ttests_untuned_results.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Testing the Untuned Results\n",
    "\n",
    "In this notebook, we test the untuned results of the paper. The following steps will be performed:\n",
    "\n",
    "1. Concatenate all the run results from our experiments into one DataFrame.  \n",
    "2. Test the model performances using t-tests.  \n",
    "\n",
    "\n",
    "Importing the necessary libraries:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import re\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "from scipy.stats import ttest_1samp\n",
    "import difflib"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Preparation\n",
    "\n",
    "\n",
    "### Prepare results\n",
    "\n",
    "First we need to put some structure into our run results. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>model</th>\n",
       "      <th>nDCG</th>\n",
       "      <th>MAP</th>\n",
       "      <th>MRR</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>F1</th>\n",
       "      <th>HR</th>\n",
       "      <th>experiment</th>\n",
       "      <th>run</th>\n",
       "      <th>file_name</th>\n",
       "      <th>run_folder</th>\n",
       "      <th>cutoff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...</td>\n",
       "      <td>0.001989</td>\n",
       "      <td>0.000370</td>\n",
       "      <td>0.001183</td>\n",
       "      <td>0.000532</td>\n",
       "      <td>0.005842</td>\n",
       "      <td>0.000937</td>\n",
       "      <td>0.010434</td>\n",
       "      <td>epinions</td>\n",
       "      <td>30</td>\n",
       "      <td>rec_NeuMF_cutoff_20_relthreshold_0_2025_01_30_...</td>\n",
       "      <td>run_30</td>\n",
       "      <td>20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>GeneralizedMF_seed=42_e=10_bs=512_lr=0$5_mffac...</td>\n",
       "      <td>0.004170</td>\n",
       "      <td>0.000567</td>\n",
       "      <td>0.002220</td>\n",
       "      <td>0.000497</td>\n",
       "      <td>0.014121</td>\n",
       "      <td>0.000938</td>\n",
       "      <td>0.024416</td>\n",
       "      <td>epinions</td>\n",
       "      <td>30</td>\n",
       "      <td>rec_cutoff_50_relthreshold_0_2025_01_30_03_56_...</td>\n",
       "      <td>run_30</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...</td>\n",
       "      <td>0.004043</td>\n",
       "      <td>0.000524</td>\n",
       "      <td>0.001745</td>\n",
       "      <td>0.000522</td>\n",
       "      <td>0.014383</td>\n",
       "      <td>0.000988</td>\n",
       "      <td>0.025876</td>\n",
       "      <td>epinions</td>\n",
       "      <td>30</td>\n",
       "      <td>rec_cutoff_50_relthreshold_0_2025_01_30_03_56_...</td>\n",
       "      <td>run_30</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...</td>\n",
       "      <td>0.004039</td>\n",
       "      <td>0.000521</td>\n",
       "      <td>0.001739</td>\n",
       "      <td>0.000522</td>\n",
       "      <td>0.014383</td>\n",
       "      <td>0.000988</td>\n",
       "      <td>0.025876</td>\n",
       "      <td>epinions</td>\n",
       "      <td>30</td>\n",
       "      <td>rec_cutoff_50_relthreshold_0_2025_01_30_03_56_...</td>\n",
       "      <td>run_30</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.004100</td>\n",
       "      <td>0.000506</td>\n",
       "      <td>0.002918</td>\n",
       "      <td>0.000476</td>\n",
       "      <td>0.012138</td>\n",
       "      <td>0.000891</td>\n",
       "      <td>0.023372</td>\n",
       "      <td>epinions</td>\n",
       "      <td>30</td>\n",
       "      <td>rec_cutoff_50_relthreshold_0_2025_01_30_03_56_...</td>\n",
       "      <td>run_30</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1966</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.022058</td>\n",
       "      <td>0.004477</td>\n",
       "      <td>0.021478</td>\n",
       "      <td>0.004116</td>\n",
       "      <td>0.052602</td>\n",
       "      <td>0.007538</td>\n",
       "      <td>0.184037</td>\n",
       "      <td>movielens</td>\n",
       "      <td>7</td>\n",
       "      <td>rec_cutoff_50_relthreshold_0_2025_01_26_23_10_...</td>\n",
       "      <td>run_7</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1967</th>\n",
       "      <td>MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...</td>\n",
       "      <td>0.034038</td>\n",
       "      <td>0.007227</td>\n",
       "      <td>0.036726</td>\n",
       "      <td>0.005963</td>\n",
       "      <td>0.075417</td>\n",
       "      <td>0.010925</td>\n",
       "      <td>0.268470</td>\n",
       "      <td>movielens</td>\n",
       "      <td>7</td>\n",
       "      <td>rec_cutoff_50_relthreshold_0_2025_01_26_23_10_...</td>\n",
       "      <td>run_7</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1968</th>\n",
       "      <td>ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...</td>\n",
       "      <td>0.025100</td>\n",
       "      <td>0.005099</td>\n",
       "      <td>0.030870</td>\n",
       "      <td>0.004142</td>\n",
       "      <td>0.052775</td>\n",
       "      <td>0.007601</td>\n",
       "      <td>0.189974</td>\n",
       "      <td>movielens</td>\n",
       "      <td>7</td>\n",
       "      <td>rec_cutoff_50_relthreshold_0_2025_01_26_23_10_...</td>\n",
       "      <td>run_7</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1969</th>\n",
       "      <td>ConvMF_seed=42_e=10_bs=32_lr=0$05_embedding_si...</td>\n",
       "      <td>0.025100</td>\n",
       "      <td>0.005099</td>\n",
       "      <td>0.030870</td>\n",
       "      <td>0.004142</td>\n",
       "      <td>0.052775</td>\n",
       "      <td>0.007601</td>\n",
       "      <td>0.189974</td>\n",
       "      <td>movielens</td>\n",
       "      <td>7</td>\n",
       "      <td>rec_cutoff_50_relthreshold_0_2025_01_26_23_10_...</td>\n",
       "      <td>run_7</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1970</th>\n",
       "      <td>EASER_neighborhood=1014_l2_norm=3$72</td>\n",
       "      <td>0.040788</td>\n",
       "      <td>0.008632</td>\n",
       "      <td>0.041187</td>\n",
       "      <td>0.007098</td>\n",
       "      <td>0.092116</td>\n",
       "      <td>0.013043</td>\n",
       "      <td>0.304090</td>\n",
       "      <td>movielens</td>\n",
       "      <td>7</td>\n",
       "      <td>rec_cutoff_50_relthreshold_0_2025_01_26_23_10_...</td>\n",
       "      <td>run_7</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1971 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                  model      nDCG       MAP  \\\n",
       "0     NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...  0.001989  0.000370   \n",
       "1     GeneralizedMF_seed=42_e=10_bs=512_lr=0$5_mffac...  0.004170  0.000567   \n",
       "2     NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...  0.004043  0.000524   \n",
