cleaning.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def merge_dfs(dfs: list[pd.DataFrame]):\n",
    "    out_df = dfs[0]\n",
    "\n",
    "    for i in range(1, len(dfs)):\n",
    "        out_df = pd.merge(out_df, dfs[i], how=\"left\", on=[\"Country\", \"Date\"])\n",
    "\n",
    "    return out_df\n",
    "\n",
    "\n",
    "def generate_indications_cols(df: pd.DataFrame, num_ind: int, label: str, filler):\n",
    "    new_cols = pd.DataFrame.from_dict([\n",
    "        {\n",
    "            f\"{label}_ind{indication_num}\": filler\n",
    "                for indication_num in range(1, num_ind+1)\n",
    "        } \n",
    "        for _ in range(len(df))\n",
    "    ], dtype=float)\n",
    "\n",
    "    return pd.concat([ df, new_cols ], axis=1)\n",
    "\n",
    "def fix_date_missing_period(df: pd.DataFrame, col_name=\"Date\"):\n",
    "    month_map = {\n",
    "        \"janv\": \"Jan\",\n",
    "        \"févr\": \"Feb\",\n",
    "        \"mars\": \"Mar\",\n",
    "        \"avr\": \"Apr\",\n",
    "        \"mai\": \"May\",\n",
    "        \"juin\": \"Jun\",\n",
    "        \"juil\": \"Jul\",\n",
    "        \"août\": \"Aug\",\n",
    "        \"sept\": \"Sep\",\n",
    "        \"oct\": \"Oct\",\n",
    "        \"nov\": \"Nov\",\n",
    "        \"déc\": \"Dec\"\n",
    "    }\n",
    "    df[col_name] = df[col_name].apply(lambda x: f\"{month_map[x.split('-')[0]]}-{x.split('-')[1]}\")\n",
    "    df[col_name] = pd.to_datetime(df[col_name], format=\"%b-%y\")\n",
    "    return df\n",
    "\n",
    "\n",
    "def clean_innovix_ex_factory_vol(df: pd.DataFrame):\n",
    "    df.drop(columns=[\"Data type\", \"Unit of measure\"], inplace=True)\n",
    "    df.rename(columns={\"Value\": \"ex_factory_volumes\"}, inplace=True)\n",
    "    return df\n",
    "\n",
    "def clean_innovix_demand_volumes(df: pd.DataFrame, products):\n",
    "    df.drop(index=df.index[df[\"Unit of measure\"] == \"Milligrams\"], inplace=True)\n",
    "    df.drop(columns=[\"Data type\", \"Unit of measure\"], inplace=True)\n",
    "\n",
    "    for prod in products:\n",
    "        df[f\"{prod}_months_of_treatment\"] = 0.0\n",
    "\n",
    "    for i, row in df.iterrows():\n",
    "        df.loc[i, f\"{row['Product']}_months_of_treatment\"] = row[\"Value\"]\n",
    "\n",
    "    df.drop(columns=[\"Value\"], inplace=True)\n",
    "\n",
    "    country = df.iloc[0][\"Country\"]\n",
    "    df = df.groupby(\"Date\").sum(numeric_only=True)\n",
    "    df[\"Country\"] = country\n",
    "\n",
    "    return df\n",
    "\n",
    "def clean_innovix_activity(df: pd.DataFrame, num_ind: int, products: list[str]):\n",
    "    channel_name_map = {\n",
    "        \"Face to face call\": \"face-to-face-call\", \n",
    "        \"Email\": \"email\", \n",
    "        \"Remote call\": \"remote-call\", \n",
    "        \"Meetings\": \"meetings\"\n",
    "    }\n",
    "    df[\"Channel\"] = df[\"Channel\"].map(channel_name_map)\n",
    "    df[\"Indication\"] = df[\"Indication\"].map(lambda x: f\"ind{x.split(' ')[1]}\")\n",
    "\n",
    "    for prod in products:\n",
    "        for channel in df[\"Channel\"].unique().tolist():\n",
    "            df = generate_indications_cols(df, num_ind, f\"{prod}_{channel}\", 0.0)\n",
    "\n",
    "    for i, row in df.iterrows():\n",
    "        df.loc[i, f\"{row['Product']}_{row['Channel']}_{row['Indication']}\"] = row[\"Value\"]\n",
    "\n",
    "    df.drop(columns=[\"Data type\", \"Product\", \"Channel\", \"Indication\", \"Value\"], inplace = True)\n",
    "    country = df.iloc[0][\"Country\"]\n",
    "    df = df.groupby(\"Date\").sum(numeric_only=True)\n",
    "    df[\"Country\"] = country\n",
    "\n",
    "    return df\n",
    "\n",
    "def clean_innovix_share_of_voice(df: pd.DataFrame, num_ind: int, products):\n",
    "    df[\"Indication\"] = df[\"Indication\"].map(lambda x: f\"ind{x.split(' ')[1]}\")\n",
    "    for prod in products:\n",
    "        df = generate_indications_cols(df, num_ind, f\"{prod}_share_of_voice\", 0.0)\n",
    "\n",
    "    for i, row in df.iterrows():\n",
    "        df.loc[i, f\"{row['Product']}_share_of_voice_{row['Indication']}\"] = row[\"Value\"]\n",
    "\n",
    "    df.drop(columns=[\"Data type\", \"Indication\", \"Product\", \"Value\", \"Product\"], inplace=True)\n",
    "    country = df.iloc[0][\"Country\"]\n",
    "    df = df.groupby(\"Date\").sum(numeric_only=True)\n",
    "    df[\"Country\"] = country\n",
    "    return df\n",
    "\n",
