Capstone_Project.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "O5NFgjYa18SG"
   },
   "source": [
    "# Capstone Project - Iowa Liquor Retail Sales (Week 2)\n",
    "### Applied Data Science Capstone by IBM / Coursera"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "NJt1CZoElZw2"
   },
   "source": [
    "## Table of contents\n",
    "* [Introduction: Business Problem](#introduction)\n",
    "* [Data](#data)\n",
    "    * [Weather data](#weather)\n",
    "    * [Feature engineering](#feature_engineering)\n",
    "    * [Methodology](#methodology)\n",
    "* [Results and Discussion](#results)\n",
    "* [Conclusion](#conclusion)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "wklMafm-lZw4"
   },
   "source": [
    "## Introduction: Business Problem <a name=\"introduction\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "jnDNegg2lZw5"
   },
   "source": [
    "I will perform an analysis of a publicly available [data set](https://data.iowa.gov/Sales-Distribution/Iowa-Liquor-Sales/m3tr-qhgy) on liquor retail sales in the state of Iowa, U.S., to try and gain some insights on some of the most common retail problems. Retailers often want to forecast their demand in order to predict when and how much goods should be stocked.\n",
    "\n",
    "Analysis of this data set can provide insights into the habits of alcohol consumption in Iowa. They might provide liquor stores with more accurate information about their demand and help them better control their stocks; and they might help distributors optimize their delivery routes by predicting the demand days ahead and thus preventing over- or under-stocking of different types of alcohols (supply-chain problems).\n",
    "\n",
    "For this project, I will assume the role of owner of a liquor store in Iowa, and it is time to reorder inventory. Before I do so, I would like to get better idea of which products are most popular so that I order the right amounts.\n",
    "\n",
    "Using the visualisation tool on the Iowa government page, I had generated a plot of sales made by store name, and have decided to investigate the sales of the store that reported the most sales for the period for which the data is available, and that is the store `\"Hy-Vee #3 / BDI / Des Moines\"` in the `Polk` country, address: `\"3221 SE 14TH ST\"` (geographical point: `POINT(41.554101 -93.596754)`) which reported over `160K` sales between `2012-2020`."
   ]
  },
  {
   "attachments": {
    "IBM_Data_Science_SalesByStore.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "rH6GlijiD9mE"
   },
   "source": [
    "![IBM_Data_Science_SalesByStore.png](attachment:IBM_Data_Science_SalesByStore.png)!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "jPMn57PElZw6"
   },
   "source": [
    "## Introduction: Data <a name=\"data\"></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "51CWpKvJD9mG"
   },
   "source": [
    "The original data set contains 24 attributes much of whom are of no interest to us, therefore I will project the initial set to only three columns: `Date` - day the sale was made, `Item Description` - which liquor brand was bought, and `Bottles Sold` - the target variable we are trying to predict."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 419
    },
    "colab_type": "code",
    "id": "e_h4jODi2CV1",
    "outputId": "27806972-e763-4274-d6b1-3c681ac83fa4"
   },
   "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>Item Description</th>\n",
       "      <th>Bottles Sold</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2012-08-27</td>\n",
       "      <td>Uv Ivory (vanilla) Vodka</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2012-10-04</td>\n",
       "      <td>Dekuyper Peachtree</td>\n",
       "      <td>72</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2015-04-23</td>\n",
       "      <td>Absolut Mandrin</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2013-12-12</td>\n",
       "      <td>Dekuyper Hot Damn!</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2014-06-02</td>\n",
       "      <td>Paramount White Rum</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167228</th>\n",
       "      <td>2014-07-07</td>\n",
       "      <td>Smirnoff Sorbet Light Raspberry Pomegranate</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167229</th>\n",
       "      <td>2012-05-09</td>\n",
       "      <td>Bartenders Hot Sex</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167230</th>\n",
       "      <td>2013-10-10</td>\n",
       "      <td>Bacardi Wolf Berry</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167231</th>\n",
       "      <td>2015-05-21</td>\n",
       "      <td>Bacardi O</td>\n",
       "      <td>36</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167232</th>\n",
       "      <td>2012-05-14</td>\n",
       "      <td>Desert Island Long Island Ice Tea Cocktail</td>\n",
       "      <td>24</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>167233 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             Date                             Item Description  Bottles Sold\n",
       "0      2012-08-27                     Uv Ivory (vanilla) Vodka            12\n",
       "1      2012-10-04                           Dekuyper Peachtree            72\n",
       "2      2015-04-23                              Absolut Mandrin            12\n",
       "3      2013-12-12                           Dekuyper Hot Damn!            12\n",
       "4      2014-06-02                          Paramount White Rum             6\n",
       "...           ...                                          ...           ...\n",
       "167228 2014-07-07  Smirnoff Sorbet Light Raspberry Pomegranate             1\n",
       "167229 2012-05-09                           Bartenders Hot Sex             1\n",
       "167230 2013-10-10                           Bacardi Wolf Berry             1\n",
       "167231 2015-05-21                                    Bacardi O            36\n",
       "167232 2012-05-14   Desert Island Long Island Ice Tea Cocktail            24\n",
       "\n",
       "[167233 rows x 3 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "csv_filepath = '<FILEPATH>'\n",
    "column_dtypes = {'Date': str,\n",
    "              'Item Description': str,\n",
    "              'Bottles Sold': np.int64}\n",
    "\n",
    "store_sales_data = pd.read_csv(filepath_or_buffer=csv_filepath, \n",
    "                     delimiter=',',\n",
    "                     usecols=column_dtypes.keys(), \n",
    "                     dtype=column_dtypes, \n",
    "                     parse_dates=['Date'])\n",
    "store_sales_data['Date'] = pd.to_datetime(store_sales_data['Date']) # cast the `Date` column to `datetime`\n",
    "store_sales_data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "BLZUN_8wD9mN"
   },
   "source": [
    "To enhance our predictive power, we are going to incorporate weather data into our analysis. [NOAA's Global Historical Climatology Network (GHCN)](https://data.nodc.noaa.gov/cgi-bin/iso?id=gov.noaa.ncdc:C00861) contains complex weather data that might help us. It provides weather observation from more than 11K stations around the world, and for some of its stations the data goes back to year 1763!\n",
    "\n",
    "In our scenario, we would like to obtain some weather parameters from the store's nearest weather stations.\n",
    "\n",
    "First, we will obtain a list of weather stations that are in 50 km radius from the store. Then, we will compare that list with the weather data for the period 2012-2020, so that we are sure that the stations are still operating."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "a0aIKZCeD9mO"
   },
   "source": [
    "### Weather data<a name='weather'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "niKyw9BwD9mP"
   },
   "source": [
    "The weather data is scattered throughout many files, and their structure is a bit messy. If you want to know more about the structuring of the files, I would kindly refer you to read the `readme` file on this [link](https://www1.ncdc.noaa.gov/pub/data/ghcn/daily/readme.txt) and look over all the data files [here](https://www1.ncdc.noaa.gov/pub/data/ghcn/daily/). The `ghcnd-stations.txt` file is csv-like file that contains information about the weather stations and their metadata. For our purpouses, we just need the station's code and coordinates (first three columns, separated by different number of spaces)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 419
    },
    "colab_type": "code",
    "id": "4IknwqOuD9mQ",
    "outputId": "e6a9673a-5eca-4b40-eeb0-26f95988c89d"
   },
   "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>Station Code</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>Longitude</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ACW00011604</td>\n",
       "      <td>17.1167</td>\n",
       "      <td>-61.7833</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ACW00011647</td>\n",
       "      <td>17.1333</td>\n",
       "      <td>-61.7833</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>AE000041196</td>\n",
       "      <td>25.3330</td>\n",
       "      <td>55.5170</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>AEM00041194</td>\n",
       "      <td>25.2550</td>\n",
       "      <td>55.3640</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>AEM00041217</td>\n",
       "      <td>24.4330</td>\n",
       "      <td>54.6510</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115077</th>\n",
       "      <td>ZI000067969</td>\n",
       "      <td>-21.0500</td>\n",
       "      <td>29.3670</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115078</th>\n",
       "      <td>ZI000067975</td>\n",
       "      <td>-20.0670</td>\n",
       "      <td>30.8670</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115079</th>\n",
       "      <td>ZI000067977</td>\n",
       "      <td>-21.0170</td>\n",
       "      <td>31.5830</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115080</th>\n",
       "      <td>ZI000067983</td>\n",
       "      <td>-20.2000</td>\n",