       "3     MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...  0.004039  0.000521   \n",
       "4     NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.004100  0.000506   \n",
       "...                                                 ...       ...       ...   \n",
       "1966  NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.022058  0.004477   \n",
       "1967  MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...  0.034038  0.007227   \n",
       "1968  ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...  0.025100  0.005099   \n",
       "1969  ConvMF_seed=42_e=10_bs=32_lr=0$05_embedding_si...  0.025100  0.005099   \n",
       "1970               EASER_neighborhood=1014_l2_norm=3$72  0.040788  0.008632   \n",
       "\n",
       "           MRR  Precision    Recall        F1        HR experiment  run  \\\n",
       "0     0.001183   0.000532  0.005842  0.000937  0.010434   epinions   30   \n",
       "1     0.002220   0.000497  0.014121  0.000938  0.024416   epinions   30   \n",
       "2     0.001745   0.000522  0.014383  0.000988  0.025876   epinions   30   \n",
       "3     0.001739   0.000522  0.014383  0.000988  0.025876   epinions   30   \n",
       "4     0.002918   0.000476  0.012138  0.000891  0.023372   epinions   30   \n",
       "...        ...        ...       ...       ...       ...        ...  ...   \n",
       "1966  0.021478   0.004116  0.052602  0.007538  0.184037  movielens    7   \n",
       "1967  0.036726   0.005963  0.075417  0.010925  0.268470  movielens    7   \n",
       "1968  0.030870   0.004142  0.052775  0.007601  0.189974  movielens    7   \n",
       "1969  0.030870   0.004142  0.052775  0.007601  0.189974  movielens    7   \n",
       "1970  0.041187   0.007098  0.092116  0.013043  0.304090  movielens    7   \n",
       "\n",
       "                                              file_name run_folder  cutoff  \n",
       "0     rec_NeuMF_cutoff_20_relthreshold_0_2025_01_30_...     run_30      20  \n",
       "1     rec_cutoff_50_relthreshold_0_2025_01_30_03_56_...     run_30      50  \n",
       "2     rec_cutoff_50_relthreshold_0_2025_01_30_03_56_...     run_30      50  \n",
       "3     rec_cutoff_50_relthreshold_0_2025_01_30_03_56_...     run_30      50  \n",
       "4     rec_cutoff_50_relthreshold_0_2025_01_30_03_56_...     run_30      50  \n",
       "...                                                 ...        ...     ...  \n",
       "1966  rec_cutoff_50_relthreshold_0_2025_01_26_23_10_...      run_7      50  \n",
       "1967  rec_cutoff_50_relthreshold_0_2025_01_26_23_10_...      run_7      50  \n",
       "1968  rec_cutoff_50_relthreshold_0_2025_01_26_23_10_...      run_7      50  \n",
       "1969  rec_cutoff_50_relthreshold_0_2025_01_26_23_10_...      run_7      50  \n",
       "1970  rec_cutoff_50_relthreshold_0_2025_01_26_23_10_...      run_7      50  \n",
       "\n",
       "[1971 rows x 13 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Disclaimer: This code cell was partly written with LLM (ChatGPT) support\n",
    "list_experiments = [\"Untuned_epinions\", \"Untuned_amazon\", \"Untuned_movieLenz\"]\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "mapping_experiment_name = {\n",
    "    \"Untuned_epinions\": \"epinions\",\n",
    "    \"Untuned_amazon\": \"amazon\",\n",
    "    \"Untuned_movieLenz\": \"movielens\",\n",
    "}\n",
    "\n",
    "data_frames = []\n",
    "for experiment in list_experiments:\n",
    "    # Set the base directory where the run folders are stored\n",
    "    base_dir = os.path.join(\"performance\", experiment)\n",
    "\n",
    "    # List to collect dataframes from each file\n",
    "\n",
    "    # Iterate over each subdirectory (run folder) in the base directory\n",
    "    for run_folder in os.listdir(base_dir):\n",
    "        run_path = os.path.join(base_dir, run_folder)\n",
    "        # Check if the item is a directory and its name starts with \"run_\"\n",
    "        if os.path.isdir(run_path) and run_folder.startswith(\"run_\"):\n",
    "            # Extract run number (assuming folder names like \"run_1\", \"run_2\", etc.)\n",
    "            run_number = int(run_folder.split(\"_\")[-1])\n",
    "\n",
    "            # Loop through files in the run folder\n",
    "            for file_name in os.listdir(run_path):\n",
    "                if file_name.endswith(\".tsv\"):\n",
    "                    file_path = os.path.join(run_path, file_name)\n",
    "                    # Read the TSV file\n",
    "                    df = pd.read_csv(file_path, sep=\"\\t\")\n",
    "                    # Add columns for the run number and file name for later identification\n",
    "                    df[\"experiment\"] = mapping_experiment_name[experiment]\n",
    "                    df[\"run\"] = run_number\n",
    "                    df[\"file_name\"] = file_name\n",
    "                    # Optionally, you could also add the run folder name:\n",
    "                    df[\"run_folder\"] = run_folder\n",
    "\n",
    "                    pattern = r\"cutoff_(\\d+)\"\n",
    "                    match = re.search(pattern, file_name)\n",
    "                    if match:\n",
    "                        cutoff_value = int(match.group(1))\n",
    "                        df[\"cutoff\"] = cutoff_value\n",
    "\n",
    "                    data_frames.append(df)\n",
    "\n",
    "    # Concatenate all dataframes along rows\n",
    "df_experiments = pd.concat(data_frames, ignore_index=True)\n",
    "\n",
    "\n",
    "df_experiments"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check for missing values outliers in duplicates. Some files in the run folders like 'rec_NeuMF_cutoff_10_relthreshold_0_2025_01_27_21_03_59' which only contain the NeuMF model are suspicous as the model is also in the \"normal\" performance file. Essentially this could mean that we have a lot of duplicates in the dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Contains nulls: False \n",
      "\n",
      "Number of duplicates: 219\n",
      "Subsample of duplicates:\n"
     ]
    },
    {
     "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>model</th>\n",
       "      <th>nDCG</th>\n",
       "      <th>MAP</th>\n",
       "      <th>MRR</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>F1</th>\n",
       "      <th>HR</th>\n",
       "      <th>experiment</th>\n",
       "      <th>run</th>\n",
       "      <th>file_name</th>\n",
       "      <th>run_folder</th>\n",
       "      <th>cutoff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1079</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.003399</td>\n",
       "      <td>0.000475</td>\n",
       "      <td>0.001893</td>\n",
       "      <td>0.000466</td>\n",
       "      <td>0.010785</td>\n",
       "      <td>0.000880</td>\n",
       "      <td>0.022862</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>rec_NeuMF_cutoff_50_relthreshold_0_2025_02_01_...</td>\n",
