feat: data-pipelines

implemented data loading pipelines, data cleaning and basic EDA along with TFIDF vectorization

notebooks/EDA.ipynb

@@ -0,0 +1,658 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## __Exploratory Data Analysis__"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 53,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "## importing libraries\n",
+    "\n",
+    "import numpy as numpy\n",
+    "import pandas as pd\n",
+    "import matplotlib.pyplot as plt\n",
+    "import seaborn as sns\n",
+    "\n",
+    "import sys\n",
+    "import os\n",
+    "\n",
+    "import re\n",
+    "import nltk\n",
+    "from nltk.corpus import stopwords\n",
+    "from nltk.stem import PorterStemmer\n",
+    "from nltk.stem import WordNetLemmatizer"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>row</th>\n",
+       "      <th>col</th>\n",
+       "      <th>latitude</th>\n",
+       "      <th>longitude</th>\n",
+       "      <th>Map Data</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>18.89433</td>\n",
+       "      <td>72.784597</td>\n",
+       "      <td>NaN</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>0</td>\n",
+       "      <td>1</td>\n",
+       "      <td>18.89433</td>\n",
+       "      <td>72.794102</td>\n",
+       "      <td>Prongs Reef is a Natural;</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>0</td>\n",
+       "      <td>2</td>\n",
+       "      <td>18.89433</td>\n",
+       "      <td>72.803607</td>\n",
+       "      <td>United Services Club Golf Course is a Leisure ...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>0</td>\n",
+       "      <td>3</td>\n",
+       "      <td>18.89433</td>\n",
+       "      <td>72.813112</td>\n",
+       "      <td>Indian Meterological Department is a Commercia...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>0</td>\n",
+       "      <td>4</td>\n",
+       "      <td>18.89433</td>\n",
+       "      <td>72.822617</td>\n",
+       "      <td>NaN</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "   row  col  latitude  longitude  \\\n",
+       "0    0    0  18.89433  72.784597   \n",
+       "1    0    1  18.89433  72.794102   \n",
+       "2    0    2  18.89433  72.803607   \n",
+       "3    0    3  18.89433  72.813112   \n",
+       "4    0    4  18.89433  72.822617   \n",
+       "\n",
+       "                                            Map Data  \n",
+       "0                                                NaN  \n",
+       "1                         Prongs Reef is a Natural;   \n",
+       "2  United Services Club Golf Course is a Leisure ...  \n",
+       "3  Indian Meterological Department is a Commercia...  \n",
+       "4                                                NaN  "
+      ]
+     },
+     "execution_count": 6,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "data_folder = os.path.join(os.path.dirname(os.getcwd()), 'data')\n",
+    "data_file = os.path.join(data_folder, 'MMR_DATA.csv')\n",
+    "df = pd.read_csv(data_file)\n",
+    "df.head()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>row</th>\n",
+       "      <th>col</th>\n",
+       "      <th>latitude</th>\n",
+       "      <th>longitude</th>\n",
+       "      <th>Map Data</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>18.89433</td>\n",
+       "      <td>72.784597</td>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>0</td>\n",
+       "      <td>1</td>\n",
+       "      <td>18.89433</td>\n",
+       "      <td>72.794102</td>\n",
+       "      <td>Prongs Reef is a Natural;</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>0</td>\n",
+       "      <td>2</td>\n",
+       "      <td>18.89433</td>\n",
+       "      <td>72.803607</td>\n",
+       "      <td>United Services Club Golf Course is a Leisure ...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>0</td>\n",
+       "      <td>3</td>\n",
+       "      <td>18.89433</td>\n",
+       "      <td>72.813112</td>\n",
+       "      <td>Indian Meterological Department is a Commercia...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>0</td>\n",
+       "      <td>4</td>\n",
+       "      <td>18.89433</td>\n",
+       "      <td>72.822617</td>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "   row  col  latitude  longitude  \\\n",
+       "0    0    0  18.89433  72.784597   \n",
+       "1    0    1  18.89433  72.794102   \n",
+       "2    0    2  18.89433  72.803607   \n",
+       "3    0    3  18.89433  72.813112   \n",
+       "4    0    4  18.89433  72.822617   \n",
+       "\n",
+       "                                            Map Data  \n",
+       "0                                                     \n",
+       "1                         Prongs Reef is a Natural;   \n",
+       "2  United Services Club Golf Course is a Leisure ...  \n",
+       "3  Indian Meterological Department is a Commercia...  \n",
+       "4                                                     "
+      ]
+     },
+     "execution_count": 7,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "## filling the NaN values in the Map Data Column with empty string\n",
+    "\n",
+    "df['Map Data'] = df['Map Data'].fillna('')\n",
+    "df.head()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "1225"
+      ]
+     },
+     "execution_count": 20,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "len(df)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 42,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df_len_explore = df.copy()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 43,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>Map Data</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>Prongs Reef is a Natural;</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>United Services Club Golf Course is a Leisure ...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>Indian Meterological Department is a Commercia...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                                            Map Data\n",
+       "0                                                   \n",
+       "1                         Prongs Reef is a Natural; \n",
+       "2  United Services Club Golf Course is a Leisure ...\n",
+       "3  Indian Meterological Department is a Commercia...\n",
+       "4                                                   "
+      ]
+     },
+     "execution_count": 43,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "## dropping the columns that are not needed for the analysis\n",
+    "\n",
+    "df_len_explore = df_len_explore.drop(columns=['row', 'col', 'latitude', 'longitude'])\n",
+    "df_len_explore.head()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 44,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "791"
