \"5T\",\n",
    "            \"target_value\": \"volume\",\n",
    "            \"delay_lag\": 1,\n",
    "            \"num_lags\": 4,\n",
    "            \"test_size\": 0.2,\n",
    "            \"shuffle\": False,\n",
    "            \"n_splits\": 5,\n",
    "        },\n",
    "    }\n",
    "    config = cconfig.Config.from_dict(param_dict)\n",
    "    return config\n",
    "\n",
    "\n",
    "config = get_cmtask1953_config()\n",
    "print(config)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "74027f12",
   "metadata": {},
   "source": [
    "# Read data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "49f1fd6d",
   "metadata": {},
   "source": [
    "## OHLCV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "fddd0618",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:32:42.007038Z",
     "start_time": "2022-05-31T19:32:40.030277Z"
    }
   },
   "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>full_symbol</th>\n",
       "      <th>close</th>\n",
       "      <th>volume</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>timestamp</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:00:00+00:00</th>\n",
       "      <td>binance::BTC_USDT</td>\n",
       "      <td>46250.00</td>\n",
       "      <td>40.57574</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:01:00+00:00</th>\n",
       "      <td>binance::BTC_USDT</td>\n",
       "      <td>46312.76</td>\n",
       "      <td>42.38106</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:02:00+00:00</th>\n",
       "      <td>binance::BTC_USDT</td>\n",
       "      <td>46368.73</td>\n",
       "      <td>51.29955</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 full_symbol     close    volume\n",
       "timestamp                                                       \n",
       "2022-01-01 00:00:00+00:00  binance::BTC_USDT  46250.00  40.57574\n",
       "2022-01-01 00:01:00+00:00  binance::BTC_USDT  46312.76  42.38106\n",
       "2022-01-01 00:02:00+00:00  binance::BTC_USDT  46368.73  51.29955"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Initiate the client.\n",
    "client = imvccdcccc.CryptoChassisHistoricalPqByTileClient(\n",
    "    **config[\"data\"][\"im_client_ohlcv\"]\n",
    ")\n",
    "# Load OHLCV data.\n",
    "df_ohlcv = client.read_data(**config[\"data\"][\"read_data\"])\n",
    "# Resample.\n",
    "df_ohlcv.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "70c0fa2b",
   "metadata": {},
   "source": [
    "## Bid ask data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "0304ad14",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:32:42.061533Z",
     "start_time": "2022-05-31T19:32:42.009840Z"
    }
   },
   "outputs": [],
   "source": [
    "def resample_and_process_bid_ask_data(df, resample_rule):\n",
    "    # Resample.\n",
    "    df = raccchap.resample_bid_ask(df, resample_rule)\n",
    "    # Convert.\n",
    "    for cols in df.columns[:-1]:\n",
    "        df[cols] = pd.to_numeric(df[cols], downcast=\"float\")\n",
    "    # Compute bid ask stats.\n",
    "    df = ramptran.calculate_bid_ask_statistics(df)\n",
    "    # Choose only necessary values (`full_symbol`).\n",
    "    df = df.swaplevel(axis=1)[\n",
    "        str(config[\"data\"][\"read_data\"][\"full_symbols\"])[2:-2]\n",
    "    ][[\"bid_size\", \"ask_size\", \"bid_price\", \"ask_price\", \"mid\", \"quoted_spread\"]]\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "c51d8ea0",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:33:04.748998Z",
     "start_time": "2022-05-31T19:32:42.064922Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1\n",
      "2\n",
      "3\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>bid_size</th>\n",
       "      <th>ask_size</th>\n",
       "      <th>bid_price</th>\n",
       "      <th>ask_price</th>\n",
       "      <th>mid</th>\n",
       "      <th>quoted_spread</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>timestamp</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:00:00+00:00</th>\n",
       "      <td>1.08286</td>\n",
       "      <td>0.24793</td>\n",
       "      <td>46214.011719</td>\n",
       "      <td>46214.019531</td>\n",
       "      <td>46214.015625</td>\n",