       "      <td>run_1</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1062</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.001947</td>\n",
       "      <td>0.000456</td>\n",
       "      <td>0.001508</td>\n",
       "      <td>0.000508</td>\n",
       "      <td>0.004759</td>\n",
       "      <td>0.000895</td>\n",
       "      <td>0.010161</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>rec_NeuMF_cutoff_20_relthreshold_0_2025_02_01_...</td>\n",
       "      <td>run_1</td>\n",
       "      <td>20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1053</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.001065</td>\n",
       "      <td>0.000391</td>\n",
       "      <td>0.001066</td>\n",
       "      <td>0.000402</td>\n",
       "      <td>0.002000</td>\n",
       "      <td>0.000648</td>\n",
       "      <td>0.004022</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>rec_NeuMF_cutoff_10_relthreshold_0_2025_02_01_...</td>\n",
       "      <td>run_1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>864</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.003185</td>\n",
       "      <td>0.000425</td>\n",
       "      <td>0.002007</td>\n",
       "      <td>0.000420</td>\n",
       "      <td>0.009402</td>\n",
       "      <td>0.000790</td>\n",
       "      <td>0.020762</td>\n",
       "      <td>amazon</td>\n",
       "      <td>2</td>\n",
       "      <td>rec_NeuMF_cutoff_50_relthreshold_0_2025_02_01_...</td>\n",
       "      <td>run_2</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>881</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.001962</td>\n",
       "      <td>0.000415</td>\n",
       "      <td>0.001675</td>\n",
       "      <td>0.000471</td>\n",
       "      <td>0.004392</td>\n",
       "      <td>0.000826</td>\n",
       "      <td>0.009418</td>\n",
       "      <td>amazon</td>\n",
       "      <td>2</td>\n",
       "      <td>rec_NeuMF_cutoff_20_relthreshold_0_2025_02_01_...</td>\n",
       "      <td>run_2</td>\n",
       "      <td>20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>890</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.001198</td>\n",
       "      <td>0.000446</td>\n",
       "      <td>0.001340</td>\n",
       "      <td>0.000385</td>\n",
       "      <td>0.001850</td>\n",
       "      <td>0.000603</td>\n",
       "      <td>0.003853</td>\n",
       "      <td>amazon</td>\n",
       "      <td>2</td>\n",
       "      <td>rec_NeuMF_cutoff_10_relthreshold_0_2025_02_01_...</td>\n",
       "      <td>run_2</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>963</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.003348</td>\n",
       "      <td>0.000448</td>\n",
       "      <td>0.001795</td>\n",
       "      <td>0.000434</td>\n",
       "      <td>0.010681</td>\n",
       "      <td>0.000819</td>\n",
       "      <td>0.021290</td>\n",
       "      <td>amazon</td>\n",
       "      <td>3</td>\n",
       "      <td>rec_NeuMF_cutoff_50_relthreshold_0_2025_02_01_...</td>\n",
       "      <td>run_3</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>954</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.002064</td>\n",
       "      <td>0.000425</td>\n",
       "      <td>0.001463</td>\n",
       "      <td>0.000516</td>\n",
       "      <td>0.005302</td>\n",
       "      <td>0.000900</td>\n",
       "      <td>0.010108</td>\n",
       "      <td>amazon</td>\n",
       "      <td>3</td>\n",
       "      <td>rec_NeuMF_cutoff_20_relthreshold_0_2025_02_01_...</td>\n",
       "      <td>run_3</td>\n",
       "      <td>20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>953</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.001091</td>\n",
       "      <td>0.000356</td>\n",
       "      <td>0.001003</td>\n",
       "      <td>0.000344</td>\n",
       "      <td>0.002183</td>\n",
       "      <td>0.000567</td>\n",
       "      <td>0.003441</td>\n",
       "      <td>amazon</td>\n",
       "      <td>3</td>\n",
       "      <td>rec_NeuMF_cutoff_10_relthreshold_0_2025_02_01_...</td>\n",
       "      <td>run_3</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>928</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.003578</td>\n",
       "      <td>0.000511</td>\n",
       "      <td>0.002107</td>\n",
       "      <td>0.000492</td>\n",
       "      <td>0.011236</td>\n",
       "      <td>0.000922</td>\n",
       "      <td>0.024390</td>\n",
       "      <td>amazon</td>\n",
       "      <td>4</td>\n",
       "      <td>rec_NeuMF_cutoff_50_relthreshold_0_2025_02_01_...</td>\n",
       "      <td>run_4</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                  model      nDCG       MAP  \\\n",
       "1079  NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.003399  0.000475   \n",
       "1062  NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.001947  0.000456   \n",
       "1053  NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.001065  0.000391   \n",
       "864   NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.003185  0.000425   \n",
       "881   NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.001962  0.000415   \n",
       "890   NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.001198  0.000446   \n",
       "963   NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.003348  0.000448   \n",
       "954   NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.002064  0.000425   \n",
       "953   NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.001091  0.000356   \n",
       "928   NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.003578  0.000511   \n",
       "\n",
       "           MRR  Precision    Recall        F1        HR experiment  run  \\\n",
       "1079  0.001893   0.000466  0.010785  0.000880  0.022862     amazon    1   \n",
       "1062  0.001508   0.000508  0.004759  0.000895  0.010161     amazon    1   \n",
       "1053  0.001066   0.000402  0.002000  0.000648  0.004022     amazon    1   \n",
       "864   0.002007   0.000420  0.009402  0.000790  0.020762     amazon    2   \n",
       "881   0.001675   0.000471  0.004392  0.000826  0.009418     amazon    2   \n",
       "890   0.001340   0.000385  0.001850  0.000603  0.003853     amazon    2   \n",
       "963   0.001795   0.000434  0.010681  0.000819  0.021290     amazon    3   \n",
       "954   0.001463   0.000516  0.005302  0.000900  0.010108     amazon    3   \n",
       "953   0.001003   0.000344  0.002183  0.000567  0.003441     amazon    3   \n",
       "928   0.002107   0.000492  0.011236  0.000922  0.024390     amazon    4   \n",
       "\n",
       "                                              file_name run_folder  cutoff  \n",
       "1079  rec_NeuMF_cutoff_50_relthreshold_0_2025_02_01_...      run_1      50  \n",
       "1062  rec_NeuMF_cutoff_20_relthreshold_0_2025_02_01_...      run_1      20  \n",