+      ]
+     },
+     "execution_count": 44,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "## dropping the rows with 0 string length and string length > 5000\n",
+    "\n",
+    "df_len_explore = df_len_explore[df_len_explore['Map Data'].str.len() > 0]\n",
+    "df_len_explore = df_len_explore[df_len_explore['Map Data'].str.len() < 5000]\n",
+    "len(df_len_explore)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 45,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Discarded rows:  434 / 1225\n"
+     ]
+    }
+   ],
+   "source": [
+    "print('Discarded rows: ', len(df) - len(df_len_explore), '/', len(df))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 52,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Mean string length:  834.7509481668774\n",
+      "80th  percentile string length:  1560.0\n"
+     ]
+    }
+   ],
+   "source": [
+    "## mean of the string length\n",
+    "\n",
+    "print('Mean string length: ', df_len_explore['Map Data'].str.len().mean())\n",
+    "print('80th  percentile string length: ', df_len_explore['Map Data'].str.len().quantile(0.8))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 51,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "<matplotlib.lines.Line2D at 0x1c5b27a9790>"
+      ]
+     },
+     "execution_count": 51,
+     "metadata": {},
+     "output_type": "execute_result"
+    },
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "## plotting lenth of strings in the Map Data Column, mean and 75th percentile\n",
+    "\n",
+    "df_len_explore['Map Data'].str.len().plot(kind='hist', bins=100)\n",
+    "plt.axvline(df_len_explore['Map Data'].str.len().mean(), color='red', linestyle='dashed', linewidth=2)\n",
+    "plt.axvline(df_len_explore['Map Data'].str.len().quantile(0.80), color='green', linestyle='dashed', linewidth=2)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Same visualization, post cleaning"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 54,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "[nltk_data] Downloading package stopwords to C:\\Users\\Akhil\n",
+      "[nltk_data]     PC\\AppData\\Roaming\\nltk_data...\n",
+      "[nltk_data]   Package stopwords is already up-to-date!\n",
+      "[nltk_data] Downloading package wordnet to C:\\Users\\Akhil\n",
+      "[nltk_data]     PC\\AppData\\Roaming\\nltk_data...\n",
+      "[nltk_data]   Package wordnet is already up-to-date!\n"
+     ]
+    }
+   ],
+   "source": [
+    "nltk.download('stopwords')\n",
+    "nltk.download('wordnet')\n",
+    "\n",
+    "stop_words = set(stopwords.words('english'))\n",
+    "stemmer = PorterStemmer()\n",
+    "lemmatizer = WordNetLemmatizer()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 55,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "## cleaning the strings, stemming and lemmatizing\n",
+    "\n",
+    "def clean_text(text):\n",
+    "    text = re.sub(r'[^\\w\\s]', '', text)\n",
+    "    text = text.lower()\n",
+    "    text = [word for word in text.split() if word not in stop_words]\n",
+    "    text = [stemmer.stem(word) for word in text] \n",
+    "    text = [lemmatizer.lemmatize(word) for word in text]\n",
+    "    return ' '.join(text)\n",
+    "\n",
+    "df_len_explore['Map Data'] = df_len_explore['Map Data'].apply(clean_text)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 56,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Mean string length:  596.3046776232617\n",
+      "80th  percentile string length:  1114.0\n"
+     ]
+    }
+   ],
+   "source": [
+    "print('Mean string length: ', df_len_explore['Map Data'].str.len().mean())\n",
+    "print('80th  percentile string length: ', df_len_explore['Map Data'].str.len().quantile(0.8))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 57,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "<matplotlib.lines.Line2D at 0x1c5b5e37b00>"
+      ]
+     },
+     "execution_count": 57,
+     "metadata": {},
+     "output_type": "execute_result"
+    },
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "df_len_explore['Map Data'].str.len().plot(kind='hist', bins=100)\n",
+    "plt.axvline(df_len_explore['Map Data'].str.len().mean(), color='red', linestyle='dashed', linewidth=2)\n",
+    "plt.axvline(df_len_explore['Map Data'].str.len().quantile(0.80), color='green', linestyle='dashed', linewidth=2)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 59,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Original data length:                     1225\n",
+      "Number of blanks:                         407\n",
+      "Number of strings with length > 5000:     27\n",
+      "Number of useful rows:                    791\n"
+     ]
+    }
+   ],
+   "source": [
+    "## Final Summary\n",
+    "\n",
+    "print('Original data length:                    ', len(df))\n",
+    "print('Number of blanks:                        ', len(df) - len(df[df['Map Data'].str.len() > 0]))\n",
+    "print('Number of strings with length > 5000:    ', len(df[df['Map Data'].str.len() > 5000]))\n",
+    "print('Number of useful rows:                   ', len(df_len_explore))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 62,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Post cleaning data length:        791\n",
+      "Avg string length:                596.3046776232617\n",
+      "Median string length:             192.0\n",
+      "25th percentile string length:    43.5\n",
+      "80th percentile string length:    1114.0\n"
+     ]
+    }
+   ],
+   "source": [
+    "print('Post cleaning data length:       ', len(df_len_explore))\n",
+    "print('Avg string length:               ', df_len_explore['Map Data'].str.len().mean())\n",
+    "print('Median string length:            ', df_len_explore['Map Data'].str.len().median())\n",
+    "print('25th percentile string length:   ', df_len_explore['Map Data'].str.len().quantile(0.25))   \n",
+    "print('80th percentile string length:   ', df_len_explore['Map Data'].str.len().quantile(0.8))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Highly uneven string length distribution.\n",
+    "- 33% of the total data is useless, ie. blank rows\n",
+    "- 25% of the strings have length less than 45 characters\n",
+    "- 50% of the strings have length less than 200 characters\n",
+    "- 30% of the strings have length between 200 - 1100 characters (huge variation)\n",
+    "- 20% of the strings have length greater than 1100 characters\n",
+    "- about 5% of strings are longer than 5000 characters (wont be used in training)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "## function to clean the given data frame\n",