       "      <td>0.007812</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:01:00+00:00</th>\n",
       "      <td>0.31064</td>\n",
       "      <td>0.47658</td>\n",
       "      <td>46250.000000</td>\n",
       "      <td>46250.011719</td>\n",
       "      <td>46250.007812</td>\n",
       "      <td>0.011719</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:02:00+00:00</th>\n",
       "      <td>0.99036</td>\n",
       "      <td>0.03050</td>\n",
       "      <td>46310.011719</td>\n",
       "      <td>46318.109375</td>\n",
       "      <td>46314.062500</td>\n",
       "      <td>8.097656</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                           bid_size  ask_size     bid_price     ask_price           mid  quoted_spread\n",
       "timestamp                                                                                             \n",
       "2022-01-01 00:00:00+00:00   1.08286   0.24793  46214.011719  46214.019531  46214.015625       0.007812\n",
       "2022-01-01 00:01:00+00:00   0.31064   0.47658  46250.000000  46250.011719  46250.007812       0.011719\n",
       "2022-01-01 00:02:00+00:00   0.99036   0.03050  46310.011719  46318.109375  46314.062500       8.097656"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "start_date = config[\"data\"][\"read_data\"][\"start_ts\"]\n",
    "end_date = config[\"data\"][\"read_data\"][\"end_ts\"]\n",
    "\n",
    "# Load bid ask from s3. Note: works only for 2022 for now.\n",
    "# TODO(Grisha, Dan): How to load the bid/ask data through ImClient?\n",
    "result = []\n",
    "\n",
    "for i in range(start_date.month, end_date.month + 1):\n",
    "    print(i)\n",
    "    tmp_df = hparque.from_parquet(\n",
    "        os.path.join(\n",
    "            config[\"data\"][\"im_client_bid_ask\"][\"base_path\"],\n",
    "            f\"month={i}/data.parquet\",\n",
    "        ),\n",
    "        aws_profile=config[\"data\"][\"im_client_bid_ask\"][\"aws_profile\"],\n",
    "    )\n",
    "    result.append(tmp_df)\n",
    "bid_ask_df = pd.concat(result)\n",
    "bid_ask_df = bid_ask_df[:end_date]\n",
    "# Add `full_symbol` (necessary param for `calculate_bid_ask_statistics`).\n",
    "bid_ask_df[\"full_symbol\"] = str(config[\"data\"][\"read_data\"][\"full_symbols\"])[2:-2]\n",
    "# Choose only valid cols.\n",
    "bid_ask_df = bid_ask_df[\n",
    "    [\"bid_price\", \"bid_size\", \"ask_price\", \"ask_size\", \"full_symbol\"]\n",
    "]\n",
    "# Resample to 1-min (to be consistent with OHLCV data).\n",
    "bid_ask_df_1min = resample_and_process_bid_ask_data(\n",
    "    bid_ask_df, config[\"data\"][\"im_client_bid_ask\"][\"resample_bid_ask\"]\n",
    ")\n",
    "\n",
    "bid_ask_df_1min.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a3aad4de",
   "metadata": {},
   "source": [
    "## Combined"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "73201765",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:33:04.810217Z",
     "start_time": "2022-05-31T19:33:04.752334Z"
    }
   },
   "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>full_symbol</th>\n",
       "      <th>close</th>\n",
       "      <th>volume</th>\n",
       "      <th>bid_size</th>\n",
       "      <th>ask_size</th>\n",
       "      <th>bid_price</th>\n",
       "      <th>ask_price</th>\n",
       "      <th>mid</th>\n",
       "      <th>quoted_spread</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>timestamp</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:00:00+00:00</th>\n",
       "      <td>binance::BTC_USDT</td>\n",
       "      <td>46250.00</td>\n",
       "      <td>40.57574</td>\n",
       "      <td>1.08286</td>\n",
       "      <td>0.24793</td>\n",
       "      <td>46214.011719</td>\n",
       "      <td>46214.019531</td>\n",
       "      <td>46214.015625</td>\n",
       "      <td>0.007812</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:01:00+00:00</th>\n",
       "      <td>binance::BTC_USDT</td>\n",
       "      <td>46312.76</td>\n",
       "      <td>42.38106</td>\n",
       "      <td>0.31064</td>\n",
       "      <td>0.47658</td>\n",
       "      <td>46250.000000</td>\n",
       "      <td>46250.011719</td>\n",
       "      <td>46250.007812</td>\n",
