TNR Uplift Framework.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "c2327fb6",
   "metadata": {},
   "source": [
    "### Uplift Modeling for Treatment Non-Responders (TNRs)\n",
    "#### Experimental Design Framework (n=50,000)\n",
    "#### Stage 1: Treatment (One-time Email) & Cost (USD 0.2/treatment cycle)\n",
    "(i) Control Group: 9,755.\n",
    "(ii) Treatment Group: 40,245 (TR: 7,973; TNR: 32,272).\n",
    "(iii) Total: 50,000.\n",
    "#### Stage 2: Treatment (Multifaceted: Email, SMS, and Coupon) & Cost (USD 1.0/treatment cycle)\n",
    "##### Only select TNRs from Stage 1.\n",
    "(i) Control Group: 8,052.\n",
    "(ii) Treatment Group: 32,034 (TR: 4,783; TNR: 27,251).\n",
    "(iii) Total: 40,086.\n",
    "#### Stage 3: Treatment (Multifaceted and Repeated) & Cost (USD 1.5/treatment cycle)\n",
    "##### Only select TNRs from Stage 2.\n",
    "(i) Control Group: 6,961.\n",
    "(ii) Treatment Group: 27,109 (TR: 2,715; TNR: 24,394).\n",
    "(iii) Total: 34,070\n",
    "#### Experimental Design Example: https://bmcresnotes.biomedcentral.com/articles/10.1186/s13104-021-05516-9?fbclid=IwAR0xHmDeQzc4HHGMDNP7n6fADmV7Fut6GYv1fQkU2SmH39Ka3IsyQuGRIvI\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "db1effcc",
   "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>gender</th>\n",
       "      <th>education</th>\n",
       "      <th>round_1</th>\n",
       "      <th>treat_1</th>\n",
       "      <th>convert_1</th>\n",
       "      <th>expcost_1</th>\n",
       "      <th>cdays_1</th>\n",
       "      <th>gain_pc_1</th>\n",
       "      <th>round_2</th>\n",
       "      <th>treat_2</th>\n",
       "      <th>convert_2</th>\n",
       "      <th>expcost_2</th>\n",
       "      <th>cdays_2</th>\n",
       "      <th>gain_pc_2</th>\n",
       "      <th>round_3</th>\n",
       "      <th>treat_3</th>\n",
       "      <th>convert_3</th>\n",
       "      <th>expcost_3</th>\n",
       "      <th>cdays_3</th>\n",
       "      <th>gain_pc_3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.2</td>\n",
       "      <td>9.0</td>\n",
       "      <td>4.9</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</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>1</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>13</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.5</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49995</th>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49996</th>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0.2</td>\n",
       "      <td>13.0</td>\n",
       "      <td>4.9</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</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>49997</th>\n",
       "      <td>0</td>\n",
       "      <td>14</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49998</th>\n",
       "      <td>0</td>\n",
       "      <td>9</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0.2</td>\n",
       "      <td>23.0</td>\n",
       "      <td>4.9</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</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>49999</th>\n",
       "      <td>0</td>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>50000 rows × 20 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       gender  education  round_1  treat_1  convert_1  expcost_1  cdays_1  \\\n",
       "0           0         12        1        1          0        0.2      NaN   \n",
       "1           1         11        1        0          1        0.2      9.0   \n",
       "2           1         12        1        1          0        0.2      NaN   \n",
       "3           1         10        1        1          0        0.2      NaN   \n",
       "4           0         13        1        1          0        0.2      NaN   \n",
       "...       ...        ...      ...      ...        ...        ...      ...   \n",
       "49995       1         10        1        1          0        0.2      NaN   \n",
       "49996       1          8        1        1          1        0.2     13.0   \n",
       "49997       0         14        1        1          0        0.2      NaN   \n",
       "49998       0          9        1        1          1        0.2     23.0   \n",
       "49999       0         12        1        1          0        0.2      NaN   \n",
       "\n",
       "       gain_pc_1  round_2  treat_2  convert_2  expcost_2  cdays_2  gain_pc_2  \\\n",
       "0            NaN        2      0.0        0.0        1.0      NaN        NaN   \n",
       "1            4.9        2      NaN        NaN        NaN      NaN        NaN   \n",
       "2            NaN        2      1.0        0.0        1.0      NaN        NaN   \n",
       "3            NaN        2      1.0        0.0        1.0      NaN        NaN   \n",
       "4            NaN        2      1.0        0.0        1.0      NaN        NaN   \n",
       "...          ...      ...      ...        ...        ...      ...        ...   \n",