       "      <td>32.6160</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115081</th>\n",
       "      <td>ZI000067991</td>\n",
       "      <td>-22.2170</td>\n",
       "      <td>30.0000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>115082 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Station Code  Latitude Longitude\n",
       "0       ACW00011604   17.1167  -61.7833\n",
       "1       ACW00011647   17.1333  -61.7833\n",
       "2       AE000041196   25.3330   55.5170\n",
       "3       AEM00041194   25.2550   55.3640\n",
       "4       AEM00041217   24.4330   54.6510\n",
       "...             ...       ...       ...\n",
       "115077  ZI000067969  -21.0500   29.3670\n",
       "115078  ZI000067975  -20.0670   30.8670\n",
       "115079  ZI000067977  -21.0170   31.5830\n",
       "115080  ZI000067983  -20.2000   32.6160\n",
       "115081  ZI000067991  -22.2170   30.0000\n",
       "\n",
       "[115082 rows x 3 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "store_location = (41.554101, -93.596754) # lat, long\n",
    "import io\n",
    "station_locations_filepath = '<FILEPATH>'\n",
    "lines = io.open(station_locations_filepath, encoding='UTF-8').read().strip().split('\\n')\n",
    "data = list()\n",
    "for line in lines:\n",
    "    splitted = line.split(' ') # split line on empty spaces\n",
    "    splitted = [x for x in splitted if x != ''] # filter out empty strings\n",
    "    data.append(splitted[:3]) # get first three parts: station code, lat coord, long coord\n",
    "station_locations = pd.DataFrame(data=data, columns=['Station Code', 'Latitude', 'Longitude'])\n",
    "station_locations"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "XYA4_mkGD9mV"
   },
   "source": [
    "Once we have our data in the proper format, we can filter out the station codes that are within radius to our store."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 419
    },
    "colab_type": "code",
    "id": "a6d1hFPsD9mW",
    "outputId": "9575927c-71e6-48c6-8710-0b068481a2cf"
   },
   "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>Station Code</th>\n",
       "      <th>Distance (km)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>US1IABN0009</td>\n",
       "      <td>39.630102</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>US1IABN0013</td>\n",
       "      <td>49.540551</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>US1IADL0001</td>\n",
       "      <td>31.437682</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>US1IADL0004</td>\n",
       "      <td>34.309993</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>US1IADL0005</td>\n",
       "      <td>24.139138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>USC00139132</td>\n",
       "      <td>44.119406</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>93</th>\n",
       "      <td>USR0000INEA</td>\n",
       "      <td>28.267976</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>94</th>\n",
       "      <td>USW00014933</td>\n",
       "      <td>5.209395</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95</th>\n",
       "      <td>USW00054902</td>\n",
       "      <td>25.959069</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>USW00094989</td>\n",
       "      <td>48.703350</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>97 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Station Code  Distance (km)\n",
       "0   US1IABN0009      39.630102\n",
       "1   US1IABN0013      49.540551\n",
       "2   US1IADL0001      31.437682\n",
       "3   US1IADL0004      34.309993\n",
       "4   US1IADL0005      24.139138\n",
       "..          ...            ...\n",
       "92  USC00139132      44.119406\n",
       "93  USR0000INEA      28.267976\n",
       "94  USW00014933       5.209395\n",
       "95  USW00054902      25.959069\n",
       "96  USW00094989      48.703350\n",
       "\n",
       "[97 rows x 2 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from geopy.distance import distance\n",
    "\n",
    "radius = 50 #km\n",
    "stations_within_radius = [(row[0], distance(store_location, (float(row[1]), float(row[2]))).km) \n",
    "                          for row in station_locations.values \n",
    "                          if distance(store_location, (float(row[1]), float(row[2]))).km < radius]\n",
    "stations_within_radius = pd.DataFrame(data=stations_within_radius, columns=['Station Code', 'Distance (km)'])\n",
    "stations_within_radius"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "aeYttRKWD9mb"
   },
   "source": [
    "There are 97 weather stations within 50km radius of the store. Let us now check how many of them are active in the period of 2012-2020 and have recorded the weather attributes of interest to us. Upon examining the `ghcnd-inventory` file in the data set, we will filter the weather stations active in period 2012-2020 who have recorded the:\n",
    "\n",
    "- Maximum temperature for the day (TMAX flag)\n",
    "- Minimum temperature for the day (TMIN flag)\n",
    "- Amount of snow fall (SNOW flag)\n",
    "- Amount of precipitation (PRCP flag)\n",
    "\n",
    "This information is contained in the `inventory` data set. For each station, it records the `first` and `last` year of measuring each `element`. We would require that each nearby station has records for all of these four elements and that their `first` year is less than `2012`, while the latest year is `2020`.\n",
    "\n",
    "So, first we read the station inventory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "JC3Xa8C9D9mc"
   },
   "outputs": [],
   "source": [
    "inventory_filepath = '<FILEPATH>'\n",
    "inventory_df_column_names = ['Station Code', 'Latitude', 'Longitude', 'Element', 'First Year', 'Last Year']\n",
    "lines = io.open(inventory_filepath, encoding='UTF-8').read().strip().split('\\n')\n",
    "data = list()\n",
    "# data = list()\n",
    "for line in lines:\n",
    "    splitted = line.split(' ') # split line on empty spaces\n",
    "    splitted = [x for x in splitted if x != ''] # filter out empty strings\n",
    "#     print (len(splitted))\n",
    "    data.append(splitted)\n",
    "inventory_df = pd.DataFrame(data=data, columns=inventory_df_column_names)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "znRNzb7aD9mh"
   },
   "source": [
    "Next, we will query this data set to consider only the stations that are nearby our store, and that have recorded the weather elements for the period of interest."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 419
    },
    "colab_type": "code",
    "id": "-R5aBJuID9mi",
    "outputId": "b3ae0384-a814-44dc-f0bc-dea953979766"
   },
   "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>Station Code</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>Longitude</th>\n",
       "      <th>Element</th>\n",
       "      <th>First Year</th>\n",
       "      <th>Last Year</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>227245</th>\n",
       "      <td>US1IADL0017</td>\n",
       "      <td>41.7392</td>\n",
       "      <td>-93.9635</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>1998</td>\n",
       "      <td>2020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>227249</th>\n",
       "      <td>US1IADL0018</td>\n",
       "      <td>41.6085</td>\n",
       "      <td>-93.8736</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>2012</td>\n",
       "      <td>2020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>228040</th>\n",
       "      <td>US1IAMD0003</td>\n",
       "      <td>41.4964</td>\n",
       "      <td>-94.1599</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>1998</td>\n",
       "      <td>2020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>228041</th>\n",
       "      <td>US1IAMD0003</td>\n",
       "      <td>41.4964</td>\n",
       "      <td>-94.1599</td>\n",
       "      <td>SNOW</td>\n",
       "      <td>1998</td>\n",
       "      <td>2020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>228167</th>\n",
       "      <td>US1IAPK0003</td>\n",
       "      <td>41.6066</td>\n",
       "      <td>-93.6956</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>2006</td>\n",
       "      <td>2020</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>676789</th>\n",
       "      <td>USW00054902</td>\n",
       "      <td>41.5561</td>\n",
       "      <td>-93.2856</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>2004</td>\n",
       "      <td>2020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>685252</th>\n",
       "      <td>USW00094989</td>\n",
       "      <td>41.9922</td>\n",
       "      <td>-93.6217</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>1996</td>\n",
       "      <td>2020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>685253</th>\n",
       "      <td>USW00094989</td>\n",
       "      <td>41.9922</td>\n",
       "      <td>-93.6217</td>\n",
       "      <td>TMIN</td>\n",
       "      <td>1996</td>\n",
       "      <td>2020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>685254</th>\n",
       "      <td>USW00094989</td>\n",
       "      <td>41.9922</td>\n",
       "      <td>-93.6217</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>1998</td>\n",
       "      <td>2020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>685255</th>\n",
       "      <td>USW00094989</td>\n",
       "      <td>41.9922</td>\n",
       "      <td>-93.6217</td>\n",
       "      <td>SNOW</td>\n",
       "      <td>2012</td>\n",
       "      <td>2020</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>65 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Station Code Latitude Longitude Element First Year Last Year\n",
       "227245  US1IADL0017  41.7392  -93.9635    PRCP       1998      2020\n",
       "227249  US1IADL0018  41.6085  -93.8736    PRCP       2012      2020\n",
       "228040  US1IAMD0003  41.4964  -94.1599    PRCP       1998      2020\n",
       "228041  US1IAMD0003  41.4964  -94.1599    SNOW       1998      2020\n",
       "228167  US1IAPK0003  41.6066  -93.6956    PRCP       2006      2020\n",