    "def clean_innovix_new_patient_share(df: pd.DataFrame, num_ind: int, products):\n",
    "    df[\"Indication\"] = df[\"Indication\"].map(lambda x: f\"ind{x.split(' ')[1]}\")\n",
    "    for prod in products:\n",
    "        df = generate_indications_cols(df, num_ind, f\"{prod}_new_patient_shape\", 0.0)\n",
    "\n",
    "    for i, row in df.iterrows():\n",
    "        df.loc[i, f\"{row['Product']}_new_patient_shape_{row['Indication']}\"] = row[\"Value\"]\n",
    "\n",
    "    df.drop(columns=[\"Data type\", \"Indication\", \"Product\", \"Sub-Indication\", \"Value\"], inplace=True)\n",
    "    country = df.iloc[0][\"Country\"]\n",
    "    df = df.groupby(\"Date\").sum(numeric_only=True)\n",
    "    df[\"Country\"] = country\n",
    "    return df\n",
    "\n",
    "\n",
    "def clean_innovix_indication_split(df: pd.DataFrame, num_ind: int, products):\n",
    "    df[\"Indication\"] = df[\"Indication\"].map(lambda x: f\"ind{x.split(' ')[1]}\")\n",
    "    for prod in products:\n",
    "        df = generate_indications_cols(df, num_ind, f\"{prod}_indication_split\", 0.0)\n",
    "\n",
    "    for i, row in df.iterrows():\n",
    "        df.loc[i, f\"{row['Product']}_indication_split_{row['Indication']}\"] = row[\"Value\"]\n",
    "\n",
    "    df.drop(columns=[\"Data type\", \"Indication\", \"Product\", \"Sub-Indication\", \"Value\"], inplace=True)\n",
    "    country = df.iloc[0][\"Country\"]\n",
    "    df = df.groupby(\"Date\").sum(numeric_only=True)\n",
    "    df[\"Country\"] = country\n",
    "    return df\n",
    "\n",
    "\n",
    "df_floresland_ex_factory_vol = pd.read_excel(\"src/data/INNOVIX_Floresland.xlsx\", sheet_name=\"Ex-Factory volumes\")\n",
    "df_floresland_demand_volumes = pd.read_excel(\"src/data/INNOVIX_Floresland.xlsx\", sheet_name=\"Demand volumes\")\n",
    "df_floresland_activity = pd.read_excel(\"src/data/INNOVIX_Floresland.xlsx\", sheet_name=\"Activity\")\n",
    "df_floresland_share_of_voice = pd.read_excel(\"src/data/INNOVIX_Floresland.xlsx\", sheet_name=\"Share of Voice\")\n",
    "df_floresland_new_patient_share = pd.read_excel(\"src/data/INNOVIX_Floresland.xlsx\", sheet_name=\"New patient share\")\n",
    "df_floresland_new_patient_share = fix_date_missing_period(df_floresland_new_patient_share)\n",
    "df_floresland_indication_split = pd.read_excel(\"src/data/INNOVIX_Floresland.xlsx\", sheet_name=\"Indication split\")\n",
    "\n",
    "\n",
    "\n",
    "NUM_IND = max(\n",
    "    max([int(x.split(\" \")[1]) for x in df_floresland_activity[\"Indication\"].unique()]),\n",
    "    max([int(x.split(\" \")[1]) for x in df_floresland_share_of_voice[\"Indication\"].unique()]),\n",
    "    max([int(x.split(\" \")[1]) for x in df_floresland_new_patient_share[\"Indication\"].unique()]),\n",
    "    max([int(x.split(\" \")[1]) for x in df_floresland_indication_split[\"Indication\"].unique()])\n",
    ")\n",
    "PRODUCTS = [\"INNOVIX\", \"YREX\"]\n",
    "\n",
    "\n",
    "df = merge_dfs([\n",
    "    clean_innovix_ex_factory_vol(df_floresland_ex_factory_vol),\n",
    "    clean_innovix_demand_volumes(df_floresland_demand_volumes, products=PRODUCTS),\n",
    "    clean_innovix_activity(df_floresland_activity, NUM_IND, products=PRODUCTS),\n",
    "    clean_innovix_share_of_voice(df_floresland_share_of_voice, NUM_IND, products=PRODUCTS),\n",
    "    clean_innovix_new_patient_share(df_floresland_new_patient_share, NUM_IND, products=PRODUCTS),\n",
    "    clean_innovix_indication_split(df_floresland_indication_split, NUM_IND, products=PRODUCTS)\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['optimized tslmx', 'optimized arima', 'univariate xgboost',\n",
       "       'multivariate xgboost', 'optimized ets', 'univariate arima',\n",
       "       'optimized tbats'], dtype=object)"
      ]
     },
     "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>Country</th>\n",
       "      <th>Product</th>\n",
       "      <th>Data type</th>\n",
       "      <th>Forecast Algorithm</th>\n",
       "      <th>Data period</th>\n",
       "      <th>Date</th>\n",
       "      <th>Value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>84</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2024-11-01</td>\n",
       "      <td>4.187574e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2024-12-01</td>\n",
       "      <td>4.227316e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>86</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-01-01</td>\n",
