Initial commit

src/test.ipynb

@@ -2,7 +2,7 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 9,
+   "execution_count": 1,
    "metadata": {
     "collapsed": true,
     "pycharm": {
@@ -25,59 +25,20 @@
   {
    "cell_type": "code",
    "execution_count": 2,
-   "outputs": [],
-   "source": [
-    "def get_data() -> tuple[Any, Any, Any]:\n",
-    "\t# получаем датасет\n",
-    "\tdataset = fetch_hillstrom(target_col='visit')\n",
-    "\tdataset, target, treatment = dataset['data'], dataset['target'], dataset['treatment']\n",
-    "\t# выбираем два сегмента\n",
-    "\tdataset = dataset[treatment != 'Mens E-Mail']\n",
-    "\ttarget = target[treatment != 'Mens E-Mail']\n",
-    "\ttreatment = treatment[treatment != 'Mens E-Mail'].map({\n",
-    "\t\t'Womens E-Mail': 1,\n",
-    "\t\t'No E-Mail':     0\n",
-    "\t})\n",
-    "\n",
-    "\treturn dataset, target, treatment"
-   ],
-   "metadata": {
-    "collapsed": false,
-    "pycharm": {
-     "name": "#%%\n"
-    }
-   }
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 3,
-   "outputs": [],
-   "source": [
-    "data, target, treatment = get_data()"
-   ],
-   "metadata": {
-    "collapsed": false,
-    "pycharm": {
-     "name": "#%%\n"
-    }
-   }
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 4,
    "outputs": [
     {
      "data": {
-      "text/plain": "   recency history_segment  history  mens  womens   zip_code  newbie channel\n0       10  2) $100 - $200   142.44     1       0  Surburban       0   Phone\n1        6  3) $200 - $350   329.08     1       1      Rural       1     Web\n2        7  2) $100 - $200   180.65     0       1  Surburban       1     Web\n4        2    1) $0 - $100    45.34     1       0      Urban       0     Web\n5        6  2) $100 - $200   134.83     0       1  Surburban       0   Phone",
-      "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>recency</th>\n      <th>history_segment</th>\n      <th>history</th>\n      <th>mens</th>\n      <th>womens</th>\n      <th>zip_code</th>\n      <th>newbie</th>\n      <th>channel</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>10</td>\n      <td>2) $100 - $200</td>\n      <td>142.44</td>\n      <td>1</td>\n      <td>0</td>\n      <td>Surburban</td>\n      <td>0</td>\n      <td>Phone</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>6</td>\n      <td>3) $200 - $350</td>\n      <td>329.08</td>\n      <td>1</td>\n      <td>1</td>\n      <td>Rural</td>\n      <td>1</td>\n      <td>Web</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>7</td>\n      <td>2) $100 - $200</td>\n      <td>180.65</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Surburban</td>\n      <td>1</td>\n      <td>Web</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>2</td>\n      <td>1) $0 - $100</td>\n      <td>45.34</td>\n      <td>1</td>\n      <td>0</td>\n      <td>Urban</td>\n      <td>0</td>\n      <td>Web</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>6</td>\n      <td>2) $100 - $200</td>\n      <td>134.83</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Surburban</td>\n      <td>0</td>\n      <td>Phone</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
+      "text/plain": "{'new_filter': 'filter'}"
      },
-     "execution_count": 4,
+     "execution_count": 2,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "data.head()"
+    "filters = {}\n",
+    "filters['new_filter'] = 'filter'\n",
+    "filters"
    ],
    "metadata": {
     "collapsed": false,
@@ -88,19 +49,22 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
-   "outputs": [
-    {
-     "data": {
-      "text/plain": "0    0\n1    0\n2    0\n4    0\n5    1\nName: visit, dtype: int64"
-     },
-     "execution_count": 5,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "execution_count": 4,
+   "outputs": [],