       "1053  rec_NeuMF_cutoff_10_relthreshold_0_2025_02_01_...      run_1      10  \n",
       "864   rec_NeuMF_cutoff_50_relthreshold_0_2025_02_01_...      run_2      50  \n",
       "881   rec_NeuMF_cutoff_20_relthreshold_0_2025_02_01_...      run_2      20  \n",
       "890   rec_NeuMF_cutoff_10_relthreshold_0_2025_02_01_...      run_2      10  \n",
       "963   rec_NeuMF_cutoff_50_relthreshold_0_2025_02_01_...      run_3      50  \n",
       "954   rec_NeuMF_cutoff_20_relthreshold_0_2025_02_01_...      run_3      20  \n",
       "953   rec_NeuMF_cutoff_10_relthreshold_0_2025_02_01_...      run_3      10  \n",
       "928   rec_NeuMF_cutoff_50_relthreshold_0_2025_02_01_...      run_4      50  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Contains nulls:\",sum(df_experiments.isna().sum())>0, \"\\n\")\n",
    "\n",
    "\n",
    "# If the values in these columns are duplicated, it means that one of the duplicated entries can be removed. \n",
    "subset_columns = [\n",
    "'model',\n",
    "'nDCG',\n",
    "'MAP',\n",
    "'MRR',\n",
    "'Precision',\n",
    "'Recall',\n",
    "'F1',\n",
    "'HR',\n",
    "'experiment',\n",
    "'run',\n",
    "'cutoff'\n",
    " ]\n",
    "\n",
    "print(\"Number of duplicates:\" ,sum(df_experiments.duplicated(subset=subset_columns)))\n",
    "\n",
    "df_experiments = df_experiments.sort_values(by=[\"experiment\",\"run\",\"file_name\"], ascending=[True,True,False])\n",
    "\n",
    "print(\"Subsample of duplicates:\")\n",
    "\n",
    "df_experiments[df_experiments.duplicated(subset=subset_columns)].head(10)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that there a multiple duplicates. First we have to check if the number of rec_NeuMF files equals the number of duplicates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "219"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(df_experiments[\"file_name\"].str.startswith(\"rec_NeuMF\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The number of rec_NeuMF files and the number of duplicates is equal. That means we can drop them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of duplicates: 0\n"
     ]
    }
   ],
   "source": [
    "df_experiments = df_experiments[~df_experiments[\"file_name\"].str.startswith(\"rec_NeuMF\")]\n",
    "print(\"Number of duplicates:\" ,sum(df_experiments.duplicated(subset=subset_columns)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We only need a subset of columns and rows for the testing the reported values. The paper uses the nDCG@10 results (cutoff value of 10)"
   ]
  },
  {
   "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>model</th>\n",
       "      <th>nDCG</th>\n",
       "      <th>experiment</th>\n",
       "      <th>run</th>\n",
       "      <th>cutoff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1071</th>\n",
       "      <td>GeneralizedMF_seed=42_e=15_bs=512_lr=0$07_mffa...</td>\n",
       "      <td>0.003706</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1072</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.001065</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1073</th>\n",
       "      <td>MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...</td>\n",
       "      <td>0.002580</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1074</th>\n",
       "      <td>MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...</td>\n",
       "      <td>0.004335</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1075</th>\n",
       "      <td>ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...</td>\n",
       "      <td>0.002285</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1173</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.008221</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1174</th>\n",
       "      <td>MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...</td>\n",
       "      <td>0.012410</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1175</th>\n",
       "      <td>ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...</td>\n",
       "      <td>0.012007</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1176</th>\n",
       "      <td>ConvMF_seed=42_e=10_bs=32_lr=0$05_embedding_si...</td>\n",
       "      <td>0.012007</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1177</th>\n",
       "      <td>EASER_neighborhood=1017_l2_norm=3$72</td>\n",
       "      <td>0.020207</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>584 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                  model      nDCG experiment  \\\n",
       "1071  GeneralizedMF_seed=42_e=15_bs=512_lr=0$07_mffa...  0.003706     amazon   \n",
       "1072  NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.001065     amazon   \n",
       "1073  MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...  0.002580     amazon   \n",
       "1074  MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...  0.004335     amazon   \n",
       "1075  ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...  0.002285     amazon   \n",
       "...                                                 ...       ...        ...   \n",
       "1173  NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.008221  movielens   \n",
       "1174  MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...  0.012410  movielens   \n",
       "1175  ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...  0.012007  movielens   \n",
       "1176  ConvMF_seed=42_e=10_bs=32_lr=0$05_embedding_si...  0.012007  movielens   \n",
       "1177               EASER_neighborhood=1017_l2_norm=3$72  0.020207  movielens   \n",
       "\n",
       "      run  cutoff  \n",
       "1071    1      10  \n",
       "1072    1      10  \n",
       "1073    1      10  \n",
       "1074    1      10  \n",
       "1075    1      10  \n",
       "...   ...     ...  \n",
       "1173   30      10  \n",
       "1174   30      10  \n",
       "1175   30      10  \n",
       "1176   30      10  \n",
       "1177   30      10  \n",
       "\n",
       "[584 rows x 5 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "columns_to_keep = [\n",
    "    \"model\",\n",
    "    \"nDCG\",\n",
    "    \"experiment\",\n",
    "    \"run\",\n",
    "    \"cutoff\"\n",
    "]\n",
    "\n",
    "df_experiments = df_experiments[columns_to_keep]\n",
    "df_experiments = df_experiments[df_experiments[\"cutoff\"] == 10]\n",
    "df_experiments"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finally we create a DataFrame with the performances noted in the paper."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of duplicates: 0\n"
     ]
    },
    {
     "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>model</th>\n",
       "      <th>nDCG</th>\n",
       "      <th>experiment</th>\n",
       "      <th>cutoff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>GeneralizedMF_seed=42_e=10_bs=512_lr=0$5_mffac...</td>\n",