       "      <td>5.214774e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>87</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-02-01</td>\n",
       "      <td>4.646591e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-03-01</td>\n",
       "      <td>4.815716e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-04-01</td>\n",
       "      <td>4.344956e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-05-01</td>\n",
       "      <td>4.406519e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>91</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-06-01</td>\n",
       "      <td>4.089633e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-07-01</td>\n",
       "      <td>4.361719e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>93</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-08-01</td>\n",
       "      <td>4.427256e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>94</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-09-01</td>\n",
       "      <td>4.404180e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-10-01</td>\n",
       "      <td>4.386894e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-11-01</td>\n",
       "      <td>4.421644e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>Floresland</td>\n",
       "      <td>INNOVIX</td>\n",
       "      <td>Forecast volumes</td>\n",
       "      <td>optimized arima</td>\n",
       "      <td>Monthly</td>\n",
       "      <td>2025-12-01</td>\n",
       "      <td>4.075436e+06</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Country  Product         Data type Forecast Algorithm Data period  \\\n",
       "84  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "85  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "86  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "87  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "88  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "89  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "90  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "91  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "92  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "93  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "94  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "95  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "96  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "97  Floresland  INNOVIX  Forecast volumes    optimized arima     Monthly   \n",
       "\n",
       "         Date         Value  \n",
       "84 2024-11-01  4.187574e+06  \n",
       "85 2024-12-01  4.227316e+06  \n",
       "86 2025-01-01  5.214774e+06  \n",
       "87 2025-02-01  4.646591e+06  \n",
       "88 2025-03-01  4.815716e+06  \n",
       "89 2025-04-01  4.344956e+06  \n",
       "90 2025-05-01  4.406519e+06  \n",
       "91 2025-06-01  4.089633e+06  \n",
       "92 2025-07-01  4.361719e+06  \n",
       "93 2025-08-01  4.427256e+06  \n",
       "94 2025-09-01  4.404180e+06  \n",
       "95 2025-10-01  4.386894e+06  \n",
       "96 2025-11-01  4.421644e+06  \n",
       "97 2025-12-01  4.075436e+06  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_truth = pd.read_excel(\"src/data/INNOVIX_Floresland.xlsx\", sheet_name=\"Forecast volumes\")\n",
    "display(df_truth[\"Forecast Algorithm\"].unique())\n",
    "\n",
    "algo = \"optimized arima\"\n",
    "\n",
    "df_truth = df_truth[ df_truth[\"Forecast Algorithm\"] == algo ]\n",
    "df_truth = df_truth[ df_truth[\"Data period\"] == \"Monthly\" ]\n",
    "\n",
    "display(df_truth)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.drop(columns=[\"Country\", \"Product\"], inplace=True)\n",
    "df = df.sort_values(by=\"Date\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "df[\"Date\"] = np.arange(0, len(df[\"Date\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Date</th>\n",
       "      <th>ex_factory_volumes</th>\n",
       "      <th>INNOVIX_months_of_treatment</th>\n",
       "      <th>YREX_months_of_treatment</th>\n",
       "      <th>INNOVIX_face-to-face-call_ind1</th>\n",
       "      <th>INNOVIX_face-to-face-call_ind2</th>\n",
       "      <th>INNOVIX_face-to-face-call_ind3</th>\n",
       "      <th>INNOVIX_face-to-face-call_ind4</th>\n",
       "      <th>INNOVIX_face-to-face-call_ind5</th>\n",
       "      <th>INNOVIX_face-to-face-call_ind6</th>\n",
       "      <th>...</th>\n",
       "      <th>YREX_indication_split_ind14</th>\n",
       "      <th>YREX_indication_split_ind15</th>\n",
       "      <th>YREX_indication_split_ind16</th>\n",