    "source": [
-    "target.head()"
+    "def get_data() -> tuple[Any, Any, Any]:\n",
+    "\t# получаем датасет\n",
+    "\tdataset = fetch_hillstrom(target_col='visit')\n",
+    "\tdataset, target, treatment = dataset['data'], dataset['target'], dataset['treatment']\n",
+    "\t# выбираем два сегмента\n",
+    "\tdataset = dataset[treatment != 'Mens E-Mail']\n",
+    "\ttarget = target[treatment != 'Mens E-Mail']\n",
+    "\ttreatment = treatment[treatment != 'Mens E-Mail'].map({\n",
+    "\t\t'Womens E-Mail': 1,\n",
+    "\t\t'No E-Mail':     0\n",
+    "\t})\n",
+    "\n",
+    "\treturn dataset, target, treatment"
    ],
    "metadata": {
     "collapsed": false,
@@ -111,19 +75,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
-   "outputs": [
-    {
-     "data": {
-      "text/plain": "0    1\n1    0\n2    1\n4    1\n5    1\nName: segment, dtype: int64"
-     },
-     "execution_count": 6,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "execution_count": 111,
+   "outputs": [],
    "source": [
-    "treatment.head()"
+    "data, target, treatment = get_data()"
    ],
    "metadata": {
     "collapsed": false,
@@ -134,20 +89,17 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 23,
-   "outputs": [
-    {
-     "data": {
-      "text/plain": "visit           0         1\nsegment                    \n0        0.893833  0.106167\n1        0.848600  0.151400",
-      "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>visit</th>\n      <th>0</th>\n      <th>1</th>\n    </tr>\n    <tr>\n      <th>segment</th>\n      <th></th>\n      <th></th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0.893833</td>\n      <td>0.106167</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>0.848600</td>\n      <td>0.151400</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
-     },
-     "execution_count": 23,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "execution_count": 112,
+   "outputs": [],
    "source": [
-    "pd.crosstab(treatment, target, normalize='index')"
+    "X_train, X_val, y_train, y_val, trmnt_train, trmnt_val = train_test_split(\n",
+    "    data, target, treatment, test_size=0.3, random_state=42\n",
+    ")\n",
+    "\n",
+    "models_results = {\n",
+    "    'approach': [],\n",
+    "    'uplift@30%': []\n",
+    "}"
    ],
    "metadata": {
     "collapsed": false,
@@ -158,19 +110,20 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 50,
+   "execution_count": 113,
    "outputs": [
     {
      "data": {
-      "text/plain": "1    50.261167\n0    49.738833\nName: newbie, dtype: float64"
+      "text/plain": "       recency   history_segment  history  mens  womens   zip_code  newbie  \\\n609          9    3) $200 - $350   212.32     1       0  Surburban       0   \n51952        3  6) $750 - $1,000   849.16     1       1  Surburban       1   \n33629        9      1) $0 - $100    43.22     1       0  Surburban       1   \n22103        8      1) $0 - $100    29.99     0       1  Surburban       0   \n21350        7    2) $100 - $200   164.07     1       0  Surburban       1   \n...        ...               ...      ...   ...     ...        ...     ...   \n9307         1    2) $100 - $200   110.45     0       1      Urban       0   \n16819        1      1) $0 - $100    88.04     0       1  Surburban       0   \n57173        1      1) $0 - $100    72.63     0       1      Rural       0   \n1282         3    4) $350 - $500   366.16     0       1      Rural       0   \n23624        9      1) $0 - $100    36.87     1       0      Urban       1   \n\n            channel  \n609    Multichannel  \n51952  Multichannel  \n33629         Phone  \n22103           Web  \n21350         Phone  \n...             ...  \n9307          Phone  \n16819         Phone  \n57173         Phone  \n1282          Phone  \n23624           Web  \n\n[29885 rows x 8 columns]",
