Untitled3.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "UIuGc9gMByOc"
   },
   "source": [
    "<h1> Hierarchical Agglomerative Clustering </h1>\n",
    "<br>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "irIi6Ms95qgE"
   },
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "5IUqAhGDGM60"
   },
   "source": [
    "#### Loading Scraped and processed data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "lz8mWtHs5sFo"
   },
   "outputs": [],
   "source": [
    "df = pd.read_csv(\"/content/drive/MyDrive/edited_mdf.csv\", encoding = \"ISO-8859-1\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "pcKgFnAE5-n3",
    "outputId": "16f5754c-1f42-4c28-983e-ef5e06f48bd4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "32\n"
     ]
    }
   ],
   "source": [
    "df.head()\n",
    "print(len(df['cuisine'].unique()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "K51nfPlX-J3q"
   },
   "outputs": [],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "from sklearn.metrics import silhouette_samples, silhouette_score\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.cm as cm\n",
    "import numpy as np\n",
    "from sklearn.cluster import AgglomerativeClustering \n",
    "from sklearn.metrics.pairwise import cosine_similarity\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "nbdYISjZGSiE"
   },
   "source": [
    "#### Vectorising the corpus for the further computations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "rE_xT0loAqCa",
    "outputId": "d9d1e375-fef8-4f74-980b-68962f2c0c02"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['aam', 'achar', 'achari', 'adai', 'adhirasam', 'akhrot', 'akki', 'allam', 'almond', 'alo', 'aloo', 'alu', 'amrakhand', 'amritsari', 'amti', 'and', 'appam', 'appe', 'apple', 'aratikaya', 'ash', 'at', 'atta', 'authentic', 'aval', 'avalakki', 'aviyal', 'baby', 'badam', 'badanekai', 'badusha', 'bagara', 'baigana', 'baingan', 'bajji', 'bajre', 'bake', 'baked', 'balehannu', 'balls', 'banana', 'banane', 'bangalore', 'bar', 'barfi', 'basic', 'basket', 'basundi', 'batata', 'bateta', 'bath', 'bati', 'batter', 'bean', 'beans', 'beerakaya', 'beetroot', 'bele', 'bendekai', 'bengali', 'benne', 'besan', 'besasn', 'bhaat', 'bhaja', 'bhaji', 'bhajiya', 'bhakri', 'bhara', 'bharwa', 'bhature', 'bhavana', 'bhel', 'bhindi', 'bhujia', 'bhurji', 'biriyani', 'biryani', 'biscuit', 'biscuits', 'bisi', 'bisibelabath', 'bites', 'bitter', 'black', 'blossom', 'bolu', 'bombay', 'bonda', 'boondi', 'boost', 'box', 'brea', 'bread', 'brinjal', 'brownies', 'bun', 'buns', 'burfi', 'burger', 'burnt', 'butter', 'butterscotch', 'cabbage', 'cak', 'cake', 'capsicum', 'caramel', 'carrot', 'cauliflower', 'chaat', 'chakkuli', 'chakli', 'cham', 'chana', 'channa', 'chapati', 'chasni', 'chatni', 'chawal', 'cheela', 'cheese', 'chekkalu', 'chhena', 'chhole', 'chikki', 'chilli', 'chintapandu', 'chip', 'chips', 'chitranna', 'chivda', 'choco', 'chocobar', 'chocolate', 'chocolaty', 'chole', 'chop', 'chunks', 'chura', 'churchur', 'churma', 'chutne', 'chutney', 'cilantro', 'classic', 'club', 'coconu', 'coconut', 'coins', 'colour', 'condensed', 'cook', 'cooke', 'cooked', 'cooker', 'cookie', 'cookies', 'corn', 'cornflakes', 'cotta', 'crea', 'cream', 'crisp', 'crispy', 