       "      <td>0.001836</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...</td>\n",
       "      <td>0.000811</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...</td>\n",
       "      <td>0.001753</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.003049</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>MultiDAE_seed=42_e=10_bs=1024_intermediate_dim...</td>\n",
       "      <td>0.015473</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>ConvMF_seed=42_e=5_bs=32_lr=0$05_embedding_siz...</td>\n",
       "      <td>0.000811</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>ConvNeuMF_seed=42_e=5_bs=16_lr=0$05_embedding_...</td>\n",
       "      <td>0.005913</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>EASER_neighborhood=8444_l2_norm=3$72</td>\n",
       "      <td>0.135068</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>GeneralizedMF_seed=42_e=15_bs=512_lr=0$07_mffa...</td>\n",
       "      <td>0.000719</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.000437</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...</td>\n",
       "      <td>0.001620</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...</td>\n",
       "      <td>0.003260</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...</td>\n",
       "      <td>0.002641</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>ConvMF_seed=42_e=10_bs=32_lr=0$07_embedding_si...</td>\n",
       "      <td>0.000437</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.000612</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>EASER_neighborhood=10025_l2_norm=2$72</td>\n",
       "      <td>0.076518</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>56</th>\n",
       "      <td>GeneralizedMF_seed=42_e=15_bs=512_lr=0$5_mffac...</td>\n",
       "      <td>0.015951</td>\n",
       "      <td>movielens</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...</td>\n",
       "      <td>0.021963</td>\n",
       "      <td>movielens</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>58</th>\n",
       "      <td>MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...</td>\n",
       "      <td>0.006450</td>\n",
       "      <td>movielens</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>59</th>\n",
       "      <td>MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...</td>\n",
       "      <td>0.071482</td>\n",
       "      <td>movielens</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>60</th>\n",
       "      <td>ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...</td>\n",
       "      <td>0.037135</td>\n",
       "      <td>movielens</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61</th>\n",
       "      <td>ConvMF_seed=42_e=10_bs=32_lr=0$05_embedding_si...</td>\n",
       "      <td>0.021963</td>\n",
       "      <td>movielens</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>62</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.013511</td>\n",
       "      <td>movielens</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>63</th>\n",
       "      <td>EASER_neighborhood=2810_l2_norm=3$72</td>\n",
       "      <td>0.248742</td>\n",
       "      <td>movielens</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                model      nDCG experiment  \\\n",
       "8   GeneralizedMF_seed=42_e=10_bs=512_lr=0$5_mffac...  0.001836   epinions   \n",
       "9   NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...  0.000811   epinions   \n",
       "10  MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...  0.001753   epinions   \n",
       "11  NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.003049   epinions   \n",
       "12  MultiDAE_seed=42_e=10_bs=1024_intermediate_dim...  0.015473   epinions   \n",
       "13  ConvMF_seed=42_e=5_bs=32_lr=0$05_embedding_siz...  0.000811   epinions   \n",
       "14  ConvNeuMF_seed=42_e=5_bs=16_lr=0$05_embedding_...  0.005913   epinions   \n",
       "15               EASER_neighborhood=8444_l2_norm=3$72  0.135068   epinions   \n",
       "40  GeneralizedMF_seed=42_e=15_bs=512_lr=0$07_mffa...  0.000719     amazon   \n",
       "41  NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.000437     amazon   \n",
       "42  MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...  0.001620     amazon   \n",
       "43  MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...  0.003260     amazon   \n",
       "44  ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...  0.002641     amazon   \n",
       "45  ConvMF_seed=42_e=10_bs=32_lr=0$07_embedding_si...  0.000437     amazon   \n",
       "46  NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.000612     amazon   \n",
       "47              EASER_neighborhood=10025_l2_norm=2$72  0.076518     amazon   \n",
       "56  GeneralizedMF_seed=42_e=15_bs=512_lr=0$5_mffac...  0.015951  movielens   \n",
       "57  NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...  0.021963  movielens   \n",
       "58  MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...  0.006450  movielens   \n",
       "59  MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...  0.071482  movielens   \n",
       "60  ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...  0.037135  movielens   \n",
       "61  ConvMF_seed=42_e=10_bs=32_lr=0$05_embedding_si...  0.021963  movielens   \n",
       "62  NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.013511  movielens   \n",
       "63               EASER_neighborhood=2810_l2_norm=3$72  0.248742  movielens   \n",
       "\n",
       "    cutoff  \n",
       "8       10  \n",
       "9       10  \n",
       "10      10  \n",
       "11      10  \n",
       "12      10  \n",
       "13      10  \n",
       "14      10  \n",
       "15      10  \n",
       "40      10  \n",
       "41      10  \n",
       "42      10  \n",
       "43      10  \n",
       "44      10  \n",
       "45      10  \n",
       "46      10  \n",
       "47      10  \n",
       "56      10  \n",
       "57      10  \n",
       "58      10  \n",
       "59      10  \n",
       "60      10  \n",
       "61      10  \n",
       "62      10  \n",
       "63      10  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Disclaimer: This code cell was partly written with LLM (ChatGPT) support\n",
    "\n",
    "\n",
    "list_experiments = [\"Epinions/Untuned\",\"Amazon/Untuned\",\"Movielens_1m/Untuned\"]\n",
    "\n",
    "\n",
    "mapping_experiment_name = {\n",
    "    \"Epinions/Untuned\": \"epinions\",\n",
    "    \"Amazon/Untuned\": \"amazon\",\n",
    "    \"Movielens_1m/Untuned\": \"movielens\",\n",
    "}\n",
    "\n",
    "\n",
    "data_frames = []\n",
    "for experiment in list_experiments:\n",
    "\n",
    "    # Set the base directory where the run folders are stored\n",
    "    results_folder = os.path.join(\"Additional material\", experiment)\n",
    "\n",