+      "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>recency</th>\n      <th>history_segment</th>\n      <th>history</th>\n      <th>mens</th>\n      <th>womens</th>\n      <th>zip_code</th>\n      <th>newbie</th>\n      <th>channel</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>609</th>\n      <td>9</td>\n      <td>3) $200 - $350</td>\n      <td>212.32</td>\n      <td>1</td>\n      <td>0</td>\n      <td>Surburban</td>\n      <td>0</td>\n      <td>Multichannel</td>\n    </tr>\n    <tr>\n      <th>51952</th>\n      <td>3</td>\n      <td>6) $750 - $1,000</td>\n      <td>849.16</td>\n      <td>1</td>\n      <td>1</td>\n      <td>Surburban</td>\n      <td>1</td>\n      <td>Multichannel</td>\n    </tr>\n    <tr>\n      <th>33629</th>\n      <td>9</td>\n      <td>1) $0 - $100</td>\n      <td>43.22</td>\n      <td>1</td>\n      <td>0</td>\n      <td>Surburban</td>\n      <td>1</td>\n      <td>Phone</td>\n    </tr>\n    <tr>\n      <th>22103</th>\n      <td>8</td>\n      <td>1) $0 - $100</td>\n      <td>29.99</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Surburban</td>\n      <td>0</td>\n      <td>Web</td>\n    </tr>\n    <tr>\n      <th>21350</th>\n      <td>7</td>\n      <td>2) $100 - $200</td>\n      <td>164.07</td>\n      <td>1</td>\n      <td>0</td>\n      <td>Surburban</td>\n      <td>1</td>\n      <td>Phone</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    </tr>\n    <tr>\n      <th>9307</th>\n      <td>1</td>\n      <td>2) $100 - $200</td>\n      <td>110.45</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Urban</td>\n      <td>0</td>\n      <td>Phone</td>\n    </tr>\n    <tr>\n      <th>16819</th>\n      <td>1</td>\n      <td>1) $0 - $100</td>\n      <td>88.04</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Surburban</td>\n      <td>0</td>\n      <td>Phone</td>\n    </tr>\n    <tr>\n      <th>57173</th>\n      <td>1</td>\n      <td>1) $0 - $100</td>\n      <td>72.63</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Rural</td>\n      <td>0</td>\n      <td>Phone</td>\n    </tr>\n    <tr>\n      <th>1282</th>\n      <td>3</td>\n      <td>4) $350 - $500</td>\n      <td>366.16</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Rural</td>\n      <td>0</td>\n      <td>Phone</td>\n    </tr>\n    <tr>\n      <th>23624</th>\n      <td>9</td>\n      <td>1) $0 - $100</td>\n      <td>36.87</td>\n      <td>1</td>\n      <td>0</td>\n      <td>Urban</td>\n      <td>1</td>\n      <td>Web</td>\n    </tr>\n  </tbody>\n</table>\n<p>29885 rows × 8 columns</p>\n</div>"
      },
+     "execution_count": 113,
      "metadata": {},
-     "output_type": "display_data"
+     "output_type": "execute_result"
     }
    ],
    "source": [
-    "res = data.newbie.value_counts(normalize=True) * 100\n",
-    "display(res)"
+    "X_train"
    ],
    "metadata": {
     "collapsed": false,
@@ -181,40 +134,20 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 54,
+   "execution_count": 114,
    "outputs": [
     {
      "data": {
-      "text/plain": "Surburban    19275\nUrban        17098\nRural         6320\nName: zip_code, dtype: int64"
+      "text/plain": "       recency history_segment  history  mens  womens   zip_code  newbie  \\\n7730         3  5) $500 - $750   503.73     0       1      Rural       1   \n17594        5  2) $100 - $200   163.39     1       0      Urban       0   \n14481       11  3) $200 - $350   287.77     1       0      Rural       1   \n20003        3    1) $0 - $100    29.99     0       1  Surburban       1   \n19981        5  2) $100 - $200   131.51     0       1      Urban       1   \n...        ...             ...      ...   ...     ...        ...     ...   \n41828       11  4) $350 - $500   457.09     0       1  Surburban       1   \n49354        2    1) $0 - $100    29.99     0       1      Rural       1   \n12063        9  2) $100 - $200   147.48     1       0      Rural       1   \n3757        10  2) $100 - $200   128.78     1       0  Surburban       1   \n10708        6  2) $100 - $200   138.72     1       0      Urban       0   \n\n            channel  \n7730   Multichannel  \n17594           Web  \n14481           Web  \n20003           Web  \n19981           Web  \n...             ...  \n41828  Multichannel  \n49354         Phone  \n12063         Phone  \n3757          Phone  \n10708           Web  \n\n[12808 rows x 8 columns]",
+      "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>recency</th>\n      <th>history_segment</th>\n      <th>history</th>\n      <th>mens</th>\n      <th>womens</th>\n      <th>zip_code</th>\n      <th>newbie</th>\n      <th>channel</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>7730</th>\n      <td>3</td>\n      <td>5) $500 - $750</td>\n      <td>503.73</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Rural</td>\n      <td>1</td>\n      <td>Multichannel</td>\n    </tr>\n    <tr>\n      <th>17594</th>\n      <td>5</td>\n      <td>2) $100 - $200</td>\n      <td>163.39</td>\n      <td>1</td>\n      <td>0</td>\n      <td>Urban</td>\n      <td>0</td>\n      <td>Web</td>\n    </tr>\n    <tr>\n      <th>14481</th>\n      <td>11</td>\n      <td>3) $200 - $350</td>\n      <td>287.77</td>\n      <td>1</td>\n      <td>0</td>\n      <td>Rural</td>\n      <td>1</td>\n      <td>Web</td>\n    </tr>\n    <tr>\n      <th>20003</th>\n      <td>3</td>\n      <td>1) $0 - $100</td>\n      <td>29.99</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Surburban</td>\n      <td>1</td>\n      <td>Web</td>\n    </tr>\n    <tr>\n      <th>19981</th>\n      <td>5</td>\n      <td>2) $100 - $200</td>\n      <td>131.51</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Urban</td>\n      <td>1</td>\n      <td>Web</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    </tr>\n    <tr>\n      <th>41828</th>\n      <td>11</td>\n      <td>4) $350 - $500</td>\n      <td>457.09</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Surburban</td>\n      <td>1</td>\n      <td>Multichannel</td>\n    </tr>\n    <tr>\n      <th>49354</th>\n      <td>2</td>\n      <td>1) $0 - $100</td>\n      <td>29.99</td>\n      <td>0</td>\n      <td>1</td>\n      <td>Rural</td>\n      <td>1</td>\n      <td>Phone</td>\n    </tr>\n    <tr>\n      <th>12063</th>\n      <td>9</td>\n      <td>2) $100 - $200</td>\n      <td>147.48</td>\n      <td>1</td>\n      <td>0</td>\n      <td>Rural</td>\n      <td>1</td>\n      <td>Phone</td>\n    </tr>\n    <tr>\n      <th>3757</th>\n      <td>10</td>\n      <td>2) $100 - $200</td>\n      <td>128.78</td>\n      <td>1</td>\n      <td>0</td>\n      <td>Surburban</td>\n      <td>1</td>\n      <td>Phone</td>\n    </tr>\n    <tr>\n      <th>10708</th>\n      <td>6</td>\n      <td>2) $100 - $200</td>\n      <td>138.72</td>\n      <td>1</td>\n      <td>0</td>\n      <td>Urban</td>\n      <td>0</td>\n      <td>Web</td>\n    </tr>\n  </tbody>\n</table>\n<p>12808 rows × 8 columns</p>\n</div>"
      },
-     "execution_count": 54,
+     "execution_count": 114,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "data.zip_code.value_counts()"
-   ],
-   "metadata": {
-    "collapsed": false,
-    "pycharm": {
-     "name": "#%%\n"
-    }
-   }
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 12,
-   "outputs": [],
-   "source": [
-    "X_train, X_val, y_train, y_val, trmnt_train, trmnt_val = train_test_split(\n",
-    "    data, target, treatment, test_size=0.3, random_state=42\n",
-    ")\n",
-    "\n",
-    "models_results = {\n",
-    "    'approach': [],\n",
-    "    'uplift@30%': []\n",
-    "}"
+    "X_val"
    ],
    "metadata": {
     "collapsed": false,