'crunchy', 'cupcake', 'curd', 'curr', 'curry', 'custard', 'cutlet', 'cutlets', 'cutting', 'daal', 'dabeli', 'dahi', 'dal', 'daler', 'daliya', 'date', 'davanagere', 'deep', 'desi', 'dessert', 'dhaba', 'dhabeli', 'dharwad', 'dhokla', 'dilkush', 'dill', 'dip', 'diwani', 'do', 'dog', 'doodh', 'doodhi', 'dos', 'dosa', 'dosai', 'dose', 'double', 'drink', 'drinks', 'dry', 'dum', 'dumplings', 'easy', 'egg', 'eggless', 'eggs', 'falafel', 'falooda', 'faluda', 'farali', 'farsi', 'fenugreek', 'filter', 'finger', 'fingers', 'firni', 'flavoured', 'flour', 'food', 'for', 'frankie', 'fried', 'fritters', 'from', 'frooti', 'fruit', 'frutti', 'fry', 'gajar', 'garam', 'garelu', 'garli', 'garlic', 'gassi', 'gathia', 'gatte', 'gelati', 'ghee', 'ginger', 'gobhi', 'gobi', 'gojju', 'gold', 'golgappa', 'golla', 'goolgoola', 'gourd', 'grapefruit', 'grav', 'gravy', 'green', 'grilled', 'groundnut', 'gujarati', 'gujiya', 'gujrati', 'gulab', 'gulla', 'hagalkayi', 'hakka', 'halbai', 'haldiram', 'halwa', 'handi', 'hannu', 'hara', 'hari', 'hesaru', 'hittu', 'hom', 'homemade', 'honey', 'hot', 'how', 'huli', 'hyderabadi', 'ice', 'idli', 'idly', 'imli', 'immune', 'in', 'indian', 'instant', 'irani', 'ja', 'jackfruit', 'jain', 'jalebi', 'jalfrez', 'jamun', 'jeera', 'jhal', 'jini', 'jowar', 'juice', 'junnu', 'ka', 'kaali', 'kaapi', 'kabab', 'kacchi', 'kacha', 'kachiyathu', 'kachori', 'kadai', 'kadhi', 'kairi', 'kaju', 'kala', 'kalakand', 'kalanda', 'kalappam', 'kara', 'karahi', 'karanji', 'karela', 'kashayam', 'kashmiri', 'kathi', 'kathirikai', 'katori', 'kayirasa', 'ke', 'kele', 'kerala', 'kesar', 'kesari', 'ketchup', 'khaja', 'khajoor', 'khajur', 'khaman', 'khana', 'khandvi', 'khara', 'khasta', 'kheer', 'kheera', 'khichdi', 'khoya', 'ki', 'kidney', 'kids', 'koddel', 'kofta', 'kokum', 'kolhapuri', 'korma', 'kosambari', 'kothimbir', 'kottambari', 'kovakkai', 'kuchumber', 'kulambu', 'kulcha', 'kulche', 'kulfi', 'kumblakai', 'kurma', 'kuska', 'kuzhambu', 'lacha', 'lachha', 'laddu', 'ladoo', 'laduu', 'lasooni', 'lassi', 'lauki', 'lava', 'leaves', 'left', 'leftover', 'lentil', 'lime', 'lollipop', 'lot', 'low', 'lunch', 'macaroni', 'maggi', 'magic', 'maharashtrian', 'maida', 'majjige', 'make', 'maker', 'makhana', 'makhanwala', 'malabar', 'malai', 'malpua', 'malpura', 'mamidikaya', 'manchow', 'manchuria', 'manchurian', 'mangalore', 'mangalorean', 'mango', 'marble', 'masal', 'masala', 'masale', 'masrum', 'matar', 'mathri', 'mattu', 'mavina', 'mavinakayi', 'maw', 'mawa', 'mayo', 'mcaloo', 'meal', 'medu', 'meetha', 'meethe', 'menasinakai', 'methi', 'mexican', 'microwave', 'milagu', 'milk', 'milkmai', 'milkmaid', 'milkshake', 'mint', 'minute', 'mirapakaya', 'mirch', 'mirchi', 'mix', 'mixed', 'mixie', 'mixture', 'moglai', 'mohanthal', 'moist', 'molten', 'momo', 'momos', 'month', 'moong', 'mor', 'more', 'moru', 'mosaranna', 'motichoor', 'mousse', 'mozzarella', 'mug', 'mumbai', 'muri', 'murmura', 'murukku', 'mushroom', 'muthiya', 'mutter', 'mysore', 'na', 'naan', 'naariyal', 'nachos', 'namak', 'nankhatai', 'nariyal', 'nasta', 'nawabi', 'neer', 'nimbu', 'no', 'non', 'noodle', 'noodles', 'nool', 'north', 'nu', 'nuggekai', 'nuggets', 'nut', 'nutty', 'oatmeal', 'oats', 'obbattu', 'of', 'okra', 'omapodi', 'omelette', 'on', 'onion', 