       "      <td>0.011719</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:02:00+00:00</th>\n",
       "      <td>binance::BTC_USDT</td>\n",
       "      <td>46368.73</td>\n",
       "      <td>51.29955</td>\n",
       "      <td>0.99036</td>\n",
       "      <td>0.03050</td>\n",
       "      <td>46310.011719</td>\n",
       "      <td>46318.109375</td>\n",
       "      <td>46314.062500</td>\n",
       "      <td>8.097656</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 full_symbol     close    volume  bid_size  ask_size     bid_price     ask_price           mid  quoted_spread\n",
       "timestamp                                                                                                                                    \n",
       "2022-01-01 00:00:00+00:00  binance::BTC_USDT  46250.00  40.57574   1.08286   0.24793  46214.011719  46214.019531  46214.015625       0.007812\n",
       "2022-01-01 00:01:00+00:00  binance::BTC_USDT  46312.76  42.38106   0.31064   0.47658  46250.000000  46250.011719  46250.007812       0.011719\n",
       "2022-01-01 00:02:00+00:00  binance::BTC_USDT  46368.73  51.29955   0.99036   0.03050  46310.011719  46318.109375  46314.062500       8.097656"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# OHLCV + bid ask\n",
    "data = pd.concat([df_ohlcv, bid_ask_df_1min], axis=1)\n",
    "data.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ace30295",
   "metadata": {},
   "source": [
    "# Compute features\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "6b46e4c7",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:33:04.864553Z",
     "start_time": "2022-05-31T19:33:04.813457Z"
    }
   },
   "outputs": [],
   "source": [
    "def get_average_intraday_value(\n",
    "    df: pd.DataFrame,\n",
    "    timestamp: pd.Timestamp,\n",
    "    lookback_days: int,\n",
    "    column_name: str,\n",
    "    delay: int = 0,\n",
    "    mode: str = \"mean\",\n",
    ") -> float:\n",
    "    \"\"\"\n",
    "    Compute a feature as the mean / median of the target variable in the same\n",
    "    period of time using data in the last `lookback_days` days.\n",
    "\n",
    "    This is useful for volume, spread, volatility.\n",
    "\n",
    "    1) Set the period that is equal `timestamp for prediciton` - N days (lookback_days).\n",
    "    2) For that period, calculate mean (or median) value for target in time during days.\n",
    "    3) Choose this mean value as an estimation for target in the given timestamp.\n",
    "\n",
    "    :param df: data that contains target value\n",
    "    :param timestamp: timestamp for prediciton\n",
    "    :param lookback_days: historical period for estimation, in days\n",
    "    :param column_name: targeted estimation value (e.g., \"quoted_spread\", \"volume\")\n",
    "    :param delay: how many minutes to substract from the lookback starting period\n",
    "    :param mode: 'mean' or 'median'\n",
    "    :return: value of predicted target\n",
    "    \"\"\"\n",
    "    # Choose sample data using lookback period (with a delay).\n",
    "    start_date = timestamp - timedelta(days=lookback_days, minutes=delay)\n",
    "    if start_date >= df.index.min() and start_date <= df.index.max():\n",
    "        sample = df.loc[start_date:timestamp].loc[timestamp.time()]\n",
    "        if mode == \"mean\":\n",
    "            value = sample[column_name].mean()\n",
    "        else:\n",
    "            value = sample[column_name].median()\n",
    "    else:\n",
    "        value = np.nan\n",
    "    return value"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "be2f5021",
   "metadata": {},
   "source": [
    "# Assemble ml_df\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ff993115",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:34:59.829350Z",
     "start_time": "2022-05-31T19:33:04.866950Z"
    },
    "run_control": {
     "marked": false
    }
   },
   "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>volume_2</th>\n",
       "      <th>volume_3</th>\n",
       "      <th>volume_4</th>\n",
       "      <th>volume_5</th>\n",
       "      <th>avg_intraday_volume</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>timestamp</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:05:00+00:00</th>\n",