    "    # List to collect dataframes from each file\n",
    "\n",
    "    # Loop through files in the run folder\n",
    "    for file_name in os.listdir(results_folder):\n",
    "        if file_name.endswith(\".tsv\"):\n",
    "            file_path = os.path.join(results_folder, file_name)\n",
    "            # Read the TSV file\n",
    "            df = pd.read_csv(file_path, sep=\"\\t\")\n",
    "            # Add columns for the run number and file name for later identification\n",
    "            df[\"experiment\"] = mapping_experiment_name[experiment]\n",
    "            df[\"file_name\"] = file_name\n",
    "            # Optionally, you could also add the run folder name:\n",
    "            df[\"results_folder\"] = results_folder\n",
    "\n",
    "\n",
    "            pattern = r\"cutoff_(\\d+)\"\n",
    "            match = re.search(pattern, file_name)\n",
    "            if match:\n",
    "                cutoff_value = int(match.group(1))\n",
    "                df[\"cutoff\"] = cutoff_value\n",
    "            \n",
    "            \n",
    "            data_frames.append(df)\n",
    "\n",
    "df_results = pd.concat(data_frames, ignore_index=True)\n",
    "\n",
    "\n",
    "# Remove 'run' as it is not present in df_results\n",
    "subset_columns.remove(\"run\")\n",
    "columns_to_keep.remove(\"run\")\n",
    "\n",
    "print(\"Number of duplicates:\" ,sum(df_results.duplicated(subset=subset_columns)))\n",
    "\n",
    "\n",
    "df_results = df_results[columns_to_keep]\n",
    "df_results = df_results[df_results[\"cutoff\"] == 10]\n",
    "\n",
    "df_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>model</th>\n",
       "      <th>nDCG</th>\n",
       "      <th>experiment</th>\n",
       "      <th>run</th>\n",
       "      <th>cutoff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1071</th>\n",
       "      <td>GeneralizedMF_seed=42_e=15_bs=512_lr=0$07_mffa...</td>\n",
       "      <td>0.003706</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1072</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.001065</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1073</th>\n",
       "      <td>MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...</td>\n",
       "      <td>0.002580</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1074</th>\n",
       "      <td>MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...</td>\n",
       "      <td>0.004335</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1075</th>\n",
       "      <td>ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...</td>\n",
       "      <td>0.002285</td>\n",
       "      <td>amazon</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1173</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.008221</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1174</th>\n",
       "      <td>MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...</td>\n",
       "      <td>0.012410</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1175</th>\n",
       "      <td>ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...</td>\n",
       "      <td>0.012007</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1176</th>\n",
       "      <td>ConvMF_seed=42_e=10_bs=32_lr=0$05_embedding_si...</td>\n",
       "      <td>0.012007</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1177</th>\n",
       "      <td>EASER_neighborhood=1017_l2_norm=3$72</td>\n",
       "      <td>0.020207</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>584 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                  model      nDCG experiment  \\\n",
       "1071  GeneralizedMF_seed=42_e=15_bs=512_lr=0$07_mffa...  0.003706     amazon   \n",
       "1072  NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.001065     amazon   \n",
       "1073  MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...  0.002580     amazon   \n",
       "1074  MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...  0.004335     amazon   \n",
       "1075  ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...  0.002285     amazon   \n",
       "...                                                 ...       ...        ...   \n",
       "1173  NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.008221  movielens   \n",
       "1174  MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...  0.012410  movielens   \n",
       "1175  ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...  0.012007  movielens   \n",
       "1176  ConvMF_seed=42_e=10_bs=32_lr=0$05_embedding_si...  0.012007  movielens   \n",
       "1177               EASER_neighborhood=1017_l2_norm=3$72  0.020207  movielens   \n",
       "\n",
       "      run  cutoff  \n",
       "1071    1      10  \n",
       "1072    1      10  \n",
       "1073    1      10  \n",
       "1074    1      10  \n",
       "1075    1      10  \n",
       "...   ...     ...  \n",
       "1173   30      10  \n",
       "1174   30      10  \n",
       "1175   30      10  \n",
       "1176   30      10  \n",
       "1177   30      10  \n",
       "\n",
       "[584 rows x 5 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_experiments"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Perform t-test"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test Epinions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>model</th>\n",
       "      <th>nDCG</th>\n",
       "      <th>experiment</th>\n",
       "      <th>cutoff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>EASER_neighborhood=8444_l2_norm=3$72</td>\n",
       "      <td>0.135068</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>MultiDAE_seed=42_e=10_bs=1024_intermediate_dim...</td>\n",
       "      <td>0.015473</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>ConvNeuMF_seed=42_e=5_bs=16_lr=0$05_embedding_...</td>\n",
       "      <td>0.005913</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.003049</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>GeneralizedMF_seed=42_e=10_bs=512_lr=0$5_mffac...</td>\n",
       "      <td>0.001836</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...</td>\n",
       "      <td>0.001753</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...</td>\n",
       "      <td>0.000811</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>ConvMF_seed=42_e=5_bs=32_lr=0$05_embedding_siz...</td>\n",
       "      <td>0.000811</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                model      nDCG experiment  \\\n",
       "15               EASER_neighborhood=8444_l2_norm=3$72  0.135068   epinions   \n",
       "12  MultiDAE_seed=42_e=10_bs=1024_intermediate_dim...  0.015473   epinions   \n",
       "14  ConvNeuMF_seed=42_e=5_bs=16_lr=0$05_embedding_...  0.005913   epinions   \n",
       "11  NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.003049   epinions   \n",
       "8   GeneralizedMF_seed=42_e=10_bs=512_lr=0$5_mffac...  0.001836   epinions   \n",