       "49995        NaN        2      1.0        0.0        1.0      NaN        NaN   \n",
       "49996        4.9        2      NaN        NaN        NaN      NaN        NaN   \n",
       "49997        NaN        2      0.0        0.0        1.0      NaN        NaN   \n",
       "49998        4.9        2      NaN        NaN        NaN      NaN        NaN   \n",
       "49999        NaN        2      1.0        0.0        1.0      NaN        NaN   \n",
       "\n",
       "       round_3  treat_3  convert_3  expcost_3  cdays_3  gain_pc_3  \n",
       "0            3      0.0        0.0        1.5      NaN        NaN  \n",
       "1            3      NaN        NaN        NaN      NaN        NaN  \n",
       "2            3      1.0        0.0        1.5      NaN        NaN  \n",
       "3            3      1.0        0.0        1.5      NaN        NaN  \n",
       "4            3      1.0        0.0        1.5      NaN        NaN  \n",
       "...        ...      ...        ...        ...      ...        ...  \n",
       "49995        3      0.0        0.0        1.5      NaN        NaN  \n",
       "49996        3      NaN        NaN        NaN      NaN        NaN  \n",
       "49997        3      1.0        0.0        1.5      NaN        NaN  \n",
       "49998        3      NaN        NaN        NaN      NaN        NaN  \n",
       "49999        3      1.0        0.0        1.5      NaN        NaN  \n",
       "\n",
       "[50000 rows x 20 columns]"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Required Packages and Data\n",
    "import pandas as pd \n",
    "import numpy as np\n",
    "\n",
    "df = pd.read_csv('T:\\\\Research Projects_MGA\\\\GitHub\\\\uplift.csv') \n",
    "df = df.loc[:, df.columns != 'id'] #id column is not required\n",
    "df\n",
    "#df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "418e5ee2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "       gender  education  treat_1  convert_1  expcost_1  cdays_1  gain_pc_1\n",
      "0           0         12        1          0        0.2      NaN        NaN\n",
      "1           1         11        0          1        0.2      9.0        4.9\n",
      "2           1         12        1          0        0.2      NaN        NaN\n",
      "3           1         10        1          0        0.2      NaN        NaN\n",
      "4           0         13        1          0        0.2      NaN        NaN\n",
      "...       ...        ...      ...        ...        ...      ...        ...\n",
      "49995       1         10        1          0        0.2      NaN        NaN\n",
      "49996       1          8        1          1        0.2     13.0        4.9\n",
      "49997       0         14        1          0        0.2      NaN        NaN\n",
      "49998       0          9        1          1        0.2     23.0        4.9\n",
      "49999       0         12        1          0        0.2      NaN        NaN\n",
      "\n",
      "[50000 rows x 7 columns]\n",
      "       gender  education  treat_2  convert_2  expcost_2  cdays_2  gain_pc_2\n",
      "0           0         12      0.0        0.0        1.0      NaN        NaN\n",
      "1           1         11      NaN        NaN        NaN      NaN        NaN\n",
      "2           1         12      1.0        0.0        1.0      NaN        NaN\n",
      "3           1         10      1.0        0.0        1.0      NaN        NaN\n",
      "4           0         13      1.0        0.0        1.0      NaN        NaN\n",
      "...       ...        ...      ...        ...        ...      ...        ...\n",
      "49995       1         10      1.0        0.0        1.0      NaN        NaN\n",
      "49996       1          8      NaN        NaN        NaN      NaN        NaN\n",
      "49997       0         14      0.0        0.0        1.0      NaN        NaN\n",
      "49998       0          9      NaN        NaN        NaN      NaN        NaN\n",
      "49999       0         12      1.0        0.0        1.0      NaN        NaN\n",
      "\n",
      "[50000 rows x 7 columns]\n",
      "       gender  education  treat_3  convert_3  expcost_3  cdays_3  gain_pc_3\n",
      "0           0         12      0.0        0.0        1.5      NaN        NaN\n",
      "1           1         11      NaN        NaN        NaN      NaN        NaN\n",
      "2           1         12      1.0        0.0        1.5      NaN        NaN\n",
      "3           1         10      1.0        0.0        1.5      NaN        NaN\n",
      "4           0         13      1.0        0.0        1.5      NaN        NaN\n",
      "...       ...        ...      ...        ...        ...      ...        ...\n",
      "49995       1         10      0.0        0.0        1.5      NaN        NaN\n",
      "49996       1          8      NaN        NaN        NaN      NaN        NaN\n",
      "49997       0         14      1.0        0.0        1.5      NaN        NaN\n",