+    "\n",
+    "def clean_text(text):\n",
+    "    text = re.sub(r'[^\\w\\s]', '', text)\n",
+    "    text = text.lower()\n",
+    "    text = [word for word in text.split() if word not in stop_words]\n",
+    "    text = [stemmer.stem(word) for word in text] \n",
+    "    text = [lemmatizer.lemmatize(word) for word in text]\n",
+    "    return ' '.join(text)\n",
+    "\n",
+    "def clean_data(df):\n",
+    "    df['Map Data'] = df['Map Data'].fillna('')\n",
+    "    df = df[df['Map Data'].str.len() > 0]\n",
+    "    df = df[df['Map Data'].str.len() < 5000]\n",
+    "    df['Map Data'] = df['Map Data'].apply(clean_text)\n",
+    "    return df"
+   ]
+  }
+ ],
+ "metadata": {
+  "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": "3.12.0"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

notebooks/data_loading.ipynb

@@ -0,0 +1,996 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## __Data Pipelines__ \n",
+    "Loading data from OpenStreetMap using overpass API"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 60,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import requests\n",
+    "import pandas as pd\n",
+    "import re\n",
+    "import math\n",
+    "from typing import Tuple, List, Dict"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 97,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def fetch_osm_data(lat: float, lon: float, radius: int) -> List[Dict]:\n",
+    "    overpass_url = \"http://overpass-api.de/api/interpreter\"\n",
+    "    overpass_query = f\"\"\"\n",
+    "    [out:json];\n",
+    "    (\n",
+    "      node[\"name\"](around:{radius},{lat},{lon});\n",
+    "      way[\"name\"](around:{radius},{lat},{lon});\n",
+    "      relation[\"name\"](around:{radius},{lat},{lon});\n",
+    "    );\n",
+    "    out center;\n",
+    "    \"\"\"\n",
+    "    \n",
+    "    response = requests.get(overpass_url, params={'data': overpass_query})\n",
+    "    data = response.json()\n",
+    "    return data['elements']\n",
+    "\n",
+    "def determine_location_type(tags: Dict[str, str]) -> str:\n",
+    "    # Residential\n",
+    "    if 'building' in tags and tags['building'] in ['residential', 'house', 'apartments', 'detached', 'terrace', 'dormitory', 'bungalow']:\n",
+    "        return 'Residential'\n",
+    "    \n",
+    "    # Commercial\n",
+    "    if any(key in tags for key in ['shop', 'office', 'craft']):\n",
+    "        return 'Commercial'\n",
+    "    if 'building' in tags and tags['building'] in ['commercial', 'office', 'retail', 'supermarket', 'kiosk']:\n",
+    "        return 'Commercial'\n",
+    "    \n",
+    "    # Industrial\n",
+    "    if 'building' in tags and tags['building'] in ['industrial', 'warehouse', 'factory', 'manufacture']:\n",
+    "        return 'Industrial'\n",
+    "    if 'industrial' in tags or 'industry' in tags:\n",
+    "        return 'Industrial'\n",
+    "    \n",
+    "    # Educational\n",
+    "    if 'amenity' in tags and tags['amenity'] in ['school', 'university', 'college', 'library', 'kindergarten', 'language_school']:\n",
+    "        return 'Educational'\n",
+    "    \n",
+    "    # Healthcare\n",
+    "    if 'amenity' in tags and tags['amenity'] in ['hospital', 'clinic', 'doctors', 'dentist', 'pharmacy', 'veterinary']:\n",
+    "        return 'Healthcare'\n",
+    "    \n",
+    "    # Food & Drink\n",
+    "    if 'amenity' in tags and tags['amenity'] in ['restaurant', 'cafe', 'bar', 'fast_food', 'pub', 'food_court']:\n",
+    "        return 'Food & Drink'\n",
+    "    \n",
+    "    # Leisure & Entertainment\n",
+    "    if 'leisure' in tags or 'tourism' in tags:\n",
+    "        return 'Leisure & Entertainment'\n",
+    "    if 'amenity' in tags and tags['amenity'] in ['theatre', 'cinema', 'nightclub', 'arts_centre', 'community_centre']:\n",
+    "        return 'Leisure & Entertainment'\n",
+    "    \n",
+    "    # Transportation\n",
+    "    if 'amenity' in tags and tags['amenity'] in ['parking', 'bicycle_parking', 'bus_station', 'ferry_terminal']:\n",
+    "        return 'Transportation'\n",
+    "    if 'highway' in tags or 'railway' in tags or 'aeroway' in tags:\n",
+    "        return 'Transportation'\n",
+    "    \n",
+    "    # Religious\n",
+    "    if 'amenity' in tags and tags['amenity'] in ['place_of_worship', 'monastery']:\n",
+    "        return 'Religious'\n",
+    "    \n",
+    "    # Government & Public Services\n",
+    "    if 'amenity' in tags and tags['amenity'] in ['townhall', 'courthouse', 'police', 'fire_station', 'post_office']:\n",
+    "        return 'Government & Public Services'\n",
+    "    \n",
+    "    # Parks & Recreation\n",
+    "    if 'leisure' in tags and tags['leisure'] in ['park', 'playground', 'sports_centre', 'stadium', 'garden']:\n",
+    "        return 'Parks & Recreation'\n",
+    "    \n",
+    "    # Natural\n",
+    "    if 'natural' in tags:\n",
+    "        return 'Natural'\n",
+    "    \n",
+    "    # Landuse\n",
+    "    if 'landuse' in tags:\n",
+    "        landuse = tags['landuse'].capitalize()\n",
+    "        if landuse in ['Residential', 'Commercial', 'Industrial', 'Retail']:\n",
+    "            return landuse\n",
+    "        else:\n",
+    "            return f'Landuse: {landuse}'\n",
+    "    \n",
+    "    # If no specific category is found, return 'Other'\n",
+    "    return 'Other'\n",
+    "\n",
+    "def parse_osm_data(elements: List[Dict]) -> pd.DataFrame:\n",
+    "    parsed_data = []\n",
+    "    for element in elements:\n",
+    "        tags = element.get('tags', {})\n",
+    "        parsed_element = {\n",
+    "            'ID': f\"{element['type']}_{element['id']}\",\n",
+    "            'Location Name': tags.get('name', ''),\n",
+    "            'Location Type': determine_location_type(tags)\n",
+    "        }\n",
+    "        parsed_data.append(parsed_element)\n",
+    "    if len(parsed_data) == 0:\n",
+    "        return pd.DataFrame(columns=['ID', 'Location Name', 'Location Type'])\n",
+    "    return pd.DataFrame(parsed_data)\n",
+    "\n",
+    "def get_osm_data(lat: float, lon: float, radius: int) -> pd.DataFrame:\n",
+    "    raw_data = fetch_osm_data(lat, lon, radius)\n",
+    "    return parse_osm_data(raw_data)\n",
+    "\n",
+    "def dms_to_decimal(coord_str):\n",
+    "    # Regular expression to match the coordinate format\n",
+    "    pattern = r'(\\d+)°(\\d+)\\'([\\d.]+)\"([NS])\\s*(\\d+)°(\\d+)\\'([\\d.]+)\"([EW])'\n",
+    "    \n",
+    "    match = re.match(pattern, coord_str)\n",
+    "    if not match:\n",
+    "        raise ValueError(\"Invalid coordinate format. Expected format: 19°03'08.6\\\"N 72°54'06.0\\\"E\")\n",