       "...             ...      ...       ...     ...        ...       ...\n",
       "676789  USW00054902  41.5561  -93.2856    PRCP       2004      2020\n",
       "685252  USW00094989  41.9922  -93.6217    TMAX       1996      2020\n",
       "685253  USW00094989  41.9922  -93.6217    TMIN       1996      2020\n",
       "685254  USW00094989  41.9922  -93.6217    PRCP       1998      2020\n",
       "685255  USW00094989  41.9922  -93.6217    SNOW       2012      2020\n",
       "\n",
       "[65 rows x 6 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# define query parameters\n",
    "weather_elements = ['TMAX', 'TMIN', 'SNOW', 'PRCP']\n",
    "first_year = '2012'\n",
    "last_year = '2020'\n",
    "nearby_station_codes = list(stations_within_radius['Station Code'])\n",
    "\n",
    "# query the dataframe\n",
    "inventory_df.query('`Station Code` in @nearby_station_codes & \\\n",
    "                        `First Year` <= @first_year & \\\n",
    "                        `Last Year` == @last_year & \\\n",
    "                        `Element` in @weather_elements', inplace=True) # modify the original inventory data set\n",
    "\n",
    "inventory_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "gdgDw_QfD9mo"
   },
   "source": [
    "This is still not our final list of stations, as some of them may have not recorded **all** of the weather attributes. To further filter our list, and obtain the final result, we group the `inventory` by `Station Code` and request that it has a length of `4`, i.e. the number of elements in our list."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 612
    },
    "colab_type": "code",
    "id": "CtpqThz7D9mp",
    "outputId": "5f13a002-698f-4df2-f47a-aa5a85ab79ad"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "       Station Code Latitude Longitude Element First Year Last Year\n",
      "419046  USC00130203  41.9519  -93.5656    TMAX       1978      2020\n",
      "419047  USC00130203  41.9519  -93.5656    TMIN       1978      2020\n",
      "419049  USC00130203  41.9519  -93.5656    PRCP       1978      2020\n",
      "419050  USC00130203  41.9519  -93.5656    SNOW       1978      2020\n",
      "       Station Code Latitude Longitude Element First Year Last Year\n",
      "419125  USC00130241  41.7322  -93.6500    TMAX       1950      2020\n",
      "419126  USC00130241  41.7322  -93.6500    TMIN       1950      2020\n",
      "419128  USC00130241  41.7322  -93.6500    PRCP       1950      2020\n",
      "419129  USC00130241  41.7322  -93.6500    SNOW       1950      2020\n",
      "       Station Code Latitude Longitude Element First Year Last Year\n",
      "420276  USC00132209  41.7367  -93.7236    TMAX       1995      2020\n",
      "420277  USC00132209  41.7367  -93.7236    TMIN       1995      2020\n",
      "420279  USC00132209  41.7367  -93.7236    PRCP       1995      2020\n",
      "420280  USC00132209  41.7367  -93.7236    SNOW       1995      2020\n",
      "       Station Code Latitude Longitude Element First Year Last Year\n",
      "421162  USC00134063  41.3664  -93.6481    TMAX       1893      2020\n",
      "421163  USC00134063  41.3664  -93.6481    TMIN       1893      2020\n",
      "421165  USC00134063  41.3664  -93.6481    PRCP       1893      2020\n",
      "421166  USC00134063  41.3664  -93.6481    SNOW       1893      2020\n",
      "       Station Code Latitude Longitude Element First Year Last Year\n",
      "421458  USC00134502  41.3247  -93.1008    TMAX       1893      2020\n",
      "421459  USC00134502  41.3247  -93.1008    TMIN       1893      2020\n",
      "421461  USC00134502  41.3247  -93.1008    PRCP       1893      2020\n",
      "421462  USC00134502  41.3247  -93.1008    SNOW       1893      2020\n",
      "       Station Code Latitude Longitude Element First Year Last Year\n",
      "662742  USW00014933  41.5339  -93.6531    TMAX       1945      2020\n",
      "662743  USW00014933  41.5339  -93.6531    TMIN       1945      2020\n",
      "662744  USW00014933  41.5339  -93.6531    PRCP       1945      2020\n",
      "662745  USW00014933  41.5339  -93.6531    SNOW       1945      2020\n",
      "       Station Code Latitude Longitude Element First Year Last Year\n",
      "685252  USW00094989  41.9922  -93.6217    TMAX       1996      2020\n",
      "685253  USW00094989  41.9922  -93.6217    TMIN       1996      2020\n",
      "685254  USW00094989  41.9922  -93.6217    PRCP       1998      2020\n",
      "685255  USW00094989  41.9922  -93.6217    SNOW       2012      2020\n"
     ]
    }
   ],
   "source": [
    "grouped_stations = inventory_df.groupby(by = 'Station Code')\n",
    "data = list()\n",
    "for _, group in grouped_stations:\n",
    "    if len(group.index) == 4:\n",
    "        data.append(group[['Station Code', 'Latitude', 'Longitude']].iloc[0]) # take the first row to get the station's info\n",
    "        print (group) # visual confirmation that the station records all four weather elements\n",
    "operating_stations = pd.DataFrame(data=data, columns=['Station Code', 'Latitude', 'Longitude'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "GBmg0JMLD9mu"
   },
   "source": [
    "There are seven weather stations in the store's vicinity that record **all** four weather attributes for the period 2012-2020. For the final step, we are going to sort them by distance to the store, and choose the nearest one as the most reliable."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 102
    },
    "colab_type": "code",
    "id": "bdrVdhlKD9mv",
    "outputId": "241068dc-d1e2-4fce-e223-968c213caf1e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Station Code     USW00014933\n",
       "Latitude             41.5339\n",
       "Longitude           -93.6531\n",
       "Distance (km)         5.2094\n",
       "Name: 5, dtype: object"
      ]
     },
     "execution_count": 8,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "closest_operating_station = pd.merge(left=operating_stations, \n",
    "                                     right=stations_within_radius, \n",
    "                                     how='inner', \n",
    "                                     on='Station Code').sort_values(by='Distance (km)', \n",
    "                                                                    ascending=True).iloc[0]\n",
    "closest_operating_station"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "j15V-gJHD9m0"
   },
   "source": [
    "Et, voilà! We finally have our closest weather station. It is located only 5.2 km from the store, at `POINT(41.5339, -93.6531)` with code `USW00014933`.\n",
    "\n",
    "For the next step, we will get the actual data associated with this station so that we can merge it with our sales data. The directory `gsn` contains a `<station-code>.dly` file with all the data associated with a particular station. We obtain that file, and filter it to get only the measurements for the period 2012-2020 for the weather elements (TMAX, TMIN, SNOW and PRCP).\n",
    "\n",
    "Again, the structure of the file is very messy and has to be studied and handled carefully and thoroughly.\n",
    "\n",
    "Each line in the file contains one month of daily data. The line structure is as follows:\n",
    "\n",
    "* Station code - first 11 characters\n",
    "* Year - (11, 15) characters\n",
    "* Month - (15, 17) characters\n",
    "* Weather element (17, 21) characters\n",
    "* Value for day 1 of the month\n",
    "* Three consecutive measurement quality flags (should be skipped)\n",
    "* Value for day 2 of the month\n",
    "* Three consecutive measurement quality flags (should be skipped)\n",
    "* ... continues for 31 days total inputting missing values (-9999) for months with smaller number of days\n",
    "\n",
    "After a few unsuccessfull try outs, I have decided to parse each line in this file by character. Again, with the help of the `readme` file as previously, I have managed to read the data, but the code is very error-prone and unreadable in regards to the character indices. Therefore, in the absence of a better solution, this will do.\n",
    "\n",
    "While reading the data, we are also cleaning it by:\n",
    "\n",
    "1. Replacing all missing values (with value `-9999`) with the first previously known value\n",
    "2. Changing the unit measured for the temperature (from tenth of `Celsius` to `Fahrenheit`)\n",
    "3. Changing the unit measured for the precipitation (from tenth of `mm` to just `mm`) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 419
    },
    "colab_type": "code",
    "id": "xFH3lMxQD9m1",
    "outputId": "3ad62e0f-f97b-4b8e-87ef-4cc3420efcb5"
   },
   "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>Weather Element</th>\n",
       "      <th>Weather Value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2012-01-01</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>28.04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2012-01-02</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>39.92</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2012-01-03</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>46.94</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2012-01-04</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>64.94</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2012-01-05</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>53.96</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12535</th>\n",
       "      <td>2020-09-26</td>\n",
       "      <td>SNOW</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12536</th>\n",
       "      <td>2020-09-27</td>\n",
       "      <td>SNOW</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12537</th>\n",
       "      <td>2020-09-28</td>\n",
       "      <td>SNOW</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12538</th>\n",
       "      <td>2020-09-29</td>\n",
       "      <td>SNOW</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12539</th>\n",
       "      <td>2020-09-30</td>\n",