       "10  MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...  0.001753   epinions   \n",
       "9   NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...  0.000811   epinions   \n",
       "13  ConvMF_seed=42_e=5_bs=32_lr=0$05_embedding_siz...  0.000811   epinions   \n",
       "\n",
       "    cutoff  \n",
       "15      10  \n",
       "12      10  \n",
       "14      10  \n",
       "11      10  \n",
       "8       10  \n",
       "10      10  \n",
       "9       10  \n",
       "13      10  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_epinions_experiments = df_experiments[df_experiments[\"experiment\"] == \"epinions\"].copy()\n",
    "df_epinions_results = df_results[df_results[\"experiment\"] == \"epinions\"]\n",
    "df_epinions_results.sort_values(\"nDCG\", ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For the Epinions dataset, the authors also reported values for the tuned EASER model, which is a shallow model. However in the paper directly they did not report the results for the untuned model. Therefore we exclude it from the tests.\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_epinions_experiments = df_epinions_experiments[~df_epinions_experiments[\"model\"].str.startswith(\"EASER\")] \n",
    "df_epinions_results = df_epinions_results[~df_epinions_results[\"model\"].str.startswith(\"EASER\")]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1500x1200 with 7 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Disclaimer: This cell was partly generated with the support of a LLM (ChatGPT)\n",
    "\n",
    "\n",
    "models = df_epinions_experiments[\"model\"].unique()\n",
    "num_models = len(models)\n",
    "\n",
    "# Define subplot grid (adjust for readability)\n",
    "cols = 3  # Number of columns in the subplot grid\n",
    "rows = int(np.ceil(num_models / cols))  # Calculate rows dynamically\n",
    "\n",
    "# Create a figure with subplots\n",
    "fig, axes = plt.subplots(rows, cols, figsize=(15, rows * 4))  \n",
    "axes = axes.flatten()  # Flatten axes for easy iteration\n",
    "\n",
    "# Loop through each model and plot the histogram in the corresponding subplot\n",
    "for i, model in enumerate(models):\n",
    "    df_plot = df_epinions_experiments[df_epinions_experiments[\"model\"] == model]\n",
    "    title = model.split(\"_\")[0]  # Extract part of the model name\n",
    "\n",
    "    axes[i].hist(df_plot[\"nDCG\"], bins=10, alpha=0.7, edgecolor=\"black\")  # Improved histogram\n",
    "    axes[i].set_title(f\"Histogram {title}\", fontsize=12)\n",
    "    axes[i].set_xlabel(\"nDCG\")\n",
    "    axes[i].set_ylabel(\"Frequency\")\n",
    "    axes[i].grid(True)\n",
    "    axes[i].tick_params(axis = \"x\",rotation = 90)\n",
    "\n",
    "# Hide any unused subplots (if total models < grid size)\n",
    "for j in range(i + 1, len(axes)):\n",
    "    fig.delaxes(axes[j])\n",
    "\n",
    "plt.tight_layout()  # Adjust layout to prevent overlap\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "While no distribution seems truly normally distributed, none appear significantly far from a normal distribution either, except maybe the GeneralizedMF distrbution. Since the tests are quite robust, we decided to proceed with the t-test for the Epinions dataset. \n",
    "\n",
    "The distributions for *ConvMF*, *MultiVAE* and *NeuMF* seem almost identical. Looking at the data through DataWrangler we can also observe that almost all values are the exact same. \n",
    "\n",
    "\n",
    "While trying to match the models from both datasets, we realized that some parameters used for the results model differ from those in the `untuned_epinions` config file (e.g., the number of epochs for MultiDAE). This should be criticized because the repository itself suggests that the results can be reproduced without any deeper configuration through the user, which is not entirely the case. As the runs where quite expensive we decide to keep our results but also keep this in mind as a limitation.\n",
    "\n",
    "\n",
    "We now perform the test. Our H0 is that our sample mean is equal to the reported results which is interpreted as the population mean. Our H1 is that they are not equal. More formally:\n",
    "\n",
    "$$\n",
    "H_0: \\ \\mu_{sample} = \\mu_0\n",
    "$$\n",
    "\n",
    "$$\n",
    "H_1: \\ \\mu_{sample} \\neq \\mu_0\n",
    "$$\n",
    "\n",
    "We perform the test at a significance level of $\\alpha = 0.05$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>Model</th>\n",
       "      <th>Mean (Sample)</th>\n",
       "      <th>SD (Sample)</th>\n",
       "      <th>Pop Mean (nDCG)</th>\n",
       "      <th>T-Statistic</th>\n",
       "      <th>p-Value</th>\n",
       "      <th>Degrees of Freedom</th>\n",
       "      <th>Significant (p &lt; 0.05)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>MultiDAE</td>\n",
       "      <td>0.01253</td>\n",
       "      <td>0.00195</td>\n",
       "      <td>0.01547</td>\n",
       "      <td>-8.12311</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>ConvNeuMF</td>\n",
       "      <td>0.00320</td>\n",
       "      <td>0.00062</td>\n",
       "      <td>0.00591</td>\n",
       "      <td>-23.57500</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GeneralizedMF</td>\n",
       "      <td>0.00233</td>\n",
       "      <td>0.00036</td>\n",
       "      <td>0.00184</td>\n",
       "      <td>7.41918</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>NGCF</td>\n",
       "      <td>0.00187</td>\n",
       "      <td>0.00016</td>\n",
       "      <td>0.00305</td>\n",
       "      <td>-39.14303</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NeuMF</td>\n",
       "      <td>0.00155</td>\n",
       "      <td>0.00047</td>\n",
       "      <td>0.00081</td>\n",
       "      <td>8.49854</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>MultiVAE</td>\n",
       "      <td>0.00155</td>\n",
       "      <td>0.00047</td>\n",
       "      <td>0.00175</td>\n",
       "      <td>-2.35932</td>\n",
       "      <td>0.02526</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>ConvMF</td>\n",
       "      <td>0.00155</td>\n",
       "      <td>0.00047</td>\n",
       "      <td>0.00081</td>\n",
       "      <td>8.49854</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           Model  Mean (Sample)  SD (Sample)  Pop Mean (nDCG)  T-Statistic  \\\n",
       "4       MultiDAE        0.01253      0.00195          0.01547     -8.12311   \n",
       "5      ConvNeuMF        0.00320      0.00062          0.00591    -23.57500   \n",
       "0  GeneralizedMF        0.00233      0.00036          0.00184      7.41918   \n",
       "3           NGCF        0.00187      0.00016          0.00305    -39.14303   \n",
       "1          NeuMF        0.00155      0.00047          0.00081      8.49854   \n",