       "      <th>YREX_indication_split_ind17</th>\n",
       "      <th>YREX_indication_split_ind18</th>\n",
       "      <th>YREX_indication_split_ind19</th>\n",
       "      <th>YREX_indication_split_ind20</th>\n",
       "      <th>YREX_indication_split_ind21</th>\n",
       "      <th>YREX_indication_split_ind22</th>\n",
       "      <th>YREX_indication_split_ind23</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>3503180</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>2611020</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2952120</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>2995280</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>2929660</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</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",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77</th>\n",
       "      <td>77</td>\n",
       "      <td>3967620</td>\n",
       "      <td>7707.728375</td>\n",
       "      <td>23359.408837</td>\n",
       "      <td>377.0</td>\n",
       "      <td>168.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.017831</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.571180</td>\n",
       "      <td>0.006816</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.045567</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78</th>\n",
       "      <td>78</td>\n",
       "      <td>4932260</td>\n",
       "      <td>9653.302392</td>\n",
       "      <td>28975.411650</td>\n",
       "      <td>158.0</td>\n",
       "      <td>287.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.013297</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.536082</td>\n",
       "      <td>0.013918</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.049469</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>79</td>\n",
       "      <td>4063420</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>80</td>\n",
       "      <td>4464140</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81</th>\n",
       "      <td>81</td>\n",
       "      <td>4556560</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>82 rows × 326 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    Date  ex_factory_volumes  INNOVIX_months_of_treatment  \\\n",
       "0      0             3503180                          NaN   \n",
       "1      1             2611020                          NaN   \n",
       "2      2             2952120                          NaN   \n",
       "3      3             2995280                          NaN   \n",
       "4      4             2929660                          NaN   \n",
       "..   ...                 ...                          ...   \n",
       "77    77             3967620                  7707.728375   \n",
       "78    78             4932260                  9653.302392   \n",
       "79    79             4063420                          NaN   \n",
       "80    80             4464140                          NaN   \n",
       "81    81             4556560                          NaN   \n",
       "\n",
       "    YREX_months_of_treatment  INNOVIX_face-to-face-call_ind1  \\\n",
       "0                        NaN                             NaN   \n",
       "1                        NaN                             NaN   \n",
       "2                        NaN                             NaN   \n",
       "3                        NaN                             NaN   \n",
       "4                        NaN                             0.0   \n",
       "..                       ...                             ...   \n",
       "77              23359.408837                           377.0   \n",
       "78              28975.411650                           158.0   \n",
       "79                       NaN                             0.0   \n",
       "80                       NaN                             0.0   \n",
       "81                       NaN                             0.0   \n",
       "\n",
       "    INNOVIX_face-to-face-call_ind2  INNOVIX_face-to-face-call_ind3  \\\n",
       "0                              NaN                             NaN   \n",
       "1                              NaN                             NaN   \n",
       "2                              NaN                             NaN   \n",
       "3                              NaN                             NaN   \n",
       "4                              0.0                             0.0   \n",