+    "\n",
+    "    lat_deg, lat_min, lat_sec, lat_dir, lon_deg, lon_min, lon_sec, lon_dir = match.groups()\n",
+    "\n",
+    "    # Convert to decimal degrees\n",
+    "    lat = float(lat_deg) + float(lat_min)/60 + float(lat_sec)/3600\n",
+    "    lon = float(lon_deg) + float(lon_min)/60 + float(lon_sec)/3600\n",
+    "\n",
+    "    # Adjust sign based on direction\n",
+    "    if lat_dir == 'S':\n",
+    "        lat = -lat\n",
+    "    if lon_dir == 'W':\n",
+    "        lon = -lon\n",
+    "\n",
+    "    return lat, lon"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 91,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Latitude: 19.015805555555556\n",
+      "Longitude: 72.89944444444446\n"
+     ]
+    }
+   ],
+   "source": [
+    "coord_str = '19°00\\'56.9\"N 72°53\\'58.0\"E'\n",
+    "radius_meters = 1000\n",
+    "try:\n",
+    "    latitude, longitude = dms_to_decimal(coord_str)\n",
+    "    print(f\"Latitude: {latitude}\")\n",
+    "    print(f\"Longitude: {longitude}\")\n",
+    "except ValueError as e:\n",
+    "    print(f\"Error: {e}\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 92,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "result_df = get_osm_data(latitude, longitude, radius_meters)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 93,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>ID</th>\n",
+       "      <th>Location Name</th>\n",
+       "      <th>Location Type</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>node_622002639</td>\n",
+       "      <td>Mahul</td>\n",
+       "      <td>Other</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>node_622005407</td>\n",
+       "      <td>Gowanpada</td>\n",
+       "      <td>Other</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>node_1646222635</td>\n",
+       "      <td>gadakary bus stop</td>\n",
+       "      <td>Transportation</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>node_1646222681</td>\n",
+       "      <td>vishnu nagar bus stop</td>\n",
+       "      <td>Other</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>node_2932495033</td>\n",
+       "      <td>Sree Dutta mandir</td>\n",
+       "      <td>Religious</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>5</th>\n",
+       "      <td>node_11954176622</td>\n",
+       "      <td>Gavhanpada</td>\n",
+       "      <td>Other</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>6</th>\n",
+       "      <td>way_25587616</td>\n",
+       "      <td>Bhikaji Damaji Patil Marg</td>\n",
+       "      <td>Transportation</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>7</th>\n",
+       "      <td>way_122289587</td>\n",
+       "      <td>Mulund - Trombay 220 KV line</td>\n",
+       "      <td>Other</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>8</th>\n",
+       "      <td>way_151783563</td>\n",
+       "      <td>Laxman Umaji Gadkari Marg</td>\n",
+       "      <td>Transportation</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>9</th>\n",
+       "      <td>way_151783570</td>\n",
+       "      <td>Vishnu Nagar Road</td>\n",
+       "      <td>Transportation</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "                 ID                 Location Name   Location Type\n",
+       "0    node_622002639                         Mahul           Other\n",
+       "1    node_622005407                     Gowanpada           Other\n",
+       "2   node_1646222635             gadakary bus stop  Transportation\n",
+       "3   node_1646222681         vishnu nagar bus stop           Other\n",
+       "4   node_2932495033             Sree Dutta mandir       Religious\n",
+       "5  node_11954176622                    Gavhanpada           Other\n",
+       "6      way_25587616     Bhikaji Damaji Patil Marg  Transportation\n",
+       "7     way_122289587  Mulund - Trombay 220 KV line           Other\n",
+       "8     way_151783563     Laxman Umaji Gadkari Marg  Transportation\n",
+       "9     way_151783570             Vishnu Nagar Road  Transportation"
+      ]
+     },
+     "execution_count": 93,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "result_df.head(10)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 94,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>ID</th>\n",
+       "      <th>Location Name</th>\n",
+       "      <th>Location Type</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>11</th>\n",
+       "      <td>way_430012316</td>\n",
+       "      <td>track</td>\n",
+       "      <td>Residential</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>12</th>\n",
+       "      <td>way_430012318</td>\n",
+       "      <td>Mumbai Refinery Mahul</td>\n",
+       "      <td>Industrial</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>13</th>\n",
+       "      <td>way_430012320</td>\n",
+       "      <td>Mumbai Refinery</td>\n",
+       "      <td>Industrial</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "               ID          Location Name Location Type\n",
+       "11  way_430012316                  track   Residential\n",
+       "12  way_430012318  Mumbai Refinery Mahul    Industrial\n",
+       "13  way_430012320        Mumbai Refinery    Industrial"
+      ]
+     },
+     "execution_count": 94,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "labelled_df = result_df[result_df['Location Type'] != 'Other']\n",
+    "labelled_df = labelled_df[labelled_df['Location Type'] != 'Religious']\n",
+    "labelled_df = labelled_df[labelled_df['Location Type'] != 'Transportation']\n",
+    "labelled_df.head(10)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 95,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>Location Name</th>\n",
+       "      <th>Location Type</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>track</td>\n",
+       "      <td>Residential</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>Mumbai Refinery Mahul</td>\n",
+       "      <td>Industrial</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>Mumbai Refinery</td>\n",
+       "      <td>Industrial</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "           Location Name Location Type\n",
+       "0                  track   Residential\n",
+       "1  Mumbai Refinery Mahul    Industrial\n",
+       "2        Mumbai Refinery    Industrial"
+      ]
+     },
+     "execution_count": 95,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "## removing duplicates\n",
+    "\n",
+    "loc_types = []\n",
+    "for row in labelled_df.iterrows():\n",
+    "    loc_type = (row[1]['Location Name'], row[1]['Location Type'])\n",
+    "    if loc_type not in loc_types:\n",
+    "        loc_types.append(loc_type)\n",
