UserBasedCF.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import the Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as pyplot\n",
    "from Evaluation.Evaluator import EvaluatorHoldout\n",
    "from Data_manager.split_functions.split_train_validation_random_holdout import split_train_in_two_percentage_global_sample\n",
    "from Recommenders.Similarity.Compute_Similarity_Python import Compute_Similarity_Python\n",
    "# Numpy cries because it is an old code so we monkey patch it\n",
    "np.int = int\n",
    "np.bool = bool\n",
    "np.float = float"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import the Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "URM_all_dataframe = pd.read_csv('data_train.csv', \n",
    "                                sep=\",\", \n",
    "                                header= 0, \n",
    "                                dtype={0:int, 1:int, 2:float},\n",
    "                                engine='python')\n",
    "\n",
    "URM_all_dataframe.columns = [\"UserID\", \"ItemID\", \"Interaction\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Building the sparse COO matrix using URM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import scipy.sparse as sps\n",
    "# Build the COO sparse matrix associated with the URM\n",
    "URM_all = sps.coo_matrix((URM_all_dataframe[\"Interaction\"].values, \n",
    "                          (URM_all_dataframe[\"UserID\"].values, URM_all_dataframe[\"ItemID\"].values))) ## .values --> numpy array, df[..] --> pd series"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Split the data in train and validation set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Warning: 614 (4.71 %) of 13025 users have no train items\n",
      "Warning: 2577 (19.79 %) of 13025 users have no sampled items\n",
      "Warning: 845 (6.49 %) of 13025 users have no train items\n",
      "Warning: 2976 (22.85 %) of 13025 users have no sampled items\n",
      "EvaluatorHoldout: Ignoring 2976 (22.8%) Users that have less than 1 test interactions\n",
      "EvaluatorHoldout: Ignoring 2577 (19.8%) Users that have less than 1 test interactions\n"
     ]
    }
   ],
   "source": [
    "URM_train, URM_test = split_train_in_two_percentage_global_sample(URM_all, train_percentage = 0.80)\n",
    "URM_train, URM_validation = split_train_in_two_percentage_global_sample(URM_train, train_percentage = 0.80)\n",
    "\n",
    "evaluator_validation = EvaluatorHoldout(URM_validation, cutoff_list=[10])\n",
    "evaluator_test = EvaluatorHoldout(URM_test, cutoff_list=[10])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Implement UserBasedCF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "UserKNNCFRecommender: URM Detected 845 ( 6.5%) users with no interactions.\n",
      "UserKNNCFRecommender: URM Detected 479 ( 2.1%) items with no interactions.\n",
      "Unable to load Cython Compute_Similarity, reverting to Python\n",
      "Similarity column 13025 (100.0%), 1255.31 column/sec. Elapsed time 10.38 sec\n",
      "EvaluatorHoldout: Processed 10049 (100.0%) in 9.40 sec. Users per second: 1069\n",
      "UserKNNCFRecommender: URM Detected 845 ( 6.5%) users with no interactions.\n",
      "UserKNNCFRecommender: URM Detected 479 ( 2.1%) items with no interactions.\n",
      "Unable to load Cython Compute_Similarity, reverting to Python\n",
      "Similarity column 13025 (100.0%), 1295.82 column/sec. Elapsed time 10.05 sec\n",
      "EvaluatorHoldout: Processed 10049 (100.0%) in 8.15 sec. Users per second: 1234\n",
      "UserKNNCFRecommender: URM Detected 845 ( 6.5%) users with no interactions.\n",
      "UserKNNCFRecommender: URM Detected 479 ( 2.1%) items with no interactions.\n",
      "Unable to load Cython Compute_Similarity, reverting to Python\n",