       "..                             ...                             ...   \n",
       "77                           168.0                             0.0   \n",
       "78                           287.0                             0.0   \n",
       "79                             0.0                             0.0   \n",
       "80                             0.0                             0.0   \n",
       "81                             0.0                             0.0   \n",
       "\n",
       "    INNOVIX_face-to-face-call_ind4  INNOVIX_face-to-face-call_ind5  \\\n",
       "0                              NaN                             NaN   \n",
       "1                              NaN                             NaN   \n",
       "2                              NaN                             NaN   \n",
       "3                              NaN                             NaN   \n",
       "4                              0.0                             0.0   \n",
       "..                             ...                             ...   \n",
       "77                             0.0                             0.0   \n",
       "78                             0.0                             0.0   \n",
       "79                             0.0                             0.0   \n",
       "80                             0.0                             0.0   \n",
       "81                             0.0                             0.0   \n",
       "\n",
       "    INNOVIX_face-to-face-call_ind6  ...  YREX_indication_split_ind14  \\\n",
       "0                              NaN  ...                          NaN   \n",
       "1                              NaN  ...                          NaN   \n",
       "2                              NaN  ...                          NaN   \n",
       "3                              NaN  ...                          NaN   \n",
       "4                              0.0  ...                          NaN   \n",
       "..                             ...  ...                          ...   \n",
       "77                             0.0  ...                     0.017831   \n",
       "78                             0.0  ...                     0.013297   \n",
       "79                             0.0  ...                          NaN   \n",
       "80                             0.0  ...                          NaN   \n",
       "81                             0.0  ...                          NaN   \n",
       "\n",
       "    YREX_indication_split_ind15  YREX_indication_split_ind16  \\\n",
       "0                           NaN                          NaN   \n",
       "1                           NaN                          NaN   \n",
       "2                           NaN                          NaN   \n",
       "3                           NaN                          NaN   \n",
       "4                           NaN                          NaN   \n",
       "..                          ...                          ...   \n",
       "77                          0.0                          0.0   \n",
       "78                          0.0                          0.0   \n",
       "79                          NaN                          NaN   \n",
       "80                          NaN                          NaN   \n",
       "81                          NaN                          NaN   \n",
       "\n",
       "    YREX_indication_split_ind17  YREX_indication_split_ind18  \\\n",
       "0                           NaN                          NaN   \n",
       "1                           NaN                          NaN   \n",
       "2                           NaN                          NaN   \n",
       "3                           NaN                          NaN   \n",
       "4                           NaN                          NaN   \n",
       "..                          ...                          ...   \n",
       "77                          0.0                          0.0   \n",
       "78                          0.0                          0.0   \n",
       "79                          NaN                          NaN   \n",
       "80                          NaN                          NaN   \n",
       "81                          NaN                          NaN   \n",
       "\n",
       "    YREX_indication_split_ind19  YREX_indication_split_ind20  \\\n",
       "0                           NaN                          NaN   \n",