+    "\n",
+    "labelled_df = pd.DataFrame(loc_types, columns=['Location Name', 'Location Type'])\n",
+    "labelled_df.head(20)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 58,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "row_of_dataset = ''\n",
+    "\n",
+    "for row in labelled_df.iterrows():\n",
+    "    row_text = row[1]['Location Name'] + ' is a ' + row[1]['Location Type']\n",
+    "    row_of_dataset += row_text + ', '"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 59,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "'Oswal Company Trees is a Natural, Newspaper stall is a Commercial, Shiv Polyclinic and Nursing Home is a Healthcare, राजपूत मेडिकल is a Healthcare, Bhabha Atomic Research Centre - BARC is a Industrial, BPCL Sports Club is a Leisure & Entertainment, New Bharat Nagar, Banjara tanda, Hasina Nagar is a Residential, Old Bharat Nagar is a Residential, Rashtriya Chemicals & Fertilizers is a Industrial, Koyna Colony is a Residential, D is a Residential, A-2 is a Residential, flip card is a Commercial, track is a Residential, Mumbai Refinery Mahul is a Industrial, Mumbai Refinery is a Industrial, Trombay Thermal Power Station is a Industrial, Vitta Sanchay Society is a Residential, E is a Residential, Acharya Sharad Narayan Udyan is a Leisure & Entertainment, bmc park is a Leisure & Entertainment, Mysore Colony Central Garden is a Leisure & Entertainment, BMC owned trees is a Natural, BMC PARK is a Leisure & Entertainment, Mysore colony eastern park is a Leisure & Entertainment, Trees owned by RCF is a Natural, Mysore Colony trees is a Natural, NAVAL KG School, TS MAHUL is a Educational, '"
+      ]
+     },
+     "execution_count": 59,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "row_of_dataset"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "This is one row of the  dataset, now writing a function to extract all these rows from a given large map area"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 61,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "## input point is at the bottom left of the map\n",
+    "\n",
+    "def calculate_distant_points(lat: float, lon: float, distance: float) -> tuple:\n",
+    "    # Earth's radius in meters\n",
+    "    R = 6371000\n",
+    "\n",
+    "    # Convert latitude and longitude to radians\n",
+    "    lat_rad = math.radians(lat)\n",
+    "    lon_rad = math.radians(lon)\n",
+    "\n",
+    "    # Calculate the point with the same latitude (moving east-west)\n",
+    "    delta_lon = distance / (R * math.cos(lat_rad))\n",
+    "    lon1 = lon + math.degrees(delta_lon)\n",
+    "    \n",
+    "    # Calculate the point with the same longitude (moving north-south)\n",
+    "    delta_lat = distance / R\n",
+    "    lat2 = lat + math.degrees(delta_lat)\n",
+    "\n",
+    "    return ((lat, lon1), (lat2, lon))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 66,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Original point: (40.7128, -74.006)\n",
+      "Point 1000m east: (40.712800, -73.709386)\n",
+      "Point 1000m north: (40.937630, -74.006000)\n"
+     ]
+    }
+   ],
+   "source": [
+    "if __name__ == \"__main__\":\n",
+    "    latitude = 40.7128  # New York City latitude\n",
+    "    longitude = -74.0060  # New York City longitude\n",
+    "    distance = 1000*25  # 1000 meters\n",
+    "\n",
+    "    result = calculate_distant_points(latitude, longitude, distance)\n",
+    "    print(f\"Original point: ({latitude}, {longitude})\")\n",
+    "    print(f\"Point 1000m east: ({result[0][0]:.6f}, {result[0][1]:.6f})\")\n",
+    "    print(f\"Point 1000m north: ({result[1][0]:.6f}, {result[1][1]:.6f})\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 69,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Bottom Left:    (40.7128, -74.006)\n",
+      "Top Left:       (40.93763040147969, -74.006)\n",
+      "Bottom Right:   (40.7128, -73.7093855252233)\n",
+      "Top Right:      (40.93763040147969, -73.7093855252233)\n"
+     ]
+    }
+   ],
+   "source": [
+    "bottom_left_latitude = 40.7128\n",
+    "bottom_left_longitude = -74.0060\n",
+    "\n",
+    "result = calculate_distant_points(bottom_left_latitude, bottom_left_longitude, 1000*25)\n",
+    "\n",
+    "top_left_latitude = result[1][0]\n",
+    "top_left_longitude = result[1][1]\n",
+    "\n",
+    "bottom_right_latitude = result[0][0]\n",
+    "bottom_right_longitude = result[0][1]\n",
+    "\n",
+    "top_right_latitude = top_left_latitude\n",
+    "top_right_longitude = bottom_right_longitude\n",
+    "\n",
+    "print(f\"Bottom Left:    ({bottom_left_latitude}, {bottom_left_longitude})\")\n",
+    "print(f\"Top Left:       ({top_left_latitude}, {top_left_longitude})\")\n",
+    "print(f\"Bottom Right:   ({bottom_right_latitude}, {bottom_right_longitude})\")\n",
+    "print(f\"Top Right:      ({top_right_latitude}, {top_right_longitude})\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 71,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "(0.008993216059187433, 0.01186457899106813)"
+      ]
+     },
+     "execution_count": 71,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "latitude_shift = top_left_latitude - bottom_left_latitude\n",
+    "longitude_shift = bottom_right_longitude - bottom_left_longitude\n",
+    "\n",
+    "latitude_unit = latitude_shift / 25\n",
+    "longitude_unit = longitude_shift / 25\n",
+    "\n",
+    "latitude_unit, longitude_unit"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 73,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "## 2d map grid (0,0) --> bottom left\n",
+    "\n",
+    "def create_map_grid(bottom_left: Tuple[float, float], top_right: Tuple[float, float], rows: int, cols: int) -> List[List[Tuple[float, float]]]:\n",
+    "    grid = []\n",
+    "    lat_unit = (top_right[0] - bottom_left[0]) / rows\n",
+    "    lon_unit = (top_right[1] - bottom_left[1]) / cols\n",
+    "    \n",
+    "    for i in range(rows):\n",
+    "        row = []\n",
+    "        for j in range(cols):\n",
+    "            lat = bottom_left[0] + i * lat_unit\n",
+    "            lon = bottom_left[1] + j * lon_unit\n",
+    "            lat = lat + lat_unit / 2\n",
+    "            lon = lon + lon_unit / 2\n",
+    "            row.append((lat, lon))\n",
+    "        grid.append(row)\n",
+    "    \n",
+    "    return grid"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 79,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "grid = create_map_grid((bottom_left_latitude, bottom_left_longitude), (top_right_latitude, top_right_longitude), 25, 25)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 108,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "grid_dataset = []\n",