src/tools.py

@@ -4,6 +4,7 @@ import pandas as pd
 import numpy as np
 from sklearn.model_selection import train_test_split
 from sklift.datasets import fetch_hillstrom
+from sklift.metrics import uplift_at_k
 from catboost import CatBoostClassifier
 import sklearn
 import streamlit as st
@@ -28,7 +29,7 @@ def get_data() -> tuple[Any, Any, Any]:
 
 
 @st.experimental_memo
-def data_split(data, treatment, target) -> tuple[Any, Any, Any, Any, Any, Any]:
+def data_split(data: pd.DataFrame, treatment: pd.DataFrame, target: pd.DataFrame) -> tuple[Any, Any, Any, Any, Any, Any]:
 	# склеиваем threatment и target для дальнейшей стратификации по ним
 	stratify_cols = pd.concat([treatment, target], axis=1)
 	# сплитим датасет
@@ -43,6 +44,99 @@ def data_split(data, treatment, target) -> tuple[Any, Any, Any, Any, Any, Any]:
 	return X_train, X_val, trmnt_train, trmnt_val, y_train, y_val
 
 
+def filter_by_newbie(data: pd.DataFrame, newbie_filter: str) -> pd.DataFrame:
+	if newbie_filter == 'Всем':
+		return data
+	elif newbie_filter == 'Только новым':
+		return data[data['newbie'] == 1]
+	elif newbie_filter == 'Только старым':
+		return data[data['newbie'] == 0]
+
+
+def filter_by_channel(data: pd.DataFrame, channel_filter: str) -> pd.DataFrame:
+	if channel_filter == 'Всем':
+		return data
+	if channel_filter == 'Phone':
+		return data[data['channel'] == channel_filter]
+	if channel_filter == 'Web':
+		return data[data['channel'] == channel_filter]
+	if channel_filter == 'Multichannel':
+		return data[data['channel'] == channel_filter]
+
+
+def filter_by_mens(data: pd.DataFrame, mens_filter: str) -> pd.DataFrame:
+	if mens_filter == 'Любые товары':
+		return data
+	if mens_filter == 'Мужские':
+		return data[data['mens'] == 1]
+	if mens_filter == 'Женские':
+		return data[data['womens'] == 1]
+
+
+def filter_by_history_segments(data: pd.DataFrame, history_segments_filter: dict) -> pd.DataFrame:
+	filtered_indexes = set()
+	if history_segments_filter.get('1) $0 - $100'):
+		filtered_indexes = filtered_indexes.union(data[data['history_segment'] == '1) $0 - $100'].index)
+	if history_segments_filter.get('2) $100 - $200'):
+		filtered_indexes = filtered_indexes.union(data[data['history_segment'] == '2) $100 - $200'].index)
+	if history_segments_filter.get('3) $200 - $350'):
+		filtered_indexes = filtered_indexes.union(data[data['history_segment'] == '3) $200 - $350'].index)
+	if history_segments_filter.get('4) $350 - $500'):
+		filtered_indexes = filtered_indexes.union(data[data['history_segment'] == '4) $350 - $500'].index)
+	if history_segments_filter.get('5) $500 - $750'):
+		filtered_indexes = filtered_indexes.union(data[data['history_segment'] == '5) $500 - $750'].index)
+	if history_segments_filter.get('6) $750 - $1,000'):
+		filtered_indexes = filtered_indexes.union(data[data['history_segment'] == '6) $750 - $1,000'].index)
+	if history_segments_filter.get('7) $1,000 +'):
+		filtered_indexes = filtered_indexes.union(data[data['history_segment'] == '7) $1,000 +'].index)
+
+	return data.loc[list(filtered_indexes)]
+
+
+def filter_by_zip_code(data: pd.DataFrame, zip_code_filter: dict) -> pd.DataFrame:
+	filterd_indexes = set()
+	if zip_code_filter.get('surburban'):
+		filterd_indexes = filterd_indexes.union(data[data['zip_code'] == 'Surburban'].index)
+	if zip_code_filter.get('urban'):
+		filterd_indexes = filterd_indexes.union(data[data['zip_code'] == 'Urban'].index)
+	if zip_code_filter.get('rural'):
+		filterd_indexes = filterd_indexes.union(data[data['zip_code'] == 'Rural'].index)
+
+	return data.loc[list(filterd_indexes)]
+
+
+def filter_by_recency(data: pd.DataFrame, recency_filter: list) -> pd.DataFrame:
+	return data[(data['recency'] >= recency_filter[0]) & (data['recency'] <= recency_filter[1])]
+
+
+def filter_data(data: pd.DataFrame, filters: dict) -> pd.DataFrame or None:
+	data = filter_by_newbie(data, filters['newbie_filter'])
+	if data.shape[0] == 0:
+		return None
+	data = filter_by_channel(data, filters['channel_filter'])
+	if data.shape[0] == 0:
+		return None
+	data = filter_by_mens(data, filters['mens_filter'])
+	if data.shape[0] == 0:
+		return None
+	data = filter_by_history_segments(data, filters['history_segments'])
+	if data.shape[0] == 0:
+		return None
+	data = filter_by_zip_code(data, filters['zip_code'])
+	if data.shape[0] == 0:
+		return None
+	data = filter_by_recency(data, filters['recency'])
+	if data.shape[0] == 0:
+		return None
+	return data
+
+
+def send_promo_and_get_res(data_train: pd.DataFrame, treatment: pd.DataFrame, target: pd.DataFrame):
+	indexes = data.index
+	target = target.loc[indexes]
+	treatment = treatment.loc[indexes]
+
+
 def get_newbie_plot(data):
 	fig = px.histogram(
 		data['newbie'],