       "      <td>SNOW</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>12540 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            Date Weather Element  Weather Value\n",
       "0     2012-01-01            TMAX          28.04\n",
       "1     2012-01-02            TMAX          39.92\n",
       "2     2012-01-03            TMAX          46.94\n",
       "3     2012-01-04            TMAX          64.94\n",
       "4     2012-01-05            TMAX          53.96\n",
       "...          ...             ...            ...\n",
       "12535 2020-09-26            SNOW           0.00\n",
       "12536 2020-09-27            SNOW           0.00\n",
       "12537 2020-09-28            SNOW           0.00\n",
       "12538 2020-09-29            SNOW           0.00\n",
       "12539 2020-09-30            SNOW           0.00\n",
       "\n",
       "[12540 rows x 3 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import datetime\n",
    "\n",
    "def replace_missing(lst, missing_value=-9999):\n",
    "    \"\"\"Replaces missing values with the first past or future known value, depending on position.\"\"\"\n",
    "    def first_non_missing_value():\n",
    "        \"\"\"Gets the first non-missing (non `-9999` value) from the list. Returns 0 if there is no such element.\"\"\"\n",
    "        for inx in range(0, len(lst)):\n",
    "            if lst[inx] != missing_value:\n",
    "                return [lst[inx]]\n",
    "        return 0\n",
    "    \n",
    "    result = first_non_missing_value() # the first element of the result is the first non-missing value in the list\n",
    "    for inx in range(1, len(lst)):\n",
    "        if lst[inx] == missing_value:\n",
    "            result.append(result[inx - 1]) # replace it with the last added value\n",
    "        else:\n",
    "            result.append(lst[inx])\n",
    "    return result\n",
    "\n",
    "station_daily_data_filepath = '<FILEPATH>'\n",
    "\n",
    "lines = io.open(station_daily_data_filepath, encoding='UTF-8').read().strip().split('\\n')\n",
    "data = list()\n",
    "for line in lines:\n",
    "    year = line[11: 15] # year as 4-char string\n",
    "    if year < first_year: # if data is older than 2012, skip the line\n",
    "        continue\n",
    "        \n",
    "    month = line[15: 17] # month as 2-char string\n",
    "    weather_element = line[17: 21] # weather element as 4-char string\n",
    "    \n",
    "    if weather_element not in weather_elements: # if weather element is of no interes, skip the line\n",
    "        continue\n",
    "    \n",
    "    daily_values = list()\n",
    "    # Read and clean the values\n",
    "    for day in range(0, 31): # iterate days\n",
    "        day_value_index = 21 + day * 8 # starting char-index\n",
    "        value = int(line[day_value_index: day_value_index + 5])\n",
    "        daily_values.append(value)\n",
    "    daily_values = replace_missing(daily_values)\n",
    "    \n",
    "    # If the values are temperatures (originally stored as VALUE*10 Celsius)\n",
    "    # Convert them to Fahrenheit\n",
    "    if weather_element in ['TMAX', 'TMIN']:\n",
    "        daily_values = [(float(x) / 10) * (9 / 5) + 32 for x in daily_values]\n",
    "    if weather_element == 'PRCP':\n",
    "        daily_values = [float(x) / 10 for x in daily_values]\n",
    "    \n",
    "    \n",
    "    for day in range(0, 31):\n",
    "        try:\n",
    "            date = datetime.date(year=int(year), month=int(month), day=day)\n",
    "        except ValueError: # A value error is thrown when we try to create an impossible date, say 2012-02-30\n",
    "            continue\n",
    "        value = daily_values[day]\n",
    "        data.append([date, weather_element, value])\n",
    "column_names = ['Date', 'Weather Element', 'Weather Value']\n",
    "station_data = pd.DataFrame(data=data, columns=column_names)\n",
    "station_data['Date'] = pd.to_datetime(station_data['Date']) # cast the `Date` column to `datetime`\n",
    "station_data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "4-3XadwQD9m6"
   },
   "source": [
    "Now that we finally have our station weather data in proper format, we can merge it with our store sales data by `Date`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 419
    },
    "colab_type": "code",
    "id": "7YJPgsbdD9m7",
    "outputId": "eefc4475-eea9-4835-cc8f-f8fb05aba1c2"
   },
   "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>Weather Element</th>\n",
       "      <th>Weather Value</th>\n",
       "      <th>Item Description</th>\n",
       "      <th>Bottles Sold</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2012-01-03</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>46.94</td>\n",
       "      <td>Three Olives Cake Flavored Vodka</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2012-01-03</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>46.94</td>\n",
       "      <td>Dekuyper 03</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2012-01-03</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>46.94</td>\n",
       "      <td>Pearl Coconut Vodka</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2012-01-03</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>46.94</td>\n",
       "      <td>Godiva Liqueur</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2012-01-03</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>46.94</td>\n",
       "      <td>Uv Ivory (vanilla) Vodka</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>655779</th>\n",
       "      <td>2020-08-27</td>\n",
       "      <td>SNOW</td>\n",
       "      <td>0.00</td>\n",
       "      <td>La Prima Gold</td>\n",
       "      <td>72</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>655780</th>\n",
       "      <td>2020-08-28</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>80.06</td>\n",
       "      <td>Western Son Watermelon Vodka</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>655781</th>\n",
       "      <td>2020-08-28</td>\n",
       "      <td>TMIN</td>\n",
       "      <td>62.96</td>\n",
       "      <td>Western Son Watermelon Vodka</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>655782</th>\n",
       "      <td>2020-08-28</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>0.00</td>\n",
       "      <td>Western Son Watermelon Vodka</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>655783</th>\n",
       "      <td>2020-08-28</td>\n",
       "      <td>SNOW</td>\n",
       "      <td>0.00</td>\n",
       "      <td>Western Son Watermelon Vodka</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>655784 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             Date  ... Bottles Sold\n",
       "0      2012-01-03  ...            1\n",
       "1      2012-01-03  ...            2\n",
       "2      2012-01-03  ...            1\n",
       "3      2012-01-03  ...           12\n",
       "4      2012-01-03  ...           12\n",
       "...           ...  ...          ...\n",
       "655779 2020-08-27  ...           72\n",
       "655780 2020-08-28  ...            1\n",
       "655781 2020-08-28  ...            1\n",
       "655782 2020-08-28  ...            1\n",
       "655783 2020-08-28  ...            1\n",
       "\n",
       "[655784 rows x 5 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged_weather = pd.merge(left=station_data, right=store_sales_data, how='inner', on='Date')\n",
    "merged_weather"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "iNd1TqO3D9nB"
   },
   "source": [
    "### Feature engineering<a name='feature_engineering'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "U-RCrlNZD9nB"
   },
   "source": [
    "In this section we will prepare the data for training a machine learning model and perform some feature engineering to increase the model's predictive performance. Forecasting models use data from the past to make predictions for the future. The data in our model will be historical sales and weather statistics, and we will use it to predict sales in the future.\n",
    "\n",
    "More specifically, we are trying to predict the `Bottles Sold` variable for each brand of liquor - `Item Description` for a given month (`Year-Month` variable), on the first of that month. For example, we take the data as of January 2020 and use it to predict the sales for February 2020, data as of February 2020 to predict March 2020, etc.\n",
    "\n",
    "The feature that we are going to use to predict sales are the following:\n",
    "\n",
    "* `Month` - the month of the year we are trying to predict\n",
    "* `Month Sales Lag X` - how many bottles of the brand of liquor were sold in the previous month (X = 1), the month before that (X = 2), the month before that (X = 3), and one year ago (X = 12)\n",
    "* `PRCP Lag X` - the average monthly precipitation last month (X = 1) and one year ago (X = 12)\n",
    "* `SNOW Lag X` - the average monthly snowfall last month (X = 1) and one year ago (X = 12)\n",
    "* `TMIN Lag X` - the average monthly minimum temperature last month (X = 1) and one year ago (X = 12)\n",
    "* `TMAX Lag X` - the average monthly maximum temperature last month (X = 1) and one year ago (X = 12)\n",
    "\n",
    "A few things to consider here.\n",
    "\n",
    "1. The features we are trying to construct require data up to 12 months prior to make prediction for a given month. Our data set starts in 2012, but since we need 12 months of data to start predictions, we can not train on 2012 data nor have it included into our training data set.\n",
    "2. Some of liquor brands are rarely stocked, probably only when the inventory gets too low. These are unpopular brands and we will not model them. Therefore, they too will be removed from the data set. We will set up a threshold parameter `popularity`, and all liquor brands must have at least `popularity` bottles sold by our store since 2012.\n",
    "\n",
    "\n",
    "Let us start by adding the `Year-Month` column to the data set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 419
    },
    "colab_type": "code",
    "id": "IRzBl26hD9nC",
    "outputId": "567307e3-7eed-4a8d-8b84-3f91aec0042b"