       "      <td>30.45894</td>\n",
       "      <td>51.29955</td>\n",
       "      <td>42.38106</td>\n",
       "      <td>40.57574</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:06:00+00:00</th>\n",
       "      <td>20.96029</td>\n",
       "      <td>30.45894</td>\n",
       "      <td>51.29955</td>\n",
       "      <td>42.38106</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:07:00+00:00</th>\n",
       "      <td>35.86682</td>\n",
       "      <td>20.96029</td>\n",
       "      <td>30.45894</td>\n",
       "      <td>51.29955</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:08:00+00:00</th>\n",
       "      <td>29.31849</td>\n",
       "      <td>35.86682</td>\n",
       "      <td>20.96029</td>\n",
       "      <td>30.45894</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:09:00+00:00</th>\n",
       "      <td>27.81847</td>\n",
       "      <td>29.31849</td>\n",
       "      <td>35.86682</td>\n",
       "      <td>20.96029</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-03-31 23:55:00+00:00</th>\n",
       "      <td>106.46153</td>\n",
       "      <td>27.45584</td>\n",
       "      <td>27.94996</td>\n",
       "      <td>43.41923</td>\n",
       "      <td>26.479994</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-03-31 23:56:00+00:00</th>\n",
       "      <td>38.34090</td>\n",
       "      <td>106.46153</td>\n",
       "      <td>27.45584</td>\n",
       "      <td>27.94996</td>\n",
       "      <td>20.250477</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-03-31 23:57:00+00:00</th>\n",
       "      <td>33.95752</td>\n",
       "      <td>38.34090</td>\n",
       "      <td>106.46153</td>\n",
       "      <td>27.45584</td>\n",
       "      <td>36.568241</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-03-31 23:58:00+00:00</th>\n",
       "      <td>64.97414</td>\n",
       "      <td>33.95752</td>\n",
       "      <td>38.34090</td>\n",
       "      <td>106.46153</td>\n",
       "      <td>16.825945</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-03-31 23:59:00+00:00</th>\n",
       "      <td>16.55855</td>\n",
       "      <td>64.97414</td>\n",
       "      <td>33.95752</td>\n",
       "      <td>38.34090</td>\n",
       "      <td>18.972656</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>129595 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                            volume_2   volume_3   volume_4   volume_5  avg_intraday_volume\n",
       "timestamp                                                                                 \n",
       "2022-01-01 00:05:00+00:00   30.45894   51.29955   42.38106   40.57574                  NaN\n",
       "2022-01-01 00:06:00+00:00   20.96029   30.45894   51.29955   42.38106                  NaN\n",
       "2022-01-01 00:07:00+00:00   35.86682   20.96029   30.45894   51.29955                  NaN\n",
       "2022-01-01 00:08:00+00:00   29.31849   35.86682   20.96029   30.45894                  NaN\n",
       "2022-01-01 00:09:00+00:00   27.81847   29.31849   35.86682   20.96029                  NaN\n",
       "...                              ...        ...        ...        ...                  ...\n",
       "2022-03-31 23:55:00+00:00  106.46153   27.45584   27.94996   43.41923            26.479994\n",
       "2022-03-31 23:56:00+00:00   38.34090  106.46153   27.45584   27.94996            20.250477\n",
       "2022-03-31 23:57:00+00:00   33.95752   38.34090  106.46153   27.45584            36.568241\n",
       "2022-03-31 23:58:00+00:00   64.97414   33.95752   38.34090  106.46153            16.825945\n",
       "2022-03-31 23:59:00+00:00   16.55855   64.97414   33.95752   38.34090            18.972656\n",
       "\n",
       "[129595 rows x 5 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Specify target value.\n",
    "target = config[\"model\"][\"target_value\"]\n",
    "# Add column with intraday time.\n",
    "data[\"time\"] = data.index.time\n",
    "# Add initial target values.\n",
    "ml_df = data[[target]]\n",
    "ml_df.columns = [f\"{target}_0\"]\n",
    "# Add lagged values.\n",
    "delay_lag = config[\"model\"][\"delay_lag\"]\n",