      "49998       0          9      NaN        NaN        NaN      NaN        NaN\n",
      "49999       0         12      1.0        0.0        1.5      NaN        NaN\n",
      "\n",
      "[50000 rows x 7 columns]\n"
     ]
    }
   ],
   "source": [
    "#New Dataframe\n",
    "df1 = df.iloc[:,np.r_[0:2, 3:8]]\n",
    "print(df1)\n",
    "\n",
    "df2 = df.iloc[:,np.r_[0:2, 9:14]]\n",
    "print(df2)\n",
    "\n",
    "df3 = df.iloc[:,np.r_[0:2, 15:20]]\n",
    "print(df3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "ca7299d9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "convert_1      0     1    All\n",
      "treat_1                      \n",
      "0           7814  1941   9755\n",
      "1          32272  7973  40245\n",
      "All        40086  9914  50000\n",
      "convert_1        0        1     All\n",
      "treat_1                            \n",
      "0          0.15628  0.03882  0.1951\n",
      "1          0.64544  0.15946  0.8049\n",
      "All        0.80172  0.19828  1.0000\n",
      "convert_2    0.0   1.0    All\n",
      "treat_2                      \n",
      "0.0         6819  1233   8052\n",
      "1.0        27251  4783  32034\n",
      "All        34070  6016  40086\n",
      "convert_2       0.0       1.0       All\n",
      "treat_2                                \n",
      "0.0        0.170109  0.030759  0.200868\n",
      "1.0        0.679813  0.119318  0.799132\n",
      "All        0.849923  0.150077  1.000000\n",
      "convert_3    0.0   1.0    All\n",
      "treat_3                      \n",
      "0.0         6283   678   6961\n",
      "1.0        24394  2715  27109\n",
      "All        30677  3393  34070\n",
      "convert_3       0.0       1.0       All\n",
      "treat_3                                \n",
      "0.0        0.184414  0.019900  0.204315\n",
      "1.0        0.715996  0.079689  0.795685\n",
      "All        0.900411  0.099589  1.000000\n"
     ]
    }
   ],
   "source": [
    "#Conversion Rate (TR & TNR)\n",
    "print(pd.crosstab(df1['treat_1'], df['convert_1'], margins = True))\n",
    "print(pd.crosstab(df1['treat_1'], df['convert_1'], margins = True, normalize=True))\n",
    "\n",
    "print(pd.crosstab(df2['treat_2'], df['convert_2'], margins = True))\n",
    "print(pd.crosstab(df2['treat_2'], df['convert_2'], margins = True, normalize=True))\n",
    "\n",
    "print(pd.crosstab(df3['treat_3'], df['convert_3'], margins = True))\n",
    "print(pd.crosstab(df3['treat_3'], df['convert_3'], margins = True, normalize=True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "982d5751",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "        convert_1       \n",
      "             mean  count\n",
      "treat_1                 \n",
      "0        0.198975   9755\n",
      "1        0.198112  40245\n",
      "ATE of Stage 1 = -0.0008633180202831103\n",
      "        convert_2       \n",
      "             mean  count\n",
      "treat_2                 \n",
      "0.0       0.15313   8052\n",
      "1.0       0.14931  32034\n",
      "ATE of Stage 2 = -0.0038195492177787727\n",
      "        convert_3       \n",
      "             mean  count\n",
      "treat_3                 \n",
      "0.0      0.097400   6961\n",
      "1.0      0.100151  27109\n",
      "ATE of Stage 3 = 0.002751442405710708\n"
     ]
    }
   ],
   "source": [
    "#Average Treatment Effect (ATE)\n",
    "conver_matrix1 = df1.groupby('treat_1').agg({'convert_1':['mean', 'count']})\n",
    "print(conver_matrix1)\n",
    "print(\"ATE of Stage 1 =\",(conver_matrix1.loc[1,('convert_1', 'mean')] - conver_matrix1.loc[0,('convert_1', 'mean')]))\n",
    "\n",
    "conver_matrix2 = df2.groupby('treat_2').agg({'convert_2':['mean', 'count']})\n",
    "print(conver_matrix2)\n",
    "print(\"ATE of Stage 2 =\",(conver_matrix2.loc[1,('convert_2', 'mean')] - conver_matrix2.loc[0,('convert_2', 'mean')]))\n",
    "\n",
    "conver_matrix3 = df3.groupby('treat_3').agg({'convert_3':['mean', 'count']})\n",
    "print(conver_matrix3)\n",
    "print(\"ATE of Stage 3 =\",(conver_matrix3.loc[1,('convert_3', 'mean')] - conver_matrix3.loc[0,('convert_3', 'mean')]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "aaee74b7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "            cdays_1\n",
      "               mean\n",
      "convert_1          \n",
      "0               NaN\n",
      "1          13.05114\n",
      "            cdays_2\n",
      "               mean\n",
      "convert_2          \n",
      "0.0             NaN\n",
      "1.0        7.933843\n",
      "            cdays_3\n",
      "               mean\n",
      "convert_3          \n",
      "0.0             NaN\n",
      "1.0        6.501916\n"
     ]
    }
   ],
   "source": [
    "#Convertion Days (CD)\n",
    "avg_cd1 = df1.groupby('convert_1').agg({'cdays_1':['mean']})\n",
    "print(avg_cd1)\n",
    "\n",
    "avg_cd2 = df2.groupby('convert_2').agg({'cdays_2':['mean']})\n",
