{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "fa4cc168",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2026-06-22 12:26:10,379 - INFO - PHASE 1: Loading data...\n",
      "2026-06-22 12:26:25,130 - INFO - PHASE 2: Filtering (User >= 20, Item >= 200)...\n",
      "2026-06-22 12:26:39,908 - INFO - Filtreleme sonrası kalan satır: 35,007,076\n",
      "2026-06-22 12:26:39,909 - INFO - PHASE 3: Performing random sampling (100000 users, 10000 items)...\n",
      "2026-06-22 12:26:45,480 - INFO - PHASE 4: Saving to Parquet format...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "=============================================\n",
      "MSD FINAL SAMPLING REPORT\n",
      "=============================================\n",
      "Number of Users          : 100,000\n",
      "Number of Items          : 10,000 (Effective: 10,000)\n",
      "Number of Interactions   : 1,282,657\n",
      "Sparsity                 : 0.1283%\n",
      "=============================================\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2026-06-22 12:26:45,744 - INFO - Successfully saved to: C:\\Users\\Hakan YILMAZER\\Downloads\\msd_ready_sampled.parquet\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import pyarrow as pa\n",
    "import pyarrow.parquet as pq\n",
    "import os\n",
    "import zipfile\n",
    "import logging\n",
    "import numpy as np\n",
    "\n",
    "# Configure logging\n",
    "logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')\n",
    "\n",
    "def extract_and_filter(min_uc=20, min_sc=200, sample_user=100000, sample_item=10000):\n",
    "    # Set your local paths\n",
    "    base_dir = r\"C:\\Users\\Hakan YILMAZER\\Downloads\"\n",
    "    txt_file = os.path.join(base_dir, \"msd_raw\", \"train_triplets.txt\")\n",
    "    parquet_output_path = os.path.join(base_dir, \"msd_ready_sampled.parquet\")\n",
    "\n",
    "    # 1. Load Data\n",
    "    logging.info(\"PHASE 1: Loading data...\")\n",
    "    df = pd.read_csv(txt_file, sep='\\t', header=None, names=['userId', 'movieId', 'rating'])\n",
    "    df['rating'] = 1  # Implicit feedback binarization\n",
    "\n",
    "    # 2. Filter First (Establish the stable core)\n",
    "    logging.info(f\"PHASE 2: Filtering (User >= {min_uc}, Item >= {min_sc})...\")\n",
    "    user_counts = df.groupby('userId').size()\n",
    "    item_counts = df.groupby('movieId').size()\n",
    "    \n",
    "    df = df[df['userId'].isin(user_counts[user_counts >= min_uc].index)]\n",
    "    df = df[df['movieId'].isin(item_counts[item_counts >= min_sc].index)]\n",
    "    logging.info(f\"Filtreleme sonrası kalan satır: {len(df):,}\")\n",
    "\n",
    "    # 3. Random Sampling (Perform on the high-quality core)\n",
    "    logging.info(f\"PHASE 3: Performing random sampling ({sample_user} users, {sample_item} items)...\")\n",
    "    \n",
    "    unique_items = df['movieId'].unique()\n",
    "    chosen_items = np.random.choice(unique_items, size=min(sample_item, len(unique_items)), replace=False)\n",
    "    df = df[df['movieId'].isin(chosen_items)]\n",
    "            \n",
    "    unique_users = df['userId'].unique()\n",
    "    chosen_users = np.random.choice(unique_users, size=min(sample_user, len(unique_users)), replace=False)\n",
    "    df = df[df['userId'].isin(chosen_users)]\n",
    "\n",
    "    # 4. Final Reporting\n",
    "    num_users = df['userId'].nunique()\n",
    "    num_items = df['movieId'].nunique()\n",
    "    num_ratings = len(df)\n",
    "    \n",
    "    # Check for padding requirement\n",
    "    if num_items % 2 != 0:\n",
    "        effective_num_items = num_items + 1\n",
    "    else:\n",
    "        effective_num_items = num_items\n",
    "\n",
    "    # Sparsity calculation\n",
    "    sparsity = (num_ratings / (num_users * effective_num_items))\n",
    "    \n",
    "    print(\"\\n\" + \"=\"*45)\n",
    "    print(\"MSD FINAL SAMPLING REPORT\")\n",
    "    print(\"=\"*45)\n",
    "    print(f\"Number of Users          : {num_users:,}\")\n",
    "    print(f\"Number of Items          : {num_items:,} (Effective: {effective_num_items:,})\")\n",
    "    print(f\"Number of Interactions   : {num_ratings:,}\")\n",
    "    print(f\"Sparsity                 : {sparsity * 100:.4f}%\")\n",
    "    print(\"=\"*45 + \"\\n\")\n",
    "\n",
    "    # 5. Save to Parquet\n",
    "    logging.info(\"PHASE 4: Saving to Parquet format...\")\n",
    "    table = pa.Table.from_pandas(df, preserve_index=False)\n",
    "    pq.write_table(table, parquet_output_path, compression='snappy')\n",
    "    logging.info(f\"Successfully saved to: {parquet_output_path}\")\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    # Ensure reproducibility with seed\n",
    "    np.random.seed(98765)\n",
    "    extract_and_filter()"
   ]
  }
 ],
 "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": "3.12.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}