'or', 'over', 'paan', 'pachadi', 'paddu', 'padhir', 'pak', 'pakoda', 'pakode', 'pakora', 'palak', 'palkova', 'palya', 'pan', 'pancakes', 'panchamrut', 'paneer', 'panha', 'pani', 'paniyaram', 'panna', 'papad', 'papadum', 'papdi', 'pappu', 'papri', 'para', 'parantha', 'parath', 'paratha', 'parathe', 'parle', 'parotta', 'parsi', 'party', 'paruppu', 'pasta', 'patta', 'patties', 'pav', 'payasam', 'paysa', 'peanut', 'peas', 'peda', 'pedha', 'peel', 'peni', 'pepper', 'phirni', 'phool', 'pickle', 'pilaf', 'pineapple', 'pinwheels', 'pista', 'pithe', 'pizza', 'plain', 'plus', 'poda', 'poha', 'poli', 'pomegranate', 'pongal', 'pooja', 'poori', 'popsicle', 'porota', 'porridge', 'potat', 'potato', 'powde', 'powder', 'prepare', 'pressure', 'protein', 'puddin', 'pudding', 'pudi', 'puff', 'puffed', 'pulao', 'pulav', 'puli', 'pulihora', 'puliyodharai', 'puliyogare', 'pumpkin', 'punjabi', 'puran', 'puri', 'puttu', 'pyaz', 'quick', 'rabdi', 'radish', 'ragi', 'rait', 'raita', 'raj', 'rajasthani', 'rajma', 'rasam', 'rasayana', 'rasmalai', 'rava', 'ready', 'recip', 'recipe', 'recipes', 'red', 'reshmi', 'restaurant', 'ribbon', 'ric', 'rice', 'ridge', 'ring', 'rings', 'roasted', 'roll', 'rolls', 'roomali', 'rosogulla', 'roti', 'rotti', 'royal', 'saag', 'sabakki', 'sabji', 'sabudana', 'sabudane', 'sabzi', 'sadam', 'saffron', 'sago', 'sagu', 'sakkarai', 'salad', 'salan', 'salna', 'salsa', 'sama', 'samba', 'sambar', 'samosa', 'sandesh', 'sandige', 'sandwic', 'sandwich', 'sandwiches', 'saravana', 'sasmi', 'satyanarayan', 'sauc', 'sauce', 'schezwan', 'seeds', 'semiya', 'semolina', 'sesame', 'sev', 'seviyan', 'shaak', 'shahi', 'shakarpara', 'shavige', 'sheer', 'sheera', 'sheet', 'shimla', 'shrikand', 'simple', 'six', 'slice', 'snacks', 'soft', 'sooj', 'sooji', 'soppu', 'soup', 'sour', 'south', 'southekai', 'soya', 'spice', 'spiced', 'spicy', 'spinach', 'sponge', 'spring', 'sprouted', 'sprouts', 'stand', 'steamed', 'stew', 'sticky', 'stir', 'stove', 'strawberry', 'street', 'student', 'stuffed', 'styl', 'style', 'sugar', 'suji', 'super', 'suran', 'sweet', 'swiss', 'system', 'szechuan', 'tadka', 'tamarind', 'tamatar', 'tambli', 'tambuli', 'tandoori', 'tangy', 'taw', 'tawa', 'tea', 'tehri', 'temple', 'thakkali', 'thali', 'thalipeeth', 'thambittu', 'thattai', 'thenkuzhal', 'thepla', 'thokku', 'thoran', 'tiffin', 'tikka', 'tikki', 'til', 'tiranga', 'tirunelveli', 'to', 'toast', 'tomato', 'top', 'tove', 'traditional', 'tri', 'triple', 'tutti', 'twister', 'uddina', 'udupi', 'unniappam', 'upkari', 'upm', 'upma', 'upvasache', 'urad', 'uttapa', 'uttapam', 'vada', 'vadai', 'vadi', 'vangi', 'vankaya', 'vazhakkai', 'veg', 'vegetable', 'vegetarian', 'veggie', 'vendakkai', 'vermicelli', 'vidh', 'wafers', 'wale', 'wali', 'walnut', 'ways', 'wedge', 'white', 'with', 'without', 'yakhni', 'yeas', 'yeast', 'yennegai', 'yoghurt', 'yogurt']\n"
     ]
    }
   ],
   "source": [
    "s = []\n",
    "X = df['title']\n",
    "vectorizer = CountVectorizer()\n",
    "X = vectorizer.fit_transform(X)\n",
    "print(vectorizer.get_feature_names())\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "lS-ZxudjFzLl"
   },
   "source": [
    "#### Find the optimal number of clusters using the sillhouette score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "mgI4HVshA0tQ",