   },
   "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>Weather Element</th>\n",
       "      <th>Weather Value</th>\n",
       "      <th>Item Description</th>\n",
       "      <th>Bottles Sold</th>\n",
       "      <th>Year-Month</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2012-01-03</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>46.94</td>\n",
       "      <td>Three Olives Cake Flavored Vodka</td>\n",
       "      <td>1</td>\n",
       "      <td>2012-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2012-01-03</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>46.94</td>\n",
       "      <td>Dekuyper 03</td>\n",
       "      <td>2</td>\n",
       "      <td>2012-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2012-01-03</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>46.94</td>\n",
       "      <td>Pearl Coconut Vodka</td>\n",
       "      <td>1</td>\n",
       "      <td>2012-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2012-01-03</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>46.94</td>\n",
       "      <td>Godiva Liqueur</td>\n",
       "      <td>12</td>\n",
       "      <td>2012-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2012-01-03</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>46.94</td>\n",
       "      <td>Uv Ivory (vanilla) Vodka</td>\n",
       "      <td>12</td>\n",
       "      <td>2012-01</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>655779</th>\n",
       "      <td>2020-08-27</td>\n",
       "      <td>SNOW</td>\n",
       "      <td>0.00</td>\n",
       "      <td>La Prima Gold</td>\n",
       "      <td>72</td>\n",
       "      <td>2020-08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>655780</th>\n",
       "      <td>2020-08-28</td>\n",
       "      <td>TMAX</td>\n",
       "      <td>80.06</td>\n",
       "      <td>Western Son Watermelon Vodka</td>\n",
       "      <td>1</td>\n",
       "      <td>2020-08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>655781</th>\n",
       "      <td>2020-08-28</td>\n",
       "      <td>TMIN</td>\n",
       "      <td>62.96</td>\n",
       "      <td>Western Son Watermelon Vodka</td>\n",
       "      <td>1</td>\n",
       "      <td>2020-08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>655782</th>\n",
       "      <td>2020-08-28</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>0.00</td>\n",
       "      <td>Western Son Watermelon Vodka</td>\n",
       "      <td>1</td>\n",
       "      <td>2020-08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>655783</th>\n",
       "      <td>2020-08-28</td>\n",
       "      <td>SNOW</td>\n",
       "      <td>0.00</td>\n",
       "      <td>Western Son Watermelon Vodka</td>\n",
       "      <td>1</td>\n",
       "      <td>2020-08</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>655784 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             Date Weather Element  ...  Bottles Sold Year-Month\n",
       "0      2012-01-03            TMAX  ...             1    2012-01\n",
       "1      2012-01-03            TMAX  ...             2    2012-01\n",
       "2      2012-01-03            TMAX  ...             1    2012-01\n",
       "3      2012-01-03            TMAX  ...            12    2012-01\n",
       "4      2012-01-03            TMAX  ...            12    2012-01\n",
       "...           ...             ...  ...           ...        ...\n",
       "655779 2020-08-27            SNOW  ...            72    2020-08\n",
       "655780 2020-08-28            TMAX  ...             1    2020-08\n",
       "655781 2020-08-28            TMIN  ...             1    2020-08\n",
       "655782 2020-08-28            PRCP  ...             1    2020-08\n",
       "655783 2020-08-28            SNOW  ...             1    2020-08\n",
       "\n",
       "[655784 rows x 6 columns]"
      ]
     },
     "execution_count": 11,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# \"{}-{}\".format(data['Date'][0].year, data['Date'][0].month)\n",
    "year_month = [\"{}-{:02d}\".format(row.year, row.month) for row in merged_weather['Date']]\n",
    "merged_weather['Year-Month'] = year_month\n",
    "merged_weather"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "P0EYt3b6D9nH"
   },
   "source": [
    "Group the data by `Year-Month`, `Item Description`, `Weather Element` and aggregate the other columns - calculate the mean values for the weather attribute, and sum up the sales."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 419
    },
    "colab_type": "code",
    "id": "amPSvKhcD9nI",
    "outputId": "fd2079de-cf49-4856-8550-db4b2aa9fd87"
   },
   "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>Year-Month</th>\n",
       "      <th>Weather Element</th>\n",
       "      <th>Weather Value</th>\n",
       "      <th>Item Description</th>\n",
       "      <th>Bottles Sold</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2012-01</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>1800 Reposado Tequila</td>\n",
       "      <td>432</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2012-01</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>1800 Silver Tequila</td>\n",
       "      <td>96</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2012-01</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>360 Double Chocolate Vodka</td>\n",
       "      <td>48</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2012-01</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>360 Georgia Peach Vodka</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2012-01</td>\n",
       "      <td>PRCP</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>360 Madagascar Vanilla Vodka</td>\n",
       "      <td>96</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>214983</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>TMIN</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>Windsor Canadian</td>\n",
       "      <td>96</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>214984</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>TMIN</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>Woodford Reserve</td>\n",
       "      <td>576</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>214985</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>TMIN</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>Woodford Reserve Double Oaked</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>214986</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>TMIN</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>Woodford Reserve Kentucky Derby 2020</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>214987</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>TMIN</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>Yukon Jack</td>\n",
       "      <td>144</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>214988 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Year-Month  ... Bottles Sold\n",
       "0         2012-01  ...          432\n",
       "1         2012-01  ...           96\n",
       "2         2012-01  ...           48\n",
       "3         2012-01  ...            4\n",
       "4         2012-01  ...           96\n",
       "...           ...  ...          ...\n",
       "214983    2020-08  ...           96\n",
       "214984    2020-08  ...          576\n",
       "214985    2020-08  ...            8\n",
       "214986    2020-08  ...            8\n",
       "214987    2020-08  ...          144\n",
       "\n",
       "[214988 rows x 5 columns]"
      ]
     },
     "execution_count": 12,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "aggregated_weather_data = merged_weather.groupby(by=['Year-Month', 'Weather Element']) \\\n",
    "                .agg({'Weather Value': np.mean}).reset_index() # aggregate weather\n",
    "aggregated_sales_data = merged_weather.groupby(by=['Year-Month', 'Item Description']) \\\n",
    "                .agg({'Bottles Sold': np.sum}).reset_index() # aggregate sales\n",
    "\n",
    "# Merge aggregations\n",
    "merged_aggregation = pd.merge(left=aggregated_weather_data, right=aggregated_sales_data, how='inner', on='Year-Month')\n",
    "merged_aggregation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "f3Z-YDVcD9nN"
   },
   "source": [
    "Transform the data set into a convenient form that would enable us to easily compute the lag values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 419
    },
    "colab_type": "code",
    "id": "80dkYCqtD9nO",
    "outputId": "dfff04c7-34a5-4e86-ac9e-123a6f82a8d4"
   },
   "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>Year-Month</th>\n",
       "      <th>Item Description</th>\n",
       "      <th>PRCP</th>\n",
       "      <th>SNOW</th>\n",
       "      <th>TMIN</th>\n",
       "      <th>TMAX</th>\n",
       "      <th>Bottles Sold</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2012-01</td>\n",
       "      <td>1800 Reposado Tequila</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>432</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>297</th>\n",
       "      <td>2012-01</td>\n",
       "      <td>Paramount Triple Sec</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>296</th>\n",
       "      <td>2012-01</td>\n",
       "      <td>Paramount Strawberry Schnapps</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>295</th>\n",
       "      <td>2012-01</td>\n",
       "      <td>Paramount Sloe Gin</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>48</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>294</th>\n",
       "      <td>2012-01</td>\n",