      "Similarity column 13025 (100.0%), 1098.51 column/sec. Elapsed time 11.86 sec\n",
      "EvaluatorHoldout: Processed 10049 (100.0%) in 10.42 sec. Users per second: 965\n",
      "UserKNNCFRecommender: URM Detected 845 ( 6.5%) users with no interactions.\n",
      "UserKNNCFRecommender: URM Detected 479 ( 2.1%) items with no interactions.\n",
      "Unable to load Cython Compute_Similarity, reverting to Python\n",
      "Similarity column 13025 (100.0%), 1000.41 column/sec. Elapsed time 13.02 sec\n",
      "EvaluatorHoldout: Processed 10049 (100.0%) in 10.91 sec. Users per second: 921\n",
      "UserKNNCFRecommender: URM Detected 845 ( 6.5%) users with no interactions.\n",
      "UserKNNCFRecommender: URM Detected 479 ( 2.1%) items with no interactions.\n",
      "Unable to load Cython Compute_Similarity, reverting to Python\n",
      "Similarity column 13025 (100.0%), 854.59 column/sec. Elapsed time 15.24 sec\n",
      "EvaluatorHoldout: Processed 10049 (100.0%) in 10.66 sec. Users per second: 943\n",
      "UserKNNCFRecommender: URM Detected 845 ( 6.5%) users with no interactions.\n",
      "UserKNNCFRecommender: URM Detected 479 ( 2.1%) items with no interactions.\n",
      "Unable to load Cython Compute_Similarity, reverting to Python\n",
      "Similarity column 13025 (100.0%), 736.80 column/sec. Elapsed time 17.68 sec\n",
      "EvaluatorHoldout: Processed 10049 (100.0%) in 11.73 sec. Users per second: 857\n"
     ]
    }
   ],
   "source": [
    "from Recommenders.KNN.UserKNNCFRecommender import UserKNNCFRecommender\n",
    "\n",
    "x_tick = [100, 200, 500, 600, 800, 1000]\n",
    "MAP_per_k = []\n",
    "\n",
    "for topK in x_tick:\n",
    "    \n",
    "    recommender = UserKNNCFRecommender(URM_train)\n",
    "    recommender.fit(shrink=0.0, topK=topK)\n",
    "    \n",
    "    result_df, _ = evaluator_validation.evaluateRecommender(recommender)\n",
    "    \n",
    "    MAP_per_k.append(result_df.loc[10][\"MAP\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pyplot.plot(x_tick, MAP_per_k)\n",
    "pyplot.ylabel('MAP')\n",
    "pyplot.xlabel('TopK')\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# The value of k where MAP is the highest is 500\n",
    "best_k = 500"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "UserKNNCFRecommender: URM Detected 845 ( 6.5%) users with no interactions.\n",
      "UserKNNCFRecommender: URM Detected 479 ( 2.1%) items with no interactions.\n",
      "Unable to load Cython Compute_Similarity, reverting to Python\n",
      "Similarity column 13025 (100.0%), 777.39 column/sec. Elapsed time 16.75 sec\n",
      "EvaluatorHoldout: Processed 10049 (100.0%) in 10.05 sec. Users per second: 1000\n",
      "UserKNNCFRecommender: URM Detected 845 ( 6.5%) users with no interactions.\n",
      "UserKNNCFRecommender: URM Detected 479 ( 2.1%) items with no interactions.\n",
      "Unable to load Cython Compute_Similarity, reverting to Python\n",
      "Similarity column 13025 (100.0%), 1117.51 column/sec. Elapsed time 11.66 sec\n",
      "EvaluatorHoldout: Processed 10049 (100.0%) in 9.79 sec. Users per second: 1026\n",
      "UserKNNCFRecommender: URM Detected 845 ( 6.5%) users with no interactions.\n",
      "UserKNNCFRecommender: URM Detected 479 ( 2.1%) items with no interactions.\n",
      "Unable to load Cython Compute_Similarity, reverting to Python\n",
      "Similarity column 13025 (100.0%), 1080.50 column/sec. Elapsed time 12.05 sec\n",
      "EvaluatorHoldout: Processed 10049 (100.0%) in 9.84 sec. Users per second: 1021\n",
      "UserKNNCFRecommender: URM Detected 845 ( 6.5%) users with no interactions.\n",
      "UserKNNCFRecommender: URM Detected 479 ( 2.1%) items with no interactions.\n",
      "Unable to load Cython Compute_Similarity, reverting to Python\n",