    "outputId": "ae8b37f1-d809-4731-f386-ab12f52fecfd"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "For n_clusters = 2 The average silhouette_score is : 0.06165364261430987\n",
      "For n_clusters = 3 The average silhouette_score is : 0.07910757004134152\n",
      "For n_clusters = 4 The average silhouette_score is : 0.08100501915592619\n",
      "For n_clusters = 5 The average silhouette_score is : 0.06373353957405292\n",
      "For n_clusters = 6 The average silhouette_score is : 0.0794486090002619\n",
      "For n_clusters = 7 The average silhouette_score is : 0.08146910800916704\n",
      "For n_clusters = 8 The average silhouette_score is : 0.061208846682004034\n",
      "For n_clusters = 9 The average silhouette_score is : 0.06280547403781743\n"
     ]
    }
   ],
   "source": [
    "for n_clusters in range(2, 10):\n",
    "    clusterer = AgglomerativeClustering(n_clusters=n_clusters, random_state=10)\n",
    "    cluster_labels = clusterer.fit_predict(X.todense())\n",
    "    silhouette_avg = silhouette_score(X.todense(), cluster_labels)\n",
    "    s.append(silhouette_avg)\n",
    "    print(\"For n_clusters =\", n_clusters,\n",
    "          \"The average silhouette_score is :\", silhouette_avg)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "1AqgFwo28OUx"
   },
   "source": [
    "#### Number of clusters through visualising a dendogram"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "YpuD7Xtx8N5c"
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "from sklearn.cluster import AgglomerativeClustering \n",
    "from sklearn.preprocessing import StandardScaler, normalize \n",
    "from sklearn.metrics import silhouette_score \n",
    "import scipy.cluster.hierarchy as shc "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 607
    },
    "id": "B-C3hUkr8YWY",
    "outputId": "a829e5c1-bae1-4647-ebae-8af13f751a81"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize =(20,10)) \n",
    "plt.title('Visualising the data') \n",
    "label=[str(i) for i in range(X.get_shape()[1])]\n",
    "linked=shc.linkage(X.toarray(), method ='complete',metric='cosine')\n",
    "Dendrogram = shc.dendrogram(linked)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "gH5GqN7q8eqA",
    "outputId": "cce74cb0-37f5-4171-edaf-88dfec8d5bcb"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7"
      ]
     },
     "execution_count": 15,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(set(Dendrogram['color_list']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "uqEjZOxjGecr"
   },
   "source": [
    "### Function to retrieve the optimal clusters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "4E0ydp2oTwRj"
   },
   "outputs": [],
   "source": [
    "def get_clusters(n_clusters):\n",
    "\n",
    "    model = AgglomerativeClustering(n_clusters=n_clusters, affinity='euclidean', linkage='ward')\n",
    "    model.fit_predict(X.todense())\n",
    "    labels = model.labels_\n",
    "    clustering_dict = model.__dict__\n",
    "    clusters = {}\n",
    "\n",
    "    for document_id,cluster_label in enumerate(labels):\n",
    "    if cluster_label not in clusters:\n",
    "        clusters[cluster_label] = []\n",