       "1                           NaN                          NaN   \n",
       "2                           NaN                          NaN   \n",
       "3                           NaN                          NaN   \n",
       "4                           NaN                          NaN   \n",
       "..                          ...                          ...   \n",
       "77                     0.571180                     0.006816   \n",
       "78                     0.536082                     0.013918   \n",
       "79                          NaN                          NaN   \n",
       "80                          NaN                          NaN   \n",
       "81                          NaN                          NaN   \n",
       "\n",
       "    YREX_indication_split_ind21  YREX_indication_split_ind22  \\\n",
       "0                           NaN                          NaN   \n",
       "1                           NaN                          NaN   \n",
       "2                           NaN                          NaN   \n",
       "3                           NaN                          NaN   \n",
       "4                           NaN                          NaN   \n",
       "..                          ...                          ...   \n",
       "77                          0.0                          0.0   \n",
       "78                          0.0                          0.0   \n",
       "79                          NaN                          NaN   \n",
       "80                          NaN                          NaN   \n",
       "81                          NaN                          NaN   \n",
       "\n",
       "    YREX_indication_split_ind23  \n",
       "0                           NaN  \n",
       "1                           NaN  \n",
       "2                           NaN  \n",
       "3                           NaN  \n",
       "4                           NaN  \n",
       "..                          ...  \n",
       "77                     0.045567  \n",
       "78                     0.049469  \n",
       "79                          NaN  \n",
       "80                          NaN  \n",
       "81                          NaN  \n",
       "\n",
       "[82 rows x 326 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.gaussian_process.kernels import RBF\n",
    "from sklearn.gaussian_process import GaussianProcessRegressor\n",
    "from sklearn.gaussian_process.kernels import WhiteKernel\n",
    "from sklearn.gaussian_process.kernels import ConstantKernel\n",
    "from sklearn.gaussian_process.kernels import ExpSineSquared\n",
    "from sklearn.gaussian_process.kernels import DotProduct\n",
    "from sklearn.gaussian_process.kernels import Matern\n",
    "\n",
    "\n",
    "# kernel = WhiteKernel(0.1)\n",
    "# kernel += ConstantKernel(1) * RBF(1, (-1e, 1e2))\n",
    "# kernel += ConstantKernel(1) * DotProduct(1)\n",
    "\n",
    "kernel = WhiteKernel(0.1, (1e-9, 1e9))\n",
    "kernel += ConstantKernel(1, (1e-9, 1e9)) * RBF(1, (1e-9, 1e9))\n",
    "kernel += ConstantKernel(1, (1e-9, 1e9)) * DotProduct(1, (1e-9, 1e9))\n",
    "\n",
    "\n",
    "# # dff = df.dropna()\n",
    "# dff = df\n",
    "# X = dff.drop(columns=[\"ex_factory_volumes\"])\n",
    "# y = dff[\"ex_factory_volumes\"]\n",
    "\n",
    "\n",
    "# print(\"ex_factory_volumes\" in X.columns, X.shape)\n",
    "\n",
    "# X = X.fillna(X.mean())\n",
    "# # X = pd.DataFrame(X[\"Date\"])\n",
    "\n",
    "# th = 10\n",
    "\n",
    "# gp = GaussianProcessRegressor(kernel=kernel, n_restarts_optimizer=9)\n",
    "# gp.fit(X.iloc[:-th], y.iloc[:-th])\n",
    "\n",
    "# pred, std = gp.predict(X.iloc[-th:], return_std=True)\n",
    "# print( pred )\n",
    "# print( std*1.96 )\n",
    "# print( y.iloc[-th:].tolist() )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 20/20 [00:11<00:00,  1.76it/s]\n"
     ]
    }
   ],
   "source": [
    "import warnings\n",
    "from tqdm import tqdm\n",
    "\n",
    "\n",
    "\n",
    "# dff = df.dropna()\n",
    "dff = df.set_index([\"Date\"], drop=False)\n",
    "\n",
    "X = dff.drop(columns=[\"ex_factory_volumes\"])\n",
    "y = dff[\"ex_factory_volumes\"]\n",
    "\n",
    "X = X.fillna(X.mean())\n",
    "# X = pd.DataFrame(X[\"Date\"])\n",
    "\n",
    "# th = 10\n",
    "X = X # Assigns X to X\n",
    "y = y # Assigns y to y\n",
    "\n",
    "X = (X-X.mean())/(X.std() + 1e-9)\n",
    "\n",
    "y_mean = y.mean()\n",
    "y_std = y.std()\n",
    "y = (y - y.mean()) / (y.std())\n",