       "      <td>Paramount Peppermint Schnapps</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>192</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53448</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Don Julio 70th Anniversary</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53447</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Domaine de Canton</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>24</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53446</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Disaronno Amaretto</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>104</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53457</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Dr McGillicuddys Vanilla</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>96</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53746</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Yukon Jack</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>144</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>53747 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Year-Month               Item Description  ...       TMAX  Bottles Sold\n",
       "0        2012-01          1800 Reposado Tequila  ...  21.871538           432\n",
       "297      2012-01           Paramount Triple Sec  ...  21.871538            12\n",
       "296      2012-01  Paramount Strawberry Schnapps  ...  21.871538           336\n",
       "295      2012-01             Paramount Sloe Gin  ...  21.871538            48\n",
       "294      2012-01  Paramount Peppermint Schnapps  ...  21.871538           192\n",
       "...          ...                            ...  ...        ...           ...\n",
       "53448    2020-08     Don Julio 70th Anniversary  ...  63.277209            20\n",
       "53447    2020-08              Domaine de Canton  ...  63.277209            24\n",
       "53446    2020-08             Disaronno Amaretto  ...  63.277209           104\n",
       "53457    2020-08       Dr McGillicuddys Vanilla  ...  63.277209            96\n",
       "53746    2020-08                     Yukon Jack  ...  63.277209           144\n",
       "\n",
       "[53747 rows x 7 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "columns = ['Year-Month', 'Item Description', 'PRCP', 'SNOW', 'TMIN', 'TMAX', 'Bottles Sold']\n",
    "data = list()\n",
    "for name, group in merged_aggregation.groupby(by=['Year-Month', 'Item Description']):\n",
    "    row = list()\n",
    "    row.append(name[0])\n",
    "    row.append(name[1]) # append `Year-Month` and `Item Description`\n",
    "    sorted_weather_elements = group[['Weather Element', 'Weather Value']] \\\n",
    "            .sort_values(by='Weather Element')['Weather Value']\n",
    "    [row.append(x) for x in sorted_weather_elements] # append **sorted** weather data\n",
    "    row.append(group[['Bottles Sold']].iloc[0]['Bottles Sold']) # append bottles sold\n",
    "    data.append(row)\n",
    "aggregated_data = pd.DataFrame(data=data, columns=columns).sort_values(by='Year-Month')\n",
    "aggregated_data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "dN5Dk1e7yH2u"
   },
   "source": [
    "Next we will filter out the brands of liquor that rarely get restocked. The default value is set to 100 - a liquor brand must have sold at least 100 total bottles by our store by 2020."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "xC3fdAnzxjLG"
   },
   "outputs": [],
   "source": [
    "minimum_bottles_sold = 100 \n",
    "popular_items = list(aggregated_data.groupby('Item Description') \\\n",
    "                     .agg({'Bottles Sold': np.sum}) \\\n",
    "                     .reset_index() \\\n",
    "                     .query(\"`Bottles Sold` > @minimum_bottles_sold\")['Item Description'])\n",
    "\n",
    "aggregated_data.query(\"`Item Description` in @popular_items\", inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ZKgUcAdZD9nS"
   },
   "source": [
    "Compute the lag values for the weather and sales columns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "H2_cT9tVD9nS"
   },
   "outputs": [],
   "source": [
    "# Define aggregation functions\n",
    "weather_aggregation_functions = {\n",
    "    'PRCP': np.mean,\n",
    "    'SNOW': np.mean,\n",
    "    'TMIN': np.mean,\n",
    "    'TMAX': np.mean\n",
    "}\n",
    "\n",
    "# ----\n",
    "# Start computing lag values for the weather\n",
    "\n",
    "weather_lag_1 = aggregated_data.groupby(by='Year-Month') \\\n",
    "    .agg(weather_aggregation_functions) \\\n",
    "    .sort_values(by='Year-Month') \\\n",
    "    .shift(periods=1,\n",
    "           fill_value=np.nan)\n",
    "weather_lag_1.columns = ['PRCP_Lag1', 'SNOW_Lag1', 'TMIN_Lag1', 'TMAX_Lag1']\n",
    "\n",
    "weather_lag_12 = aggregated_data.groupby(by='Year-Month') \\\n",
    "    .agg(weather_aggregation_functions) \\\n",
    "    .sort_values(by='Year-Month') \\\n",
    "    .shift(periods=12,\n",
    "           fill_value=np.nan)\n",
    "    \n",
    "weather_lag_12.columns = ['PRCP_Lag12', 'SNOW_Lag12', 'TMIN_Lag12', 'TMAX_Lag12']\n",
    "\n",
    "# ----\n",
    "# Start computing lag values for the sales\n",
    "# Help from: https://stackoverflow.com/questions/23664877/pandas-equivalent-of-oracle-lead-lag-function\n",
    "\n",
    "sales_lag_1 = aggregated_data.sort_values(by=['Year-Month']) \\\n",
    "      .groupby(by = ['Item Description'])['Bottles Sold'] \\\n",
    "      .shift(periods=1, fill_value=0)\n",
    "\n",
    "sales_lag_2 = aggregated_data.sort_values(by=['Year-Month']) \\\n",
    "      .groupby(by = ['Item Description'])['Bottles Sold'] \\\n",
    "      .shift(periods=2, fill_value=0)\n",
    "sales_lag_2.columns = ['Bottles Sold (Lag_2)']\n",
    "\n",
    "sales_lag_3 = aggregated_data.sort_values(by=['Year-Month']) \\\n",
    "      .groupby(by = ['Item Description'])['Bottles Sold'] \\\n",
    "      .shift(periods=3, fill_value=0)\n",
    "sales_lag_3.columns = ['Bottles Sold (Lag_3)']\n",
    "\n",
    "sales_lag_12 = aggregated_data.sort_values(by=['Year-Month']) \\\n",
    "      .groupby(by = ['Item Description'])['Bottles Sold'] \\\n",
    "      .shift(periods=12, fill_value=0)\n",
    "sales_lag_12.columns = ['Bottles Sold (Lag_12)']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "_1R2CMcdD9nX"
   },
   "source": [
    "Once we have our lag values calculated, we can merge them in one dataframe."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 711
    },
    "colab_type": "code",
    "id": "21bmL2guD9nY",
    "outputId": "7dc1003f-d40d-4435-d944-4398c3209632"
   },
   "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>Year-Month</th>\n",
       "      <th>Item Description</th>\n",
       "      <th>PRCP</th>\n",
       "      <th>SNOW</th>\n",
       "      <th>TMIN</th>\n",
       "      <th>TMAX</th>\n",
       "      <th>Bottles Sold</th>\n",
       "      <th>Bottles Sold (Lag_1)</th>\n",
       "      <th>Bottles Sold (Lag_2)</th>\n",
       "      <th>Bottles Sold (Lag_3)</th>\n",
       "      <th>Bottles Sold (Lag_12)</th>\n",
       "      <th>PRCP_Lag1</th>\n",
       "      <th>SNOW_Lag1</th>\n",
       "      <th>TMIN_Lag1</th>\n",
       "      <th>TMAX_Lag1</th>\n",
       "      <th>PRCP_Lag12</th>\n",
       "      <th>SNOW_Lag12</th>\n",
       "      <th>TMIN_Lag12</th>\n",
       "      <th>TMAX_Lag12</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2012-01</td>\n",
       "      <td>1800 Reposado Tequila</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>432</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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>2012-01</td>\n",
       "      <td>Paramount Triple Sec</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>12</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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>2012-01</td>\n",
       "      <td>Paramount Strawberry Schnapps</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>336</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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>2012-01</td>\n",
       "      <td>Paramount Sloe Gin</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>48</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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>2012-01</td>\n",
       "      <td>Paramount Peppermint Schnapps</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>192</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51020</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Don Julio 70th Anniversary</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>20</td>\n",
       "      <td>72</td>\n",
       "      <td>24</td>\n",
       "      <td>48</td>\n",
       "      <td>72</td>\n",
       "      <td>2.16398</td>\n",
       "      <td>0.0</td>\n",
       "      <td>88.220987</td>\n",
       "      <td>68.431349</td>\n",
       "      <td>6.904757</td>\n",
       "      <td>0.0</td>\n",
       "      <td>83.03419</td>\n",
       "      <td>63.73083</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51021</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Domaine de Canton</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>24</td>\n",
       "      <td>24</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>2.16398</td>\n",
       "      <td>0.0</td>\n",
       "      <td>88.220987</td>\n",
       "      <td>68.431349</td>\n",
       "      <td>6.904757</td>\n",
       "      <td>0.0</td>\n",
       "      <td>83.03419</td>\n",
       "      <td>63.73083</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51022</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Disaronno Amaretto</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>104</td>\n",
       "      <td>48</td>\n",
       "      <td>96</td>\n",
       "      <td>48</td>\n",
       "      <td>192</td>\n",
       "      <td>2.16398</td>\n",
       "      <td>0.0</td>\n",
       "      <td>88.220987</td>\n",
       "      <td>68.431349</td>\n",
       "      <td>6.904757</td>\n",
       "      <td>0.0</td>\n",
       "      <td>83.03419</td>\n",
       "      <td>63.73083</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51023</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Dr McGillicuddys Vanilla</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>96</td>\n",
       "      <td>48</td>\n",
       "      <td>96</td>\n",
       "      <td>48</td>\n",
       "      <td>96</td>\n",
       "      <td>2.16398</td>\n",
       "      <td>0.0</td>\n",
       "      <td>88.220987</td>\n",