    "num_lags = config[\"model\"][\"num_lags\"]\n",
    "ml_df, info = cofeatur.compute_lagged_features(\n",
    "    ml_df, f\"{target}_0\", delay_lag, num_lags\n",
    ")\n",
    "# Add lookback estimator.\n",
    "ml_df[f\"avg_intraday_{target}\"] = ml_df.index\n",
    "ml_df[f\"avg_intraday_{target}\"] = ml_df[f\"avg_intraday_{target}\"].apply(\n",
    "    lambda x: get_average_intraday_value(data, x, 14, target)\n",
    ")\n",
    "# Drop the column with `{target}_0`, since Y-var will be added later (resampled).\n",
    "ml_df = ml_df.drop(columns=f\"{target}_0\")\n",
    "ml_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bfae799c",
   "metadata": {},
   "source": [
    "# Y pre-processing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "12610576",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:34:59.873681Z",
     "start_time": "2022-05-31T19:34:59.831746Z"
    }
   },
   "outputs": [],
   "source": [
    "def attach_resampled_y_var(\n",
    "    resampled_df: pd.DataFrame, estimators_df: pd.DataFrame, target: str\n",
    ") -> pd.DataFrame:\n",
    "    \"\"\"\n",
    "    :param resampled_df: Data for Y-var with >1-min frequency\n",
    "    :param estimators_df: Data for X-vars with 1-min frequency\n",
    "    :param target: i.e., \"spread\" or \"volume\"\n",
    "    \"\"\"\n",
    "    if target == \"spread\":\n",
    "        # Choose Y-var.\n",
    "        resampled_df = resampled_df[[f\"quoted_{target}\"]]\n",
    "        # Rename Y-var.\n",
    "        resampled_df = resampled_df.rename(columns={\"quoted_spread\": \"spread_0\"})\n",
    "    elif target == \"volume\":\n",
    "        # Choose Y-var.\n",
    "        resampled_df = resampled_df[[f\"{target}\"]]\n",
    "        # Rename Y-var.\n",
    "        resampled_df = resampled_df.rename(columns={\"volume\": \"volume_0\"})\n",
    "    else:\n",
    "        raise ValueError(f\"Invalid target='{target}'\")\n",
    "    # Attach Y-var to the computed estimators.\n",
    "    yx_df = pd.merge(\n",
    "        resampled_df, estimators_df, left_index=True, right_index=True\n",
    "    )\n",
    "    return yx_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "64154228",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:04.093494Z",
     "start_time": "2022-05-31T19:34:59.875413Z"
    }
   },
   "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>volume_0</th>\n",
       "      <th>volume_2</th>\n",
       "      <th>volume_3</th>\n",
       "      <th>volume_4</th>\n",
       "      <th>volume_5</th>\n",
       "      <th>avg_intraday_volume</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>timestamp</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:05:00+00:00</th>\n",
       "      <td>123.43577</td>\n",
       "      <td>30.45894</td>\n",
       "      <td>51.29955</td>\n",
       "      <td>42.38106</td>\n",
       "      <td>40.57574</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:10:00+00:00</th>\n",
       "      <td>77.54574</td>\n",
       "      <td>15.96245</td>\n",
       "      <td>27.81847</td>\n",
       "      <td>29.31849</td>\n",
       "      <td>35.86682</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:15:00+00:00</th>\n",
       "      <td>101.14315</td>\n",
       "      <td>23.15078</td>\n",
       "      <td>11.62620</td>\n",
       "      <td>19.01268</td>\n",
       "      <td>10.65734</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:20:00+00:00</th>\n",
       "      <td>135.32479</td>\n",
       "      <td>24.88379</td>\n",
       "      <td>11.99892</td>\n",
       "      <td>24.76988</td>\n",
       "      <td>8.40902</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-01 00:25:00+00:00</th>\n",
       "      <td>83.53179</td>\n",
       "      <td>9.44557</td>\n",
       "      <td>35.61633</td>\n",
       "      <td>38.03832</td>\n",
       "      <td>26.61854</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                            volume_0  volume_2  volume_3  volume_4  volume_5  avg_intraday_volume\n",
       "timestamp                                                                                        \n",
       "2022-01-01 00:05:00+00:00  123.43577  30.45894  51.29955  42.38106  40.57574                  NaN\n",