src/web_app.py

@@ -7,6 +7,8 @@ import tools
 
 dataset, target, treatment = tools.get_data()
 
+data_train, data_val, treatment_train, treatment_val, target_train, target_val = tools.data_split(dataset, treatment, target)
+
 st.title('Uplift lab')
 
 st.markdown(
@@ -17,6 +19,8 @@ st.markdown(
 	
 	Этот набор данных содержит 42 693 строк с данными клиентов, которые в последний раз совершали покупки в течение двенадцати месяцев.
 	
+	Из данных уже отделена тестовая выборка в виде 30% записей клиентов, так что данных в предоставленной выборке будет меньше.
+	
 	Среди клиентов была проведена рекламная кампания с помощью _email_ рассылки:
 	- 1/2 клиентов были выбраны случайным образом для получения электронного письма, рекламирующего женскую продукцию;
 	- С оставшейся 1/2 коммуникацию не проводили.
@@ -28,11 +32,11 @@ st.markdown(
 	"""
 )
 refresh = st.button('Обновить выборку')
-title_subsample = dataset.sample(7)
+title_subsample = data_train.sample(7)
 if refresh:
-	title_subsample = dataset.sample(7)
+	title_subsample = data_train.sample(7)
 st.dataframe(title_subsample, width=700)
-st.write( f"Всего записей: {dataset.shape[0]}")
+st.write(f"Всего записей: {data_train.shape[0]}")
 
 st.write('Описание данных')
 st.markdown(
@@ -56,59 +60,113 @@ st.write("Для того, чтобы лучше понять на какую а
 
 with st.expander('Развернуть блок анализа данных'):
 
-	st.plotly_chart(tools.get_newbie_plot(dataset), use_container_width=True)
+	st.plotly_chart(tools.get_newbie_plot(data_train), use_container_width=True)
 	st.write(f'В данных примерно одинаковое количество новых и "старых клиентов". '
-			 f'Отношение новых клиентов к старым: {(dataset["newbie"] == 1).sum() / (dataset["newbie"] == 0).sum():.2f}')
+			 f'Отношение новых клиентов к старым: {(data_train["newbie"] == 1).sum() / (data_train["newbie"] == 0).sum():.2f}')
 