       "      <td>68.431349</td>\n",
       "      <td>6.904757</td>\n",
       "      <td>0.0</td>\n",
       "      <td>83.03419</td>\n",
       "      <td>63.73083</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51024</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Yukon Jack</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>144</td>\n",
       "      <td>144</td>\n",
       "      <td>144</td>\n",
       "      <td>144</td>\n",
       "      <td>144</td>\n",
       "      <td>2.16398</td>\n",
       "      <td>0.0</td>\n",
       "      <td>88.220987</td>\n",
       "      <td>68.431349</td>\n",
       "      <td>6.904757</td>\n",
       "      <td>0.0</td>\n",
       "      <td>83.03419</td>\n",
       "      <td>63.73083</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>51025 rows × 19 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Year-Month               Item Description  ...  TMIN_Lag12  TMAX_Lag12\n",
       "0        2012-01          1800 Reposado Tequila  ...         NaN         NaN\n",
       "1        2012-01           Paramount Triple Sec  ...         NaN         NaN\n",
       "2        2012-01  Paramount Strawberry Schnapps  ...         NaN         NaN\n",
       "3        2012-01             Paramount Sloe Gin  ...         NaN         NaN\n",
       "4        2012-01  Paramount Peppermint Schnapps  ...         NaN         NaN\n",
       "...          ...                            ...  ...         ...         ...\n",
       "51020    2020-08     Don Julio 70th Anniversary  ...    83.03419    63.73083\n",
       "51021    2020-08              Domaine de Canton  ...    83.03419    63.73083\n",
       "51022    2020-08             Disaronno Amaretto  ...    83.03419    63.73083\n",
       "51023    2020-08       Dr McGillicuddys Vanilla  ...    83.03419    63.73083\n",
       "51024    2020-08                     Yukon Jack  ...    83.03419    63.73083\n",
       "\n",
       "[51025 rows x 19 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "weather_dataframe = weather_lag_1.join(weather_lag_12).reset_index()\n",
    "aggregated_data['Bottles Sold (Lag_1)'] = sales_lag_1\n",
    "aggregated_data['Bottles Sold (Lag_2)'] = sales_lag_2\n",
    "aggregated_data['Bottles Sold (Lag_3)'] = sales_lag_3\n",
    "aggregated_data['Bottles Sold (Lag_12)'] = sales_lag_12\n",
    "features_dataframe = pd.merge(left=aggregated_data, right=weather_dataframe, how='inner', on='Year-Month')\n",
    "features_dataframe"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "wOCH8GCvD9nc"
   },
   "source": [
    "Up to this point, we have our data in the proper format and we are *almost* ready to start training our machine learning model. Because there are no historical sales for the 2012 data, we will drop it and use the data from 2013 onward to train our model. Lastly, we will take out the month from `Year-Month` as a separate column."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 711
    },
    "colab_type": "code",
    "id": "FdRUmMJ0D9nd",
    "outputId": "f8734d31-ba2f-4ff7-81a8-9d883a34dbe9"
   },
   "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>Year-Month</th>\n",
       "      <th>Item Description</th>\n",
       "      <th>PRCP</th>\n",
       "      <th>SNOW</th>\n",
       "      <th>TMIN</th>\n",
       "      <th>TMAX</th>\n",
       "      <th>Bottles Sold</th>\n",
       "      <th>Bottles Sold (Lag_1)</th>\n",
       "      <th>Bottles Sold (Lag_2)</th>\n",
       "      <th>Bottles Sold (Lag_3)</th>\n",
       "      <th>Bottles Sold (Lag_12)</th>\n",
       "      <th>PRCP_Lag1</th>\n",
       "      <th>SNOW_Lag1</th>\n",
       "      <th>TMIN_Lag1</th>\n",
       "      <th>TMAX_Lag1</th>\n",
       "      <th>PRCP_Lag12</th>\n",
       "      <th>SNOW_Lag12</th>\n",
       "      <th>TMIN_Lag12</th>\n",
       "      <th>TMAX_Lag12</th>\n",
       "      <th>Month</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>5072</th>\n",
       "      <td>2013-01</td>\n",
       "      <td>Paramount Strawberry Schnapps</td>\n",
       "      <td>0.214424</td>\n",
       "      <td>0.969517</td>\n",
       "      <td>42.133665</td>\n",
       "      <td>20.302275</td>\n",
       "      <td>96</td>\n",
       "      <td>240</td>\n",
       "      <td>288</td>\n",
       "      <td>240</td>\n",
       "      <td>336</td>\n",
       "      <td>0.462702</td>\n",
       "      <td>12.110866</td>\n",
       "      <td>41.715374</td>\n",
       "      <td>22.505595</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5073</th>\n",
       "      <td>2013-01</td>\n",
       "      <td>Paramount Creme De Cacao/white</td>\n",
       "      <td>0.214424</td>\n",
       "      <td>0.969517</td>\n",
       "      <td>42.133665</td>\n",
       "      <td>20.302275</td>\n",
       "      <td>96</td>\n",
       "      <td>96</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>0.462702</td>\n",
       "      <td>12.110866</td>\n",
       "      <td>41.715374</td>\n",
       "      <td>22.505595</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5074</th>\n",
       "      <td>2013-01</td>\n",
       "      <td>Paramount Creme De Menthe/green</td>\n",
       "      <td>0.214424</td>\n",
       "      <td>0.969517</td>\n",
       "      <td>42.133665</td>\n",
       "      <td>20.302275</td>\n",
       "      <td>48</td>\n",
       "      <td>96</td>\n",
       "      <td>48</td>\n",
       "      <td>96</td>\n",
       "      <td>48</td>\n",
       "      <td>0.462702</td>\n",
       "      <td>12.110866</td>\n",
       "      <td>41.715374</td>\n",
       "      <td>22.505595</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5075</th>\n",
       "      <td>2013-01</td>\n",
       "      <td>Paramount Gin</td>\n",
       "      <td>0.214424</td>\n",
       "      <td>0.969517</td>\n",
       "      <td>42.133665</td>\n",
       "      <td>20.302275</td>\n",
       "      <td>240</td>\n",
       "      <td>192</td>\n",
       "      <td>288</td>\n",
       "      <td>288</td>\n",
       "      <td>240</td>\n",
       "      <td>0.462702</td>\n",
       "      <td>12.110866</td>\n",
       "      <td>41.715374</td>\n",
       "      <td>22.505595</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5076</th>\n",
       "      <td>2013-01</td>\n",
       "      <td>Paramount Gold Rum</td>\n",
       "      <td>0.214424</td>\n",
       "      <td>0.969517</td>\n",
       "      <td>42.133665</td>\n",
       "      <td>20.302275</td>\n",
       "      <td>8</td>\n",
       "      <td>52</td>\n",
       "      <td>20</td>\n",
       "      <td>12</td>\n",
       "      <td>16</td>\n",
       "      <td>0.462702</td>\n",
       "      <td>12.110866</td>\n",
       "      <td>41.715374</td>\n",
       "      <td>22.505595</td>\n",
       "      <td>0.353469</td>\n",
       "      <td>7.757919</td>\n",
       "      <td>39.626923</td>\n",
       "      <td>21.871538</td>\n",
       "      <td>1</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>51020</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Don Julio 70th Anniversary</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>20</td>\n",
       "      <td>72</td>\n",
       "      <td>24</td>\n",
       "      <td>48</td>\n",
       "      <td>72</td>\n",
       "      <td>2.163980</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>88.220987</td>\n",
       "      <td>68.431349</td>\n",
       "      <td>6.904757</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>83.034190</td>\n",
       "      <td>63.730830</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51021</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Domaine de Canton</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>24</td>\n",
       "      <td>24</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>48</td>\n",
       "      <td>2.163980</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>88.220987</td>\n",
       "      <td>68.431349</td>\n",
       "      <td>6.904757</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>83.034190</td>\n",
       "      <td>63.730830</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51022</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Disaronno Amaretto</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>104</td>\n",
       "      <td>48</td>\n",
       "      <td>96</td>\n",
       "      <td>48</td>\n",
       "      <td>192</td>\n",
       "      <td>2.163980</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>88.220987</td>\n",
       "      <td>68.431349</td>\n",
       "      <td>6.904757</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>83.034190</td>\n",
       "      <td>63.730830</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51023</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Dr McGillicuddys Vanilla</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>96</td>\n",
       "      <td>48</td>\n",
       "      <td>96</td>\n",
       "      <td>48</td>\n",
       "      <td>96</td>\n",
       "      <td>2.163980</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>88.220987</td>\n",
       "      <td>68.431349</td>\n",
       "      <td>6.904757</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>83.034190</td>\n",
       "      <td>63.730830</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51024</th>\n",
       "      <td>2020-08</td>\n",
       "      <td>Yukon Jack</td>\n",
       "      <td>0.116451</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>86.771318</td>\n",
       "      <td>63.277209</td>\n",
       "      <td>144</td>\n",
       "      <td>144</td>\n",
       "      <td>144</td>\n",
       "      <td>144</td>\n",
       "      <td>144</td>\n",
       "      <td>2.163980</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>88.220987</td>\n",
       "      <td>68.431349</td>\n",
       "      <td>6.904757</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>83.034190</td>\n",
       "      <td>63.730830</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>45953 rows × 20 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Year-Month                 Item Description  ...  TMAX_Lag12  Month\n",