       "2022-01-01 00:10:00+00:00   77.54574  15.96245  27.81847  29.31849  35.86682                  NaN\n",
       "2022-01-01 00:15:00+00:00  101.14315  23.15078  11.62620  19.01268  10.65734                  NaN\n",
       "2022-01-01 00:20:00+00:00  135.32479  24.88379  11.99892  24.76988   8.40902                  NaN\n",
       "2022-01-01 00:25:00+00:00   83.53179   9.44557  35.61633  38.03832  26.61854                  NaN"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# The goal is to set Y-var with 5-min frequency and\n",
    "# attach previously calculated features with 1-min frequency.\n",
    "resampling_rule = config[\"model\"][\"resampling_rule\"]\n",
    "# Resample initial bid ask.\n",
    "bid_ask_df_5min = resample_and_process_bid_ask_data(bid_ask_df, resampling_rule)\n",
    "# Resample initial OHLCV.\n",
    "df_ohlcv_5min = df_ohlcv.resample(resampling_rule).agg(\n",
    "    {\"close\": \"last\", \"volume\": \"sum\"}\n",
    ")\n",
    "# Combine resampled data.\n",
    "df_5min = pd.concat([df_ohlcv_5min, bid_ask_df_5min], axis=1)\n",
    "# Update DataFrame with ML features (add resampled y-var).\n",
    "ml_df = attach_resampled_y_var(df_5min, ml_df, target)\n",
    "\n",
    "# Now we have Y-var that is sampled by 5-min, while X-vars are 1-mins.\n",
    "ml_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d005654f",
   "metadata": {},
   "source": [
    "# Train / test separation\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "2f62c172",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:04.212066Z",
     "start_time": "2022-05-31T19:35:04.105115Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO  removed rows with nans: 4031 / 25919 = 15.55%\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(21888, 6)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>volume_0</th>\n",
       "      <th>volume_2</th>\n",
       "      <th>volume_3</th>\n",
       "      <th>volume_4</th>\n",
       "      <th>volume_5</th>\n",
       "      <th>avg_intraday_volume</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>timestamp</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2022-01-15 00:00:00+00:00</th>\n",
       "      <td>116.30611</td>\n",
       "      <td>5.01042</td>\n",
       "      <td>4.82233</td>\n",
       "      <td>13.28030</td>\n",
       "      <td>3.90642</td>\n",
       "      <td>34.664149</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-15 00:05:00+00:00</th>\n",
       "      <td>101.01256</td>\n",
       "      <td>24.37564</td>\n",
       "      <td>22.57418</td>\n",
       "      <td>19.15164</td>\n",
       "      <td>27.88350</td>\n",
       "      <td>22.349086</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2022-01-15 00:10:00+00:00</th>\n",
       "      <td>49.98275</td>\n",
       "      <td>39.70073</td>\n",
       "      <td>10.44870</td>\n",
       "      <td>17.31996</td>\n",
       "      <td>25.61504</td>\n",
       "      <td>18.089273</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                            volume_0  volume_2  volume_3  volume_4  volume_5  avg_intraday_volume\n",
       "timestamp                                                                                        \n",
       "2022-01-15 00:00:00+00:00  116.30611   5.01042   4.82233  13.28030   3.90642            34.664149\n",
       "2022-01-15 00:05:00+00:00  101.01256  24.37564  22.57418  19.15164  27.88350            22.349086\n",
       "2022-01-15 00:10:00+00:00   49.98275  39.70073  10.44870  17.31996  25.61504            18.089273"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Set of prediciton features = ['volume_2', 'volume_3', 'volume_4', 'volume_5', 'avg_intraday_volume']\n"
     ]
    }
   ],
   "source": [
    "# Specify modelling data.\n",
    "ml_df = hpandas.dropna(ml_df, report_stats=True)\n",
    "display(ml_df.shape)\n",
    "display(ml_df.head(3))\n",
    "print(f\"Set of prediciton features = {list(ml_df.columns[1:])}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "1b588e13",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:04.274276Z",
     "start_time": "2022-05-31T19:35:04.219615Z"
    }
   },
   "outputs": [],