    "    clusters[cluster_label].append(document_id)\n",
    "\n",
    "    return(clusters)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "orHLgHaGGmjC"
   },
   "source": [
    "### Generate the recommendations for the query"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "KQE5QQycUH8w"
   },
   "outputs": [],
   "source": [
    "def get_recommendation(query, clusters, n):\n",
    "    q = vectorizer.transform([query])\n",
    "    mat = cosine_similarity(q,X)\n",
    "    idx = list(np.argsort(mat[0]))\n",
    "    idx.reverse()\n",
    "\n",
    "    ms = idx[0]\n",
    "    for k, v in clusters.items():\n",
    "    if(ms in v):\n",
    "        cluster_number = k\n",
    "        break\n",
    "\n",
    "    rec_list = [i for i in idx if i in clusters[cluster_number]][0:n+1]\n",
    "    return(rec_list)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "8ov9P2JmFV-7"
   },
   "source": [
    "<h1> Search with query and find similar recommendations </h1>\n",
    "<br>\n",
    "Enter the required recipe to be found"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "WV-nZqqHTNWl",
    "outputId": "0198d653-814c-444a-8c1d-063bda093970"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Enter Query..paneer\n"
     ]
    }
   ],
   "source": [
    "clusters = get_clusters(7)\n",
    "no_rec=10\n",
    "q = input(\"Enter Query..\")\n",
    "rl = get_recommendation(q, clusters, no_rec)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "_Q1MvhwOVWje",
    "outputId": "a552f3cb-3b21-4f32-91c6-5713e4443c3f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Most relevant recipe   karahi paneer  ->  ---------- punjabi\n",
      "You may also like :\n",
      "paneer schezwan -> ---- indo chinese\n",
      "paneer cheela -> ---- north indian\n",
      "paneer momo -> ---- nepali\n",
      "paneer bites -> ---- Indian\n",
      "karahi paneer -> ---- punjabi\n",
      "paneer pakoda -> ---- Indian\n",
      "nawabi paneer -> ---- hyderabadi\n",
      "paneer tikka gravy -> ---- punjabi\n",
      "paneer tikki recipe -> ---- north indian\n",
      "paneer reshmi recipe -> ---- north indian\n"
     ]
    }
   ],
   "source": [
    "item=rl[0]\n",
    "print(\"Most relevant recipe  \",df[\"title\"][item],\" ->  ----------\",df[\"cuisine\"][item])\n",
    "print(\"You may also like :\")\n",
    "for i in range(1,len(rl)):\n",
    "    print(df[\"title\"][rl[i]],\"-> ----\",df[\"cuisine\"][rl[i]])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mF5UwtpWEIgb"
   },
   "source": [
    "### Precision at K"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "8Tq91a1QMXxt",
    "outputId": "e4a0915f-b31f-4a7a-813f-bb7e2f6c9fcd"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision at  10 is  0.2\n"
     ]
    }
   ],
   "source": [
    "#prec_at_k\n",
    "\n",
    "relev=0\n",
    "for i in range(1,len(rl)):\n",
    "    if df[\"cuisine\"][item]==df[\"cuisine\"][rl[i]]:\n",
    "        relev+=1\n",
    "prec_at_k=relev/no_rec\n",
    "print(\"Precision at \",no_rec,\"is \",prec_at_k)\n"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "IR_paper_implementation.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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}