    "\n",
    "\n",
    "\n",
    "with warnings.catch_warnings(action=\"ignore\"):\n",
    "\n",
    "    gp = GaussianProcessRegressor(kernel=kernel, n_restarts_optimizer=2)\n",
    "    # gp.fit(X, y)\n",
    "    stds = []\n",
    "\n",
    "    # Test scenario\n",
    "    # new_sample = {\n",
    "    #     col: np.random.normal(X[col].mean(), X[col].std(), (1,))[0]\n",
    "    #     for col in X.columns\n",
    "    # }\n",
    "    # new_sample[\"INNOVIX_months_of_treatment\"] *= (1 - 0.1)\n",
    "    # new_sample[\"Date\"] = 81 + 12\n",
    "\n",
    "    # sampled_in = pd.DataFrame.from_dict([new_sample])\n",
    "\n",
    "    # pred, std = gp.predict(sampled_in, return_std=True)\n",
    "    # pred = pred[0]\n",
    "\n",
    "    # stds.append(std)\n",
    "\n",
    "    start_time = int(X.iloc[-1][\"Date\"]) + 1\n",
    "\n",
    "\n",
    "    \n",
    "    # X = pd.concat([ X, pd.DataFrame.from_dict([new_sample]) ])\n",
    "    # y = pd.concat([ y, pd.Series([pred]) ])\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "    for _ in tqdm(range(20)):\n",
    "        # if start_time == 81 + 12:\n",
    "        #     start_time += 1\n",
    "        #     continue\n",
    "            \n",
    "        gp.fit(X, y)\n",
    "\n",
    "        new_sample = {\n",
    "            col: np.random.normal(X[col].mean(), X[col].std(), (1,))[0]\n",
    "            for col in X.columns\n",
    "        }\n",
    "        new_sample[\"Date\"] = start_time\n",
    "\n",
    "        sampled_in = pd.DataFrame.from_dict([new_sample])\n",
    "\n",
    "        pred, std = gp.predict(sampled_in, return_std=True)\n",
    "        pred = pred[0]\n",
    "\n",
    "        stds.append(std)\n",
    "\n",
    "        X = pd.concat([ X, pd.DataFrame.from_dict([new_sample]) ])\n",
    "        y = pd.concat([ y, pd.Series([pred]) ])\n",
    "\n",
    "        start_time += 1\n",
    "\n",
    "\n",
    "        # print((pd.DataFrame.from_dict([new_sample]).columns))\n",
    "        # print((dff.columns))\n",
    "\n",
    "        # dff.loc[len(dff)] = pd.DataFrame.from_dict([new_sample])\n",
    "\n",
    "\n",
    "        # dff.append(pd.DataFrame.from_dict([new_sample]))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1600x800 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_denorm = (y * y_std) + y_mean\n",
    "\n",
    "plot_gp(\n",
    "    pred = pd.Series(index=range(len(y_denorm[-len(stds):])), data=y_denorm[-len(stds):].values), \n",
    "    samples = pd.Series(index=range(len(y_denorm[-len(stds):])), data=y_denorm[-len(stds):].values), \n",
    "    std = np.array([x[0] for x in stds]) * y_std,\n",
    "    figsize = (16, 8)\n",
    ")\n",
    "\n",
    "# plt.ylim((-1e7, 1e7))\n",
    "# plt.plot(range(len(df_truth[\"Value\"])), df_truth[\"Value\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_gp(target=None, pred=None, std=None, samples=None,\n",
    "        target_samples=None, figsize=None):\n",
    "    plt.figure(figsize=figsize)\n",
    "    if target is not None:\n",
    "        plt.plot(target.index, target, c='black', label='target')\n",
    "    if pred is not None:\n",
    "        plt.scatter(pred.index, pred, c='tab:blue',\n",
    "                label='predictions')\n",
    "    if std is not None:\n",
    "        plt.fill_between(pred.index, pred-1.96*std, pred+1.96*std,\n",
    "                alpha=.3, fc='tab:blue', ec='None',\n",
    "                label='95% C.I.')\n",
    "    # Add scatter plots\n",
    "    if samples is not None:\n",
    "        try:\n",
    "            x = samples.index\n",
    "            y = samples.values\n",
    "        except AttributeError:\n",
    "            x = samples[0]\n",
    "            y = samples[1]\n",
    "        plt.scatter(x, y, color='tab:orange',\n",
    "              label='samples', marker='x')\n",
    "    if target_samples is not None:\n",
    "        try:\n",
    "            x = target_samples.index\n",
    "            y = target_samples.values\n",
    "        except AttributeError:\n",
    "            x = target_samples[0]\n",
    "            y = target_samples[1]\n",
    "        plt.scatter(x, y,\n",
    "                color='black', label='target', s=5)\n",
    "    plt.legend()\n",
    "    plt.grid(':')\n",
    "    plt.tight_layout()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "ml",
   "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": "3.12.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}