      "Similarity column 13025 (100.0%), 1025.76 column/sec. Elapsed time 12.70 sec\n",
      "EvaluatorHoldout: Processed 10049 (100.0%) in 10.27 sec. Users per second: 979\n"
     ]
    }
   ],
   "source": [
    "# Let's find the optimal shrinkage\n",
    "x_tick = [0,0.5,1,2]\n",
    "MAP_per_shrinkage = []\n",
    "\n",
    "for shrink in x_tick:\n",
    "    \n",
    "    recommender = UserKNNCFRecommender(URM_train)\n",
    "    recommender.fit(shrink=shrink, topK=best_k)\n",
    "    \n",
    "    result_df, _ = evaluator_validation.evaluateRecommender(recommender)\n",
    "    \n",
    "    MAP_per_shrinkage.append(result_df.loc[10][\"MAP\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pyplot.plot(x_tick, MAP_per_shrinkage)\n",
    "pyplot.ylabel('MAP')\n",
    "pyplot.xlabel('Shrinkage')\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "best_shrinkage = 0.5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "UserKNNCFRecommender: URM Detected 845 ( 6.5%) users with no interactions.\n",
      "UserKNNCFRecommender: URM Detected 479 ( 2.1%) items with no interactions.\n",
      "Unable to load Cython Compute_Similarity, reverting to Python\n",
      "Similarity column 13025 (100.0%), 705.86 column/sec. Elapsed time 18.45 sec\n"
     ]
    }
   ],
   "source": [
    "recommender = UserKNNCFRecommender(URM_train)\n",
    "recommender.fit(shrink=best_shrinkage, topK=best_k)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   user_id                                          item_list\n",
      "0        1                36 101 15 66 506 354 342 183 83 592\n",
      "1        2                             3 2 8 4 15 5 6 17 19 9\n",
      "2        3                       59 9 7 259 1 584 536 648 5 4\n",
      "3        4                         28 27 50 1 37 6 7 5 14 136\n",
      "4        5                      4 51 88 341 6 26 87 16 36 177\n",
      "5        6                            8 2 1 7 6 81 14 88 26 3\n",
      "6        8                        210 20 80 9 11 8 578 2 1 22\n",
      "7        9  11175 3727 3726 3725 3724 3723 3722 3721 3720 ...\n",
      "8       10                  3 2 1668 1411 89 809 8 1446 68 11\n",
      "9       11                           4 1 67 2 44 40 6 3 31 39\n"
     ]
    }
   ],
   "source": [
    "def write_predictions(recommender_object, at=10):\n",
    "    prediction_df = pd.read_csv('data_target_users_test.csv', sep= \",\",\n",
    "                                header=0, \n",
    "                                dtype={0:int},\n",
    "                                engine='python')\n",
    "    #prediction_df['mapped user'] = prediction_df['user_id'].map(user_original_ID_to_index).fillna(-1).astype(int)\n",
    "    #prediction_df['item_list'] = prediction_df['mapped user'].apply(recommender_object.recommend)\n",
    "    prediction_df['item_list'] = prediction_df['user_id'].apply(lambda user_id: recommender_object.recommend(user_id)[:at])\n",
    "    #def map_items(item_list):\n",
    "    #    return [index_to_item_original.get(item, item) for item in item_list]\n",
    "\n",
    "    #prediction_df['item_list'] = prediction_df['item_list'].apply(map_items)\n",
    "    def transform_items_to_string(item_list):\n",
    "        return ' '.join(map(str, item_list))\n",
    "\n",
    "    prediction_df['item_list'] = prediction_df['item_list'].apply(transform_items_to_string)\n",
    "    print(prediction_df.head(10))\n",
    "    #del prediction_df['mapped user']\n",
    "    prediction_df.to_csv('submission.csv',index=False)\n",
    "\n",
    "write_predictions(recommender)"
   ]
  }
 ],
 "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": "3.11.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}