+    "for i, row in enumerate(grid):\n",
+    "    for j, point in enumerate(row):\n",
+    "        \n",
+    "        grid_row = {\"row\": i, \"col\": j, \"latitude\": point[0], \"longitude\": point[1]}\n",
+    "        grid_dataset.append(grid_row)\n",
+    "\n",
+    "grid_df = pd.DataFrame(grid_dataset)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 83,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "left_lat = 18.889833\n",
+    "left_lon = 72.779844"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 84,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "res1 =  calculate_distant_points(left_lat, left_lon, 1000*35)\n",
+    "\n",
+    "right_lat = res1[1][0]\n",
+    "right_lon = res1[0][1]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 85,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "grid = create_map_grid((left_lat, left_lon), (right_lat, right_lon), 35, 35)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "grid_dataset = []\n",
+    "for i, row in enumerate(grid):\n",
+    "    for j, point in enumerate(row):\n",
+    "        grid_row = {\"row\": i, \"col\": j, \"latitude\": point[0], \"longitude\": point[1]}\n",
+    "        grid_dataset.append(grid_row)\n",
+    "\n",
+    "grid_df = pd.DataFrame(grid_dataset)\n",
+    "grid_df.head(25)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 106,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "## entire pipeline\n",
+    "\n",
+    "left_lat = 18.889833\n",
+    "left_lon = 72.779844\n",
+    "dist = 35\n",
+    "\n",
+    "res1 =  calculate_distant_points(left_lat, left_lon, 1000*dist)\n",
+    "\n",
+    "right_lat = res1[1][0]\n",
+    "right_lon = res1[0][1]\n",
+    "grid = create_map_grid((left_lat, left_lon), (right_lat, right_lon), dist, dist)\n",
+    "\n",
+    "grid_dataset = []\n",
+    "for i, row in enumerate(grid):\n",
+    "    for j, point in enumerate(row):\n",
+    "        result_df = get_osm_data(point[0], point[1], 710)\n",
+    "        # print(result_df.head(3))\n",
+    "        labelled_df = result_df[result_df['Location Type'] != 'Other']\n",
+    "        labelled_df = labelled_df[labelled_df['Location Type'] != 'Religious']\n",
+    "        labelled_df = labelled_df[labelled_df['Location Type'] != 'Transportation']\n",
+    "        loc_types = []\n",
+    "        for row in labelled_df.iterrows():\n",
+    "            loc_type = (row[1]['Location Name'], row[1]['Location Type'])\n",
+    "            if loc_type not in loc_types:\n",
+    "                loc_types.append(loc_type)\n",
+    "\n",
+    "        labelled_df = pd.DataFrame(loc_types, columns=['Location Name', 'Location Type'])\n",
+    "\n",
+    "        row_of_dataset = ''\n",
+    "\n",
+    "        for row in labelled_df.iterrows():\n",
+    "            row_text = row[1]['Location Name'] + ' is a ' + row[1]['Location Type']\n",
+    "            row_of_dataset += row_text + '; '\n",
+    "        ## replacing any coma in the text with a blank space\n",
+    "\n",
+    "        row_of_dataset = row_of_dataset.replace(',', ' ')\n",
+    "        \n",
+    "        grid_row = {\"row\": i, \"col\": j, \"latitude\": point[0], \"longitude\": point[1], \"Map Data\": row_of_dataset}\n",
+    "        grid_dataset.append(grid_row)\n",
+    "\n",
+    "grid_df = pd.DataFrame(grid_dataset)\n",
+    "grid_df.to_csv('MMR_DATASET.csv', index=False)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 107,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style scoped>\n",
+       "    .dataframe tbody tr th:only-of-type {\n",
+       "        vertical-align: middle;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>row</th>\n",
+       "      <th>col</th>\n",
+       "      <th>latitude</th>\n",
+       "      <th>longitude</th>\n",
+       "      <th>Map Data</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>18.894330</td>\n",
+       "      <td>72.784597</td>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>0</td>\n",
+       "      <td>1</td>\n",
+       "      <td>18.894330</td>\n",
+       "      <td>72.794102</td>\n",
+       "      <td>Prongs Reef is a Natural,</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>0</td>\n",
+       "      <td>2</td>\n",
+       "      <td>18.894330</td>\n",
+       "      <td>72.803607</td>\n",
+       "      <td>United Services Club Golf Course is a Leisure ...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>0</td>\n",
+       "      <td>3</td>\n",
+       "      <td>18.894330</td>\n",
+       "      <td>72.813112</td>\n",
+       "      <td>Indian Meterological Department is a Commercia...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>1</td>\n",
+       "      <td>0</td>\n",
+       "      <td>18.903323</td>\n",
+       "      <td>72.784597</td>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>5</th>\n",
+       "      <td>1</td>\n",
+       "      <td>1</td>\n",
+       "      <td>18.903323</td>\n",
+       "      <td>72.794102</td>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>6</th>\n",
+       "      <td>1</td>\n",
+       "      <td>2</td>\n",
+       "      <td>18.903323</td>\n",
+       "      <td>72.803607</td>\n",
+       "      <td>Jagadish Canteen is a Food &amp; Drink, Maratha St...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>7</th>\n",
+       "      <td>1</td>\n",
+       "      <td>3</td>\n",
+       "      <td>18.903323</td>\n",
+       "      <td>72.813112</td>\n",
+       "      <td>Indian Meterological Department is a Commercia...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>8</th>\n",
+       "      <td>2</td>\n",
+       "      <td>0</td>\n",
+       "      <td>18.912316</td>\n",
+       "      <td>72.784597</td>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>9</th>\n",
+       "      <td>2</td>\n",
+       "      <td>1</td>\n",
+       "      <td>18.912316</td>\n",
+       "      <td>72.794102</td>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>10</th>\n",
+       "      <td>2</td>\n",
+       "      <td>2</td>\n",
+       "      <td>18.912316</td>\n",
+       "      <td>72.803607</td>\n",
+       "      <td>Jagadish Canteen is a Food &amp; Drink, Maratha St...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>11</th>\n",
+       "      <td>2</td>\n",
+       "      <td>3</td>\n",
+       "      <td>18.912316</td>\n",
+       "      <td>72.813112</td>\n",
+       "      <td>Cafe Coffee Day is a Food &amp; Drink, King Plaza ...</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>12</th>\n",
+       "      <td>3</td>\n",
+       "      <td>0</td>\n",
+       "      <td>18.921309</td>\n",