   "source": [
    "# Construct X and y.\n",
    "y_var_col = f\"{target}_0\"\n",
    "y = ml_df[[y_var_col]]\n",
    "X = ml_df.drop(columns=[y_var_col])\n",
    "# Split into train and test sets.\n",
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    X,\n",
    "    y,\n",
    "    test_size=config[\"model\"][\"test_size\"],\n",
    "    shuffle=config[\"model\"][\"shuffle\"],\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9bbd5f4e",
   "metadata": {},
   "source": [
    "# Model set-up"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "b8451b38",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:04.331248Z",
     "start_time": "2022-05-31T19:35:04.276813Z"
    }
   },
   "outputs": [],
   "source": [
    "def regression_results(y_true, y_pred, mae_only: bool = True):\n",
    "    # Regression metrics\n",
    "    mean_absolute_error = metrics.mean_absolute_error(y_true, y_pred)\n",
    "    print(\"MAE: \", round(mean_absolute_error, 4))\n",
    "    if not mae_only:\n",
    "        explained_variance = metrics.explained_variance_score(y_true, y_pred)\n",
    "        mse = metrics.mean_squared_error(y_true, y_pred)\n",
    "        mean_squared_log_error = metrics.mean_squared_log_error(y_true, y_pred)\n",
    "        metrics.median_absolute_error(y_true, y_pred)\n",
    "        r2 = metrics.r2_score(y_true, y_pred)\n",
    "        print(\"mean_squared_log_error: \", round(mean_squared_log_error, 4))\n",
    "        print(\"explained_variance: \", round(explained_variance, 4))\n",
    "        print(\"r2: \", round(r2, 4))\n",
    "        print(\"MAE: \", round(mean_absolute_error, 4))\n",
    "        print(\"MSE: \", round(mse, 4))\n",
    "        print(\"RMSE: \", round(np.sqrt(mse), 4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "5e24d6a9",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:04.405362Z",
     "start_time": "2022-05-31T19:35:04.339765Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of splits: 5\n"
     ]
    }
   ],
   "source": [
    "n_splits = config[\"model\"][\"n_splits\"]\n",
    "print(f\"Number of splits: {n_splits}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "74b4b890",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:04.468717Z",
     "start_time": "2022-05-31T19:35:04.407365Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('LR', LinearRegression())]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create a set of various estimation modes.\n",
    "models = []\n",
    "models.append((\"LR\", LinearRegression()))\n",
    "# models.append((\"NN\", MLPRegressor(solver=\"lbfgs\")))  # neural network\n",
    "# models.append((\"KNN\", KNeighborsRegressor()))\n",
    "# models.append(\n",
    "#    (\"RF\", RandomForestRegressor(n_estimators=10))\n",
    "# )  # Ensemble method - collection of many decision trees\n",
    "# models.append((\"SVR\", SVR(gamma=\"auto\")))  # kernel = linear\n",
    "models"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c79e3ba2",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-30T17:52:03.449332Z",
     "start_time": "2022-05-30T17:52:03.402861Z"
    }
   },
   "source": [
    "# Learn / evaluate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "783f84fa",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:04.719237Z",
     "start_time": "2022-05-31T19:35:04.471215Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LR: 0.332109 (0.063132)\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>mean_perf</th>\n",
       "      <th>std_dev_perf</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>LR</th>\n",
       "      <td>0.332109</td>\n",
       "      <td>0.063132</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    mean_perf  std_dev_perf\n",
       "LR   0.332109      0.063132"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Evaluate each model in turn\n",
    "results = []\n",
    "names = []\n",
    "results_stats = pd.DataFrame()\n",
    "for name, model in models:\n",
    "    # TimeSeries Cross validation\n",
    "    tscv = TimeSeriesSplit(n_splits=n_splits)\n",
    "\n",
    "    cv_results = cross_val_score(model, X_train, y_train, cv=tscv, scoring=\"r2\")\n",
    "    results.append(cv_results)\n",