       "5072     2013-01    Paramount Strawberry Schnapps  ...   21.871538      1\n",
       "5073     2013-01   Paramount Creme De Cacao/white  ...   21.871538      1\n",
       "5074     2013-01  Paramount Creme De Menthe/green  ...   21.871538      1\n",
       "5075     2013-01                    Paramount Gin  ...   21.871538      1\n",
       "5076     2013-01               Paramount Gold Rum  ...   21.871538      1\n",
       "...          ...                              ...  ...         ...    ...\n",
       "51020    2020-08       Don Julio 70th Anniversary  ...   63.730830      8\n",
       "51021    2020-08                Domaine de Canton  ...   63.730830      8\n",
       "51022    2020-08               Disaronno Amaretto  ...   63.730830      8\n",
       "51023    2020-08         Dr McGillicuddys Vanilla  ...   63.730830      8\n",
       "51024    2020-08                       Yukon Jack  ...   63.730830      8\n",
       "\n",
       "[45953 rows x 20 columns]"
      ]
     },
     "execution_count": 17,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "months = [int(row[-2:]) for row in features_dataframe['Year-Month']]\n",
    "features_dataframe['Month'] = months\n",
    "features_dataframe.query(expr=\"`Year-Month` >= \\\"2013-01\\\"\", inplace=True)\n",
    "features_dataframe"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "VlXzMwwcEeEk"
   },
   "source": [
    "### Methodology<a name='predicting'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "S_3v1wB4J8SH"
   },
   "source": [
    "At this point, we have our data prepared for modelling. In this section we will construct a simple linear regression model on our data. We will use all attributes (except `Year-Month` which serves as filtering tool) to predict our monthly sales. We will use the R-squared value as an indicator of how good our model is.\n",
    "\n",
    "In the example laid out here, we are trying to predict the sales for each month in 2020. For that reason, we will train our model on data up to January 2020. In general, we want to use as much data from the past and up to the period for which we are trying to predict.\n",
    "\n",
    "We will try two more models, Ridge and Lasso regression. We will compare their R-squared values, and determine which is best."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 153
    },
    "colab_type": "code",
    "id": "Ix1Kr7h6AZ9u",
    "outputId": "e3e92a54-0b4f-4c1d-9a38-0cfdf45c5f82"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Linear regression: 0.8731054228476386\t Lasso regression: 0.8737874268271347\t Ridge regression: 0.873105455078551\n",
      "Linear regression: 0.8461537308032105\t Lasso regression: 0.8462549159353274\t Ridge regression: 0.8461537365657577\n",
      "Linear regression: 0.7181275217149209\t Lasso regression: 0.7180643042943\t Ridge regression: 0.7181275231264839\n",
      "Linear regression: -0.6544073075540742\t Lasso regression: -0.6518343312651926\t Ridge regression: -0.6544071997091796\n",
      "Linear regression: -1.3093815195168879\t Lasso regression: -1.3200278681155315\t Ridge regression: -1.3093817143945046\n",
      "Linear regression: 0.626885874873285\t Lasso regression: 0.6269548098791404\t Ridge regression: 0.6268858779986898\n",
      "Linear regression: 0.7392663251562485\t Lasso regression: 0.7392483779908186\t Ridge regression: 0.7392663296267408\n",
      "Linear regression: 0.7848443714557752\t Lasso regression: 0.7848898611347741\t Ridge regression: 0.7848443721248666\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import LinearRegression, Ridge, Lasso\n",
    "\n",
    "descriptive_columns = list(features_dataframe.columns)\n",
    "label = 'Bottles Sold'\n",
    "descriptive_columns.remove('Year-Month')\n",
    "descriptive_columns.remove('Item Description')\n",
    "descriptive_columns.remove(label)\n",
    "months_2020 = ['2020-01', '2020-02', '2020-03', '2020-04', '2020-05', '2020-06', '2020-07', '2020-08']\n",
    "\n",
    "training_data = features_dataframe.query(\"`Year-Month` <= \\\"2019-12\\\"\")[descriptive_columns]\n",
    "training_targets = features_dataframe.query(\"`Year-Month` <= \\\"2019-12\\\"\")[label]\n",
    "\n",
    "for month_2020 in months_2020:\n",
    "  lr = LinearRegression().fit(X=training_data, y=training_targets)\n",
    "  lasso = Lasso().fit(X=training_data, y=training_targets)\n",
    "  ridge = Ridge().fit(X=training_data, y=training_targets)\n",
    "\n",
    "  test_data = features_dataframe.query(\"`Year-Month` == @month_2020\")[descriptive_columns]\n",
    "  test_targets = features_dataframe.query(\"`Year-Month` == @month_2020\")[label]\n",
    "\n",
    "  lr_r2 =lr.score(X=test_data, y=test_targets)\n",
    "  lr_lasso = lasso.score(X=test_data, y=test_targets)\n",
    "  lr_ridge = ridge.score(X=test_data, y=test_targets)\n",
    "\n",
    "  print (\"Linear regression: {}\\t Lasso regression: {}\\t Ridge regression: {}\".format(lr_r2, lr_lasso, lr_ridge))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "M_FAoWNLAtQc"
   },
   "source": [
    "In general, we would have expected the R-squared value to start high and then continously decrease as the months progress. This is because the sales are highly correlated with their immediate previous values, and because our model would not be trained on data immediately preceeding it, therefore, it should perform worse.\n",
    "\n",
    "That is true for our case here, but for April and May 2020 the R-squared values deviate substantially. This is of course because of the COVID-19 pandemic. The plot below plots the monthly sales for the entire data set.\n",
    "\n",
    "We clearly see the steep drop down at the right end of the plot corresponding to the COVID-19 period. This kind of deviation could not have been predicted with any machine learning model, due to the fact that it is cause by an inpredictable \"higher-power\". Literally no one in the world expected something like that, so we would not hold it against simple regresion models for not being able to predict it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 351
    },
    "colab_type": "code",
    "id": "VfWJeWr4BQKf",
    "outputId": "7e8a57ea-48d2-40cd-9efe-6ac4305057dc"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f8604b6d0f0>"
      ]
     },
     "execution_count": 29,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "features_dataframe.groupby(by=['Year-Month']).agg({'Bottles Sold': np.sum}).plot(figsize=(10, 5))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "VQy6xytxDV4E"
   },
   "source": [
    "All three models have similar predictive performance. Therefore, we would recommend using the linear regression model, for simplicity."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "im1beo5eECeQ"
   },
   "source": [
    "## Results and discussion<a name='results'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ajjHqyxmEJ3-"
   },
   "source": [
    "Our analysis show how relatively easy we can build a machine learning model to help us with our decisions for restocking our store inventory.\n",
    "\n",
    "First, we identified our business problem. We tried to predict our monthly sales by using historical sales and weather data. The historical sales data is our own, and there would be no need to query an external source for it; a store has records of its sales. The weather data was freely available for download from the [NOAA's Global Historical Climatology Network (GHCN)](https://console.cloud.google.com/marketplace/details/noaa-public/ghcn-d).\n",
    "\n",
    "GHCN maintains a vastly complex database of weather information, and we needed quite some time to be able find the relevant data for us. We had to find the weather stations that are in the vicinity of the store, which are still operating in 2020 (some of stations had data going back to the 1700s!) and are recording the weather information that we need, that is temperature, snowfall and precipitation.\n",
    "\n",
    "Once we found the station closest to us, we queried its data and merged it with our historical sales to obtain our final data set that is ready for modeling.\n",
    "\n",
    "Then, we apply three regression models: linear, Ridge and Lasso. All models exhibit similar predictive performance, and we recommend using the linear regression for its simplicity.\n",
    "\n",
    "The models perform rather good, with their R-squared values ranging around 0.88.\n",
    "\n",
    "As everything in the world, the model's performance suffered substantially when trying to predict sales for the period of the COVID-19 pandemic in the U.S. (April and May 2020). The sales for this period drop very low, something that has not been observed in our data set and could not have been modeled by any other machine model."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "FnLa0Vj8GpmT"
   },
   "source": [
    "## Conclusion<a name='conclusion'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "PtJFFeCzGtdh"
   },
   "source": [
    "To conclude this work, I will say a few words about the application of this project.\n",
    "\n",
    "Predict store sales based on historical data is very common and known retail problem and there are already some very known solutions to it. In general, a store owner could just export his data to a machine learning expert, and have him return his predictions. He would then use this data to make decision on restocking his inventory.\n",
    "\n",
    "This notebook served more as an education tool for me, than as a real-world project that can be applied. I am not saying that this does not have to potential or that it can not be implemented in an actual case study, but I am having a hard time believing that a store owner from Iowa would trust my work.\n",
    "\n",
    "Maybe I will show it to some local stores in my country (Macedonia), and see if there is any interest to see this work being applied here."
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "Capstone_Project.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}