+       "      <td>72.784597</td>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>13</th>\n",
+       "      <td>3</td>\n",
+       "      <td>1</td>\n",
+       "      <td>18.921309</td>\n",
+       "      <td>72.794102</td>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>14</th>\n",
+       "      <td>3</td>\n",
+       "      <td>2</td>\n",
+       "      <td>18.921309</td>\n",
+       "      <td>72.803607</td>\n",
+       "      <td></td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>15</th>\n",
+       "      <td>3</td>\n",
+       "      <td>3</td>\n",
+       "      <td>18.921309</td>\n",
+       "      <td>72.813112</td>\n",
+       "      <td>Cafe Coffee Day is a Food &amp; Drink, King Plaza ...</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "    row  col   latitude  longitude  \\\n",
+       "0     0    0  18.894330  72.784597   \n",
+       "1     0    1  18.894330  72.794102   \n",
+       "2     0    2  18.894330  72.803607   \n",
+       "3     0    3  18.894330  72.813112   \n",
+       "4     1    0  18.903323  72.784597   \n",
+       "5     1    1  18.903323  72.794102   \n",
+       "6     1    2  18.903323  72.803607   \n",
+       "7     1    3  18.903323  72.813112   \n",
+       "8     2    0  18.912316  72.784597   \n",
+       "9     2    1  18.912316  72.794102   \n",
+       "10    2    2  18.912316  72.803607   \n",
+       "11    2    3  18.912316  72.813112   \n",
+       "12    3    0  18.921309  72.784597   \n",
+       "13    3    1  18.921309  72.794102   \n",
+       "14    3    2  18.921309  72.803607   \n",
+       "15    3    3  18.921309  72.813112   \n",
+       "\n",
+       "                                             Map Data  \n",
+       "0                                                      \n",
+       "1                          Prongs Reef is a Natural,   \n",
+       "2   United Services Club Golf Course is a Leisure ...  \n",
+       "3   Indian Meterological Department is a Commercia...  \n",
+       "4                                                      \n",
+       "5                                                      \n",
+       "6   Jagadish Canteen is a Food & Drink, Maratha St...  \n",
+       "7   Indian Meterological Department is a Commercia...  \n",
+       "8                                                      \n",
+       "9                                                      \n",
+       "10  Jagadish Canteen is a Food & Drink, Maratha St...  \n",
+       "11  Cafe Coffee Day is a Food & Drink, King Plaza ...  \n",
+       "12                                                     \n",
+       "13                                                     \n",
+       "14                                                     \n",
+       "15  Cafe Coffee Day is a Food & Drink, King Plaza ...  "
+      ]
+     },
+     "execution_count": 107,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "grid_df.head(20)"
+   ]
+  }
+ ],
+ "metadata": {
+  "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": "3.12.0"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

src/main.py

@@ -0,0 +1,33 @@
+######################################## IMPORTING REQUIRED LIBRARIES ####################################
+import os
+import sys
+sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+data_folder = os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), 'data')
+from utilities import get_data, input_filter, clean_data
+
+
+################################################## INPUTS ################################################
+
+left_lat = 18.889833
+left_lon = 72.779844
+dist = 35
+
+def data_sourcing():
+    lat, lon = input_filter(lat = left_lat, lon=left_lon)
+    df = get_data(lat, lon, dist)
+    df.to_csv(f'{data_folder}/MMR_DATA.csv', index=False)
+    return df
+
+def data_clean_for_training(df):
+    df = clean_data(df)
+    df.to_csv(f'{data_folder}/MMR_DATA_CLEAN.csv', index=False)
+    return df
+
+
+if __name__ == '__main__':
+
+    df = data_sourcing() ## testing the data sourcing endpoint
+    if df:
+        print("Data loaded successfully !!")
+
+    clean_df = data_clean_for_training(df)

utilities/__init__.py

@@ -0,0 +1,2 @@
+from .data_loader import get_data, input_filter
+from .data_cleaner import clean_data

utilities/data_cleaner.py

@@ -0,0 +1,27 @@
+import re
+import nltk
+from nltk.corpus import stopwords
+from nltk.stem import PorterStemmer
+from nltk.stem import WordNetLemmatizer
+
+def clean_text(text):
+    nltk.download('stopwords')
+    nltk.download('wordnet')
+
+    stop_words = set(stopwords.words('english'))
+    stemmer = PorterStemmer()
+    lemmatizer = WordNetLemmatizer()
+    
+    text = re.sub(r'[^\w\s]', '', text)
+    text = text.lower()
+    text = [word for word in text.split() if word not in stop_words]
+    text = [stemmer.stem(word) for word in text] 
+    text = [lemmatizer.lemmatize(word) for word in text]
+    return ' '.join(text)
+
+def clean_data(df):
+    df['Map Data'] = df['Map Data'].fillna('')
+    df = df[df['Map Data'].str.len() > 0]
+    df = df[df['Map Data'].str.len() < 5000]
+    # df['Map Data'] = df['Map Data'].apply(clean_text)
+    return df

utilities/data_loader.py

@@ -0,0 +1,222 @@
+import requests
+import pandas as pd
+import re
+import math
+from typing import Tuple, List, Dict
+
+def fetch_osm_data(lat: float, lon: float, radius: int) -> List[Dict]:
+    overpass_url = "http://overpass-api.de/api/interpreter"
+    overpass_query = f"""
+    [out:json];
+    (
+      node["name"](around:{radius},{lat},{lon});
+      way["name"](around:{radius},{lat},{lon});
+      relation["name"](around:{radius},{lat},{lon});
+    );
+    out center;
+    """
+    
+    response = requests.get(overpass_url, params={'data': overpass_query})
+    data = response.json()
+    return data['elements']
+
+def determine_location_type(tags: Dict[str, str]) -> str:
+    # Residential
+    if 'building' in tags and tags['building'] in ['residential', 'house', 'apartments', 'detached', 'terrace', 'dormitory', 'bungalow']:
+        return 'Residential'
+    
+    # Commercial
+    if any(key in tags for key in ['shop', 'office', 'craft']):
+        return 'Commercial'
+    if 'building' in tags and tags['building'] in ['commercial', 'office', 'retail', 'supermarket', 'kiosk']:
+        return 'Commercial'
+    
+    # Industrial
+    if 'building' in tags and tags['building'] in ['industrial', 'warehouse', 'factory', 'manufacture']:
+        return 'Industrial'
+    if 'industrial' in tags or 'industry' in tags:
+        return 'Industrial'
+    
+    # Educational
+    if 'amenity' in tags and tags['amenity'] in ['school', 'university', 'college', 'library', 'kindergarten', 'language_school']:
+        return 'Educational'
+    
+    # Healthcare
+    if 'amenity' in tags and tags['amenity'] in ['hospital', 'clinic', 'doctors', 'dentist', 'pharmacy', 'veterinary']:
+        return 'Healthcare'
+    
+    # Food & Drink
+    if 'amenity' in tags and tags['amenity'] in ['restaurant', 'cafe', 'bar', 'fast_food', 'pub', 'food_court']:
+        return 'Food & Drink'
+    
+    # Leisure & Entertainment
+    if 'leisure' in tags or 'tourism' in tags:
+        return 'Leisure & Entertainment'
+    if 'amenity' in tags and tags['amenity'] in ['theatre', 'cinema', 'nightclub', 'arts_centre', 'community_centre']:
+        return 'Leisure & Entertainment'
+    
+    # Transportation
+    if 'amenity' in tags and tags['amenity'] in ['parking', 'bicycle_parking', 'bus_station', 'ferry_terminal']:
+        return 'Transportation'
+    if 'highway' in tags or 'railway' in tags or 'aeroway' in tags:
+        return 'Transportation'
+    
+    # Religious
+    if 'amenity' in tags and tags['amenity'] in ['place_of_worship', 'monastery']:
+        return 'Religious'
+    
+    # Government & Public Services
+    if 'amenity' in tags and tags['amenity'] in ['townhall', 'courthouse', 'police', 'fire_station', 'post_office']:
+        return 'Government & Public Services'
+    
+    # Parks & Recreation
+    if 'leisure' in tags and tags['leisure'] in ['park', 'playground', 'sports_centre', 'stadium', 'garden']:
+        return 'Parks & Recreation'
+    
+    # Natural
+    if 'natural' in tags:
+        return 'Natural'
+    
+    # Landuse
+    if 'landuse' in tags:
+        landuse = tags['landuse'].capitalize()
+        if landuse in ['Residential', 'Commercial', 'Industrial', 'Retail']:
+            return landuse
+        else:
+            return f'Landuse: {landuse}'
+    
+    # If no specific category is found, return 'Other'
+    return 'Other'
+
+def parse_osm_data(elements: List[Dict]) -> pd.DataFrame:
+    parsed_data = []
+    for element in elements:
+        tags = element.get('tags', {})
+        parsed_element = {
+            'ID': f"{element['type']}_{element['id']}",
+            'Location Name': tags.get('name', ''),
+            'Location Type': determine_location_type(tags)
+        }
+        parsed_data.append(parsed_element)
+    if len(parsed_data) == 0:
+        return pd.DataFrame(columns=['ID', 'Location Name', 'Location Type'])
+    return pd.DataFrame(parsed_data)
+
+def get_osm_data(lat: float, lon: float, radius: int) -> pd.DataFrame:
+    raw_data = fetch_osm_data(lat, lon, radius)
+    return parse_osm_data(raw_data)
+
+def dms_to_decimal(coord_str):
+    # Regular expression to match the coordinate format
+    pattern = r'(\d+)°(\d+)\'([\d.]+)"([NS])\s*(\d+)°(\d+)\'([\d.]+)"([EW])'
+    
+    match = re.match(pattern, coord_str)
+    if not match:
+        raise ValueError("Invalid coordinate format. Expected format: 19°03'08.6\"N 72°54'06.0\"E")
+
+    lat_deg, lat_min, lat_sec, lat_dir, lon_deg, lon_min, lon_sec, lon_dir = match.groups()
+
+    # Convert to decimal degrees
+    lat = float(lat_deg) + float(lat_min)/60 + float(lat_sec)/3600
+    lon = float(lon_deg) + float(lon_min)/60 + float(lon_sec)/3600
+
+    # Adjust sign based on direction
+    if lat_dir == 'S':
+        lat = -lat
+    if lon_dir == 'W':
+        lon = -lon
+
+    return lat, lon
+
+
+def calculate_distant_points(lat: float, lon: float, distance: float) -> tuple:
+    # Earth's radius in meters
+    R = 6371000
+
+    # Convert latitude and longitude to radians
+    lat_rad = math.radians(lat)
+    lon_rad = math.radians(lon)
+
+    # Calculate the point with the same latitude (moving east-west)
+    delta_lon = distance / (R * math.cos(lat_rad))
+    lon1 = lon + math.degrees(delta_lon)
+    
+    # Calculate the point with the same longitude (moving north-south)
+    delta_lat = distance / R
+    lat2 = lat + math.degrees(delta_lat)
+
+    return ((lat, lon1), (lat2, lon))
+
+## 2d map grid (0,0) --> bottom left
+
+def create_map_grid(bottom_left: Tuple[float, float], top_right: Tuple[float, float], rows: int, cols: int) -> List[List[Tuple[float, float]]]:
+    grid = []
+    lat_unit = (top_right[0] - bottom_left[0]) / rows
+    lon_unit = (top_right[1] - bottom_left[1]) / cols
+    
+    for i in range(rows):
+        row = []
+        for j in range(cols):
+            lat = bottom_left[0] + i * lat_unit
+            lon = bottom_left[1] + j * lon_unit
+            lat = lat + lat_unit / 2
+            lon = lon + lon_unit / 2
+            row.append((lat, lon))
+        grid.append(row)
+    
+    return grid
+
+## entire pipeline
+
+left_lat = 18.889833
+left_lon = 72.779844
+dist = 35
+
+def input_filter(lat=None, lon=None, string=None):
+    if lat != None:
+        return (lat, lon)
+    elif string != None:
+        latitude, longitude = dms_to_decimal(string)
+        return (latitude, longitude)
+    else:
+        return None
+
+def get_data(bottom_left_lat, bottom_left_lon, dist):
+
+    result =  calculate_distant_points(bottom_left_lat, bottom_left_lon, 1000*dist)
+
+    top_right_lat = result[1][0]
+    top_right_lon = result[0][1]
+    grid = create_map_grid((left_lat, left_lon), (top_right_lat, top_right_lon), dist, dist)
+
+    grid_dataset = []
+    for i, row in enumerate(grid):
+        for j, point in enumerate(row):
+            result_df = get_osm_data(point[0], point[1], 710)
+            # print(result_df.head(3))
+            labelled_df = result_df[result_df['Location Type'] != 'Other']
+            labelled_df = labelled_df[labelled_df['Location Type'] != 'Religious']
+            labelled_df = labelled_df[labelled_df['Location Type'] != 'Transportation']
+            loc_types = []
+            for row in labelled_df.iterrows():
+                loc_type = (row[1]['Location Name'], row[1]['Location Type'])
+                if loc_type not in loc_types:
+                    loc_types.append(loc_type)
+
+            labelled_df = pd.DataFrame(loc_types, columns=['Location Name', 'Location Type'])
+
+            row_of_dataset = ''
+
+            for row in labelled_df.iterrows():
+                row_text = row[1]['Location Name'] + ' is a ' + row[1]['Location Type']
+                row_of_dataset += row_text + '; '
+            ## replacing any coma in the text with a blank space
+
+            row_of_dataset = row_of_dataset.replace(',', ' ')
+            
+            grid_row = {"row": i, "col": j, "latitude": point[0], "longitude": point[1], "Map Data": row_of_dataset}
+            grid_dataset.append(grid_row)
+
+    grid_df = pd.DataFrame(grid_dataset)
+    return grid_df
+    # grid_df.to_csv('MMR_DATASET.csv', index=False)