-	st.plotly_chart(tools.get_zipcode_plot(dataset), use_container_width=True)
-	tmp_res = dataset.zip_code.value_counts(normalize=True) * 100
+	st.plotly_chart(tools.get_zipcode_plot(data_train), use_container_width=True)
+	tmp_res = data_train.zip_code.value_counts(normalize=True) * 100
 	st.write(f'Большинство клиентов из пригорода: {tmp_res["Surburban"]:.2f}%, из города: {tmp_res["Urban"]:.2f}% и из села: {tmp_res["Rural"]:.2f}%')
 
-	tmp_res = dataset.channel.value_counts(normalize=True) * 100
-	st.plotly_chart(tools.get_channel_plot(dataset), use_container_width=True)
+	tmp_res = data_train.channel.value_counts(normalize=True) * 100
+	st.plotly_chart(tools.get_channel_plot(data_train), use_container_width=True)
 	st.write(f'В прошлом году почти одинаковое количество клиентов покупало товары через телефон и сайт, {tmp_res["Phone"]:.2f}% и {tmp_res["Web"]:.2f}% соответственно,'
 	         f' а {tmp_res["Multichannel"]:.2f}% клиентов покупали товары воспользовавшись двумя платформами.')
 
-	tmp_res = dataset.history_segment.value_counts(normalize=True) * 100
-	st.plotly_chart(tools.get_history_segment_plot(dataset), use_container_width=True)
+	tmp_res = data_train.history_segment.value_counts(normalize=True) * 100
+	st.plotly_chart(tools.get_history_segment_plot(data_train), use_container_width=True)
 	st.write(f'Как мы видим, большинство пользователей относится к сегменту \$0-\$100 ({tmp_res[0]:.2f}%), второй и '
 	         f'третий по количеству пользователей сегменты \$100-\$200 ({tmp_res[1]:.2f}%) и \$200-\$350 ({tmp_res[2]:.2f}%).')
 	st.write(f'К сегментам \$350-\$500 и \$500-\$750 относится {tmp_res[3]:.2f}% и {tmp_res[4]:.2f}% пользователей соответственно.')
 	st.write(f'Меньше всего пользователей в сегментах \$750-\$1.000 ({tmp_res[-2]:.2f}%) и \$1.000+ ({tmp_res[-1]:.2f}%).')
 
-	tmp_res = list(dataset.recency.value_counts(normalize=True) * 100)
-	st.plotly_chart(tools.get_recency_plot(dataset), use_container_width=True)
+	tmp_res = list(data_train.recency.value_counts(normalize=True) * 100)
+	st.plotly_chart(tools.get_recency_plot(data_train), use_container_width=True)
 	st.write(f'Большинство клиентов являются активными клиентами платформы, и совершали покупки в течен��е месяца ({tmp_res[0]:.2f}%)')
 	st.write('Также заметно, что 9 и 10 месяцев назад, много клиентов совершали покупки. Это может свидетельствовать о проведении'
 	         'рекламной кампании в это время или чего-то еще.')
 	st.write('Также интересно понаблюдать за долями новых клиентов в данном распределении.')
 
-	st.plotly_chart(tools.get_history_plot(dataset), use_container_width=True)
-	st.markdown('_График интерактивный_')
+	st.plotly_chart(tools.get_history_plot(data_train), use_container_width=True)
+	st.markdown('_График интерактивный. Двойной клик вернет в начальное состояние._')
 	st.write('Абсолютное большинство клиентов тратят \$25-\$35 на покупки, но есть и малая доля тех, кто тратит более \$3.000')
 	st.write('Интересный факт: все покупки более \$500 совершают только новые клиенты')
 
-st.subheader('Какие данные возьмем для проведения рекламной кампании?')
-clients_filter = st.radio('На каких клиентов будем воздействовать?', options=['Все', 'Только на новых', 'Только на старых'])
+filters = {}
+
+st.subheader('Выберем клиентов, которым отправим рекламу.')
+newbie_filter = st.radio('Каким клиентам отправим рекламу?', options=['Всем', 'Только новым', 'Только старым'])
+filters['newbie_filter'] = newbie_filter
+
+channel_filter = st.radio('Канал, по которому клиент покупал в прошлом году', options=['Всем', 'Phone', 'Web', 'Multichannel'])
+filters['channel_filter'] = channel_filter
 