    "print(avg_cd2)\n",
    "\n",
    "avg_cd3 = df3.groupby('convert_3').agg({'cdays_3':['mean']})\n",
    "print(avg_cd3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "70c269ea",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total Cost of Stage 1 = 8049.0 USD\n",
      "Total Cost of Stage 2 = 32034.0 USD\n",
      "Total Cost of Stage 3 = 40663.5 USD\n"
     ]
    }
   ],
   "source": [
    "#Cost of Marketing Campaign\n",
    "total_cost_r1 = 40245*0.2   #stage 1\n",
    "print('Total Cost of Stage 1 =',(total_cost_r1),\"USD\")\n",
    "\n",
    "total_cost_r2 = 32034*1.0   #stage 2\n",
    "print('Total Cost of Stage 2 =',(total_cost_r2),\"USD\")\n",
    "\n",
    "total_cost_r3 = 27109*1.5   #stage 3\n",
    "print('Total Cost of Stage 3 =',(total_cost_r3),\"USD\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "76888266",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total Revenue from Stage 1 = 150000.0 USD\n",
      "Additional Revenue from Stage 2 = 120258.0 USD\n",
      "Additional Revenue from Stage 3 = 102210.0 USD\n",
      "Expected Earning from TNRs = 149770.5 USD\n"
     ]
    }
   ],
   "source": [
    "#Expected Revenues from Marketing Campaign (Expected Revenue: USD 3/person, if converted to TR)\n",
    "print(\"Total Revenue from Stage 1 =\",(df1.treat_1.count())*3.0,\"USD\")\n",
    "print(\"Additional Revenue from Stage 2 =\",(df2.treat_2.count())*3.0,\"USD\")\n",
    "print(\"Additional Revenue from Stage 3 =\",(df3.treat_3.count())*3.0,\"USD\")\n",
    "\n",
    "#Expected Earning from TNRs for Two Additional Treatment Rounds/Stages\n",
    "print(\"Expected Earning from TNRs =\",((df2.treat_2.count()*3)+(df3.treat_3.count()*3)-(total_cost_r2)-(total_cost_r3)),\"USD\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "63a0a3ef",
   "metadata": {},
   "source": [
    "Comment: Stages 2 and 3 can generate additional revenue as the expected earnings from TNRs of Stage 1 are positive (149770.5 USD). As a result, repeated multifaceted market campaigns can be as good as alternative strategies."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "70d569b2",
   "metadata": {},
   "source": [
    "## Logistic Regression\n",
    "Odds ratios (ORs) are calculated using a logistic regression model. In the model, convert_1 represents the target/outcome variable (y), while treat_1, gender, and education represent feature/independent variables (Xs)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "51eb0a1e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.497959\n",
      "         Iterations 5\n",
      "                           Logit Regression Results                           \n",
      "==============================================================================\n",
      "Dep. Variable:              convert_1   No. Observations:                50000\n",
      "Model:                          Logit   Df Residuals:                    49996\n",
      "Method:                           MLE   Df Model:                            3\n",
      "Date:                Tue, 13 Dec 2022   Pseudo R-squ.:               9.965e-05\n",
      "Time:                        14:06:37   Log-Likelihood:                -24898.\n",
      "converged:                       True   LL-Null:                       -24900.\n",
      "Covariance Type:            nonrobust   LLR p-value:                    0.1746\n",
      "==============================================================================\n",
      "                 coef    std err          z      P>|z|      [0.025      0.975]\n",
      "------------------------------------------------------------------------------\n",
      "Intercept     -1.3752      0.068    -20.271      0.000      -1.508      -1.242\n",
      "treat_1       -0.0053      0.028     -0.188      0.851      -0.061       0.050\n",
      "gender         0.0474      0.022      2.114      0.035       0.003       0.091\n",
      "education     -0.0038      0.006     -0.672      0.501      -0.015       0.007\n",
      "==============================================================================\n",
      "Odds Ratios from Model 1:\n",
      "Intercept    0.252782\n",
      "treat_1      0.994699\n",
      "gender       1.048571\n",
      "education    0.996234\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "#Required Packages and Data\n",
    "import statsmodels.api as sm\n",
    "import statsmodels.formula.api as smf\n",
    "\n",
    "model1 = smf.logit(formula='convert_1 ~ treat_1 + gender + education', data=df1).fit()\n",
    "print(model1.summary())\n",
    "\n",
    "print('Odds Ratios from Model 1:')\n",
    "print(np.exp(model1.params))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d0a5625a",
   "metadata": {},
   "source": [
    "Comment: OR>1 (OR<1) indicates Positive (Negative) association between feature and target variable. The p-value is also another model evaluation criteria; however, it depends upon the business needs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "a7d4a98f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.422816\n",