    "    names.append(name)\n",
    "    print(\"%s: %f (%f)\" % (name, cv_results.mean(), cv_results.std()))\n",
    "\n",
    "    results_stats.loc[name, \"mean_perf\"] = cv_results.mean()\n",
    "    results_stats.loc[name, \"std_dev_perf\"] = cv_results.std()\n",
    "\n",
    "display(results_stats.sort_values(\"mean_perf\", ascending=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "527b5e5c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:05.037412Z",
     "start_time": "2022-05-31T19:35:04.735242Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Compare Algorithms\n",
    "plt.boxplot(results, labels=names)\n",
    "plt.title(\"Algorithm Comparison\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "2f65f457",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:05.094141Z",
     "start_time": "2022-05-31T19:35:05.040354Z"
    }
   },
   "outputs": [],
   "source": [
    "# TODO(Max): consider adding the hyperparameters tuning step."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "81cd750a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:05.145789Z",
     "start_time": "2022-05-31T19:35:05.096239Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAE:  76.1126\n"
     ]
    }
   ],
   "source": [
    "# Train the model.\n",
    "model = LinearRegression()\n",
    "model = model.fit(X_train, y_train)\n",
    "\n",
    "# Estimate testing results.\n",
    "y_true = y_test.values\n",
    "y_pred = model.predict(X_test)\n",
    "regression_results(y_true, y_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "221049b0",
   "metadata": {},
   "source": [
    "# Model analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "5868bf98",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T20:52:29.721202Z",
     "start_time": "2022-05-31T20:52:29.670709Z"
    }
   },
   "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>coef_value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>avg_intraday_volume</th>\n",
       "      <td>1.878690</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>volume_2</th>\n",
       "      <td>1.066354</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>volume_5</th>\n",
       "      <td>0.607097</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>volume_3</th>\n",
       "      <td>0.532742</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>volume_4</th>\n",
       "      <td>0.456847</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     coef_value\n",
       "avg_intraday_volume    1.878690\n",
       "volume_2               1.066354\n",
       "volume_5               0.607097\n",
       "volume_3               0.532742\n",
       "volume_4               0.456847"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# TODO(Max): consider displaying p-values.\n",
    "# Check coefficients.\n",
    "coef = pd.DataFrame(\n",
    "    {\"coef_value\": model.coef_.ravel()},\n",
    "    index=model.feature_names_in_,\n",
    ")\n",
    "coef = coef.sort_values(by=\"coef_value\", ascending=False)\n",
    "coef"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "67de5b2d",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:05.788451Z",
     "start_time": "2022-05-31T19:35:05.292145Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='timestamp'>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the results of predicting on testing sample.\n",
    "lr_test = pd.concat(\n",
    "    [pd.Series(y_true.ravel()), pd.Series(y_pred.ravel())], axis=1\n",
    ")\n",
    "lr_test.columns = [\"true\", \"predicted\"]\n",
    "lr_test.index = y_test.index\n",
    "lr_test.plot(figsize=(15, 7))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "2b0619a0",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-05-31T19:35:06.231085Z",
     "start_time": "2022-05-31T19:35:05.790275Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='timestamp'>"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the difference between true and predicted values.\n",
    "lr_test[\"diff\"] = lr_test[\"true\"] - lr_test[\"predicted\"]\n",
    "lr_test[\"diff\"].plot(figsize=(15, 7))"
   ]
  }
 ],
 "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": "3.8.10"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}