+mens_filter = st.radio('Клиенты, приобретавшие', options=['Любые товары', 'Мужские', 'Женские'])
+filters['mens_filter'] = mens_filter
 
-st.write('Выберите класс клиентов, по объему денег, потраченных в прошлом году (history segments).')
+st.write('Выберите класс клиентов, по объему денег, потраченных в прошлом году (history segments)')
+filters['history_segments'] = {}
 first_group = st.checkbox('$0-$100', value=True)
+if first_group:
+	filters['history_segments']['1) $0 - $100'] = True
 second_group = st.checkbox('$100-$200', value=True)
+if second_group:
+	filters['history_segments']['2) $100 - $200'] = True
 third_group = st.checkbox('$200-$350', value=True)
+if third_group:
+	filters['history_segments']['3) $200 - $350'] = True
 fourth_group = st.checkbox('$350-$500', value=True)
+if fourth_group:
+	filters['history_segments']['4) $350 - $500'] = True
 fifth_group = st.checkbox('$500-$750', value=True)
+if fifth_group:
+	filters['history_segments']['5) $500 - $750'] = True
 sixth_group = st.checkbox('$750-$1.000', value=True)
+if sixth_group:
+	filters['history_segments']['6) $750 - $1,000'] = True
 seventh_group = st.checkbox('$1.000+', value=True)
+if seventh_group:
+	filters['history_segments']['7) $1,000 +'] = True
 
 st.write('Каких пользователей по почтовому коду выберем')
+filters['zip_code'] = {}
 surburban = st.checkbox('Surburban', value=True)
+if surburban:
+	filters['zip_code']['surburban'] = True
 urban = st.checkbox('Urban', value=True)
+if urban:
+	filters['zip_code']['urban'] = True
 rural = st.checkbox('Rural', value=True)
+if rural:
+	filters['zip_code']['rural'] = True
 
-recency = st.slider(label='Месяцев с момента покупки', min_value=int(dataset.recency.min()), max_value=int(dataset.recency.max()), value=(int(dataset.recency.min()), int(dataset.recency.max())))
+recency = st.slider(label='Месяцев с момента покупки', min_value=int(data_train.recency.min()), max_value=int(data_train.recency.max()), value=(int(data_train.recency.min()), int(data_train.recency.max())))
+filters['recency'] = recency
 
+disabled = False
 if not first_group and not second_group and not third_group and not fourth_group and not fifth_group and not sixth_group and not seventh_group:
-	pass
+	st.error('Необходимо выбрать хотя бы один класс')
+	disabled = True
+elif not surburban and not urban and not rural:
+	st.error('Необходимо выбрать хотя бы один почтовый индекс')
+	disabled = True
+
+filter_data = st.button(label='Отфильтровать', disabled=disabled)
+
+if filter_data:
+	filtered_dataset = tools.filter_data(data_train, filters)
+	if filtered_dataset is None:
+		st.error('Нет подходящих под выбранный фильтр клиентов. Поп��обуйте изменить фильтр.')
+	sample_size = 7 if filtered_dataset.shape[0] >= 7 else filtered_dataset.shape[0]
+	example = filtered_dataset.sample(sample_size)
+	st.write('Пример пользователей, которым будет отправлена реклама')
+	st.dataframe(example)
+	st.info(f'Количество клиентов, которым реклама будет отправлена: _**{filtered_dataset.shape[0]}**_ ({filtered_dataset.shape[0] / data_train.shape[0] * 100 :.2f}% от всех клиентов)')
 else:
-	st.button('Отфильтровать')
+	# нельзя отправить рекламу, до фильтрации
+	disabled = True
+
+send_promo = st.button('Отправить рекламу и посмотреть результат', disabled=disabled)
+if send_promo:
+	pass
+