      "         Iterations 6\n",
      "                           Logit Regression Results                           \n",
      "==============================================================================\n",
      "Dep. Variable:              convert_2   No. Observations:                40086\n",
      "Model:                          Logit   Df Residuals:                    40082\n",
      "Method:                           MLE   Df Model:                            3\n",
      "Date:                Tue, 13 Dec 2022   Pseudo R-squ.:               6.444e-05\n",
      "Time:                        14:06:38   Log-Likelihood:                -16949.\n",
      "converged:                       True   LL-Null:                       -16950.\n",
      "Covariance Type:            nonrobust   LLR p-value:                    0.5350\n",
      "==============================================================================\n",
      "                 coef    std err          z      P>|z|      [0.025      0.975]\n",
      "------------------------------------------------------------------------------\n",
      "Intercept     -1.7365      0.084    -20.603      0.000      -1.902      -1.571\n",
      "treat_2       -0.0297      0.035     -0.857      0.392      -0.098       0.038\n",
      "gender        -0.0302      0.028     -1.080      0.280      -0.085       0.025\n",
      "education      0.0037      0.007      0.535      0.592      -0.010       0.017\n",
      "==============================================================================\n",
      "Odds Ratios from Model 2:\n",
      "Intercept    0.176133\n",
      "treat_2      0.970710\n",
      "gender       0.970245\n",
      "education    1.003749\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "model2 = smf.logit(formula='convert_2 ~ treat_2 + gender + education', data=df2).fit()\n",
    "print(model2.summary())\n",
    "\n",
    "print('Odds Ratios from Model 2:')\n",
    "print(np.exp(model2.params))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "95b3353c",
   "metadata": {},
   "source": [
    "Comment: OR>1 (OR<1) indicates Positive (Negative) association between feature and target variable. The p-value is also another model evaluation criteria; however, it depends upon the business needs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "18793266",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.324119\n",
      "         Iterations 6\n",
      "                           Logit Regression Results                           \n",
      "==============================================================================\n",
      "Dep. Variable:              convert_3   No. Observations:                34070\n",
      "Model:                          Logit   Df Residuals:                    34066\n",
      "Method:                           MLE   Df Model:                            3\n",
      "Date:                Tue, 13 Dec 2022   Pseudo R-squ.:               0.0001868\n",
      "Time:                        14:06:40   Log-Likelihood:                -11043.\n",
      "converged:                       True   LL-Null:                       -11045.\n",
      "Covariance Type:            nonrobust   LLR p-value:                    0.2482\n",
      "==============================================================================\n",
      "                 coef    std err          z      P>|z|      [0.025      0.975]\n",
      "------------------------------------------------------------------------------\n",
      "Intercept     -2.1217      0.109    -19.498      0.000      -2.335      -1.908\n",
      "treat_3        0.0310      0.045      0.687      0.492      -0.058       0.120\n",
      "gender        -0.0639      0.036     -1.764      0.078      -0.135       0.007\n",
      "education     -0.0067      0.009     -0.737      0.461      -0.024       0.011\n",
      "==============================================================================\n",
      "Odds Ratios from Model 3:\n",
      "Intercept    0.119824\n",
      "treat_3      1.031525\n",
      "gender       0.938141\n",
      "education    0.993357\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "model3 = smf.logit(formula='convert_3 ~ treat_3 + gender + education', data=df3).fit()\n",
    "print(model3.summary())\n",
    "\n",
    "print('Odds Ratios from Model 3:')\n",
    "print(np.exp(model3.params))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "becc0225",
   "metadata": {},
   "source": [
    "Comment: OR>1 (OR<1) indicates Positive (Negative) association between feature and target variable. The p-value is also another model evaluation criteria; however, it depends upon the business needs."
   ]
  }
 ],
 "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": "3.10.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}