       "2       MultiVAE        0.00155      0.00047          0.00175     -2.35932   \n",
       "6         ConvMF        0.00155      0.00047          0.00081      8.49854   \n",
       "\n",
       "   p-Value  Degrees of Freedom Significant (p < 0.05)  \n",
       "4  0.00000                  29                      ✅  \n",
       "5  0.00000                  29                      ✅  \n",
       "0  0.00000                  29                      ✅  \n",
       "3  0.00000                  29                      ✅  \n",
       "1  0.00000                  29                      ✅  \n",
       "2  0.02526                  29                      ✅  \n",
       "6  0.00000                  29                      ✅  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "# Disclaimer: This Notebook Cell was partly created with an LLM (ChatGPT). It was used to improve the visual output of the test.\n",
    "\n",
    "\n",
    "# To get rid of pandas warnings\n",
    "df_epinions_experiments = df_epinions_experiments.copy()\n",
    "df_epinions_results = df_epinions_results.copy()\n",
    "df_epinions_experiments.loc[:,\"model_name\"] = df_epinions_experiments.loc[:,\"model\"].str.split(\"_\").str[0]\n",
    "df_epinions_results.loc[:,\"model_name\"] = df_epinions_results.loc[:,\"model\"].str.split(\"_\").str[0]\n",
    "\n",
    "# Create a list to store results\n",
    "ttest_results = []\n",
    "\n",
    "for model in df_epinions_experiments[\"model_name\"].unique():\n",
    "    result = df_epinions_results.loc[df_epinions_results[\"model_name\"] == model, \"nDCG\"].item() \n",
    "    samples = df_epinions_experiments.loc[df_epinions_experiments[\"model_name\"] == model, \"nDCG\"].to_numpy()\n",
    "    \n",
    "    # Perform t-test\n",
    "    ttest_result = ttest_1samp(a=samples, popmean=result)\n",
    "    \n",
    "    # Store formatted results\n",
    "    ttest_results.append({\n",
    "        \"Model\": model,\n",
    "        \"Mean (Sample)\": np.round(np.mean(samples), 5),\n",
    "        \"SD (Sample)\" : np.round(np.std(samples),5),\n",
    "        \"Pop Mean (nDCG)\": np.round(result, 5),\n",
    "        \"T-Statistic\": np.round(ttest_result.statistic, 5),\n",
    "        \"p-Value\": np.round(ttest_result.pvalue, 5),\n",
    "        \"Degrees of Freedom\":(ttest_result.df),\n",
    "        \"Significant (p < 0.05)\": \"✅\" if ttest_result.pvalue < 0.05 else \"❌\"\n",
    "    })\n",
    "\n",
    "# Convert to DataFrame for better display\n",
    "df_ttest_results = pd.DataFrame(ttest_results)\n",
    "\n",
    "df_ttest_results.sort_values(by = \"Mean (Sample)\", ascending=False)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can reject $H_0$ for all models meaning our results are siginficantly different from the ones reported in the paper. See the report for more details.\n",
    "\n",
    "Finally check which parameters differed between the models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==============================\n",
      "GeneralizedMF\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "NeuMF\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "MultiVAE\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "NGCF\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "MultiDAE\n",
      "Differences found:\n",
      "\n",
      "Parameter: e\n",
      "   - Config File: 5\n",
      "\n",
      "Parameter: e\n",
      "   - Results File: 10\n",
      "\n",
      "==============================\n",
      "ConvNeuMF\n",
      "Differences found:\n",
      "\n",
      "Parameter: e\n",
      "   - Config File: 10\n",
      "\n",
      "Parameter: e\n",
      "   - Results File: 5\n",
      "\n",
      "==============================\n",
      "ConvMF\n",
      "Differences found:\n",
      "\n",
      "Parameter: e\n",
      "   - Config File: 10\n",
      "\n",
      "Parameter: e\n",
      "   - Results File: 5\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Disclaimer: This cell was partly generated with the support of a LLM (ChatGPT)\n",
    "\n",
    "\n",
    "for model in df_epinions_experiments[\"model_name\"].unique():\n",
    "\n",
    "    \n",
    "    \n",
    "    print(\"=\" * 30)\n",
    "    print(model)\n",
    "\n",
    "    model1 = df_epinions_experiments.loc[df_epinions_experiments[\"model_name\"] == model, \"model\"].unique().item()\n",
    "    model2 = df_epinions_results.loc[df_epinions_results[\"model_name\"] == model, \"model\"].unique().item()\n",
    "\n",
    "    # Use difflib to compare\n",
    "    diff = list(difflib.ndiff(model1.split(\"_\"), model2.split(\"_\")))\n",
    "\n",
    "    # Extract and print differences\n",
    "    differences = []\n",
    "    for d in diff:\n",
    "        if d.startswith('- ') or d.startswith('+ '):\n",
    "            key = d[2:].split(\"=\")[0] if \"=\" in d else \"Unknown\"\n",
    "            value = d[2:].split(\"=\")[1] if \"=\" in d else d[2:]\n",
    "            model_source = \"Config File\" if d.startswith(\"- \") else \"Results File\"\n",
    "            differences.append((key, value, model_source))\n",
    "\n",
    "    # Print results\n",
    "    if differences:\n",
    "        print(\"Differences found:\\n\")\n",
    "        for param, value, model_source in differences:\n",
    "            print(f\"Parameter: {param}\")\n",
    "            print(f\"   - {model_source}: {value}\\n\")\n",
    "    else:\n",
    "        print(\"No Differences Found Between Models\")\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GeneralizedMF_seed=42_e=10_bs=512_lr=0$5_mffactors=256_isedgeweighttrain=True\n",
      "1\n",
      "NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256_drop=0_batch_eval=256_mftrain=True_mlptrain=True_m=8\n",
      "1\n",
      "MultiVAE_seed=42_e=10_bs=256_intermediate_dim=800_latent_dim=400_reg_lambda=0$6_lr=0$6_dropout_pkeep=0$5\n",
      "1\n",
      "NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_w=0$4_weight_size=64_node_dropout=_message_dropout=0$4_n_fold=5\n",
      "1\n",
      "MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=600_latent_dim=200_reg_lambda=0$6_lr=0$7_dropout_pkeep=0$0\n",
      "0\n",
      "ConvNeuMF_seed=42_e=10_bs=16_lr=0$05_embedding_size=50_cnn_channels=1-16-16_cnn_kernels=2-2_cnn_strides=2-2_dropout_prob=0$5_l_w=0$08_l_b=0$07\n",
      "0\n",
      "ConvMF_seed=42_e=10_bs=32_lr=0$05_embedding_size=100_cnn_channels=1-32-32_cnn_kernels=2-2_cnn_strides=2-2_dropout_prob=0$4_l_w=0$03_l_b=0$04\n",
      "0\n"
     ]
    }
   ],
   "source": [
    "for model in models:\n",
    "    print(model)\n",
    "    print(sum(df_epinions_results[\"model\"] == model))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'+ 5\\n- 1\\n- 0'"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Define the two parameter strings\n",
    "param_str1 = \"ConvNeuMF_seed=42_e=10_bs=16_lr=0$05_embedding_size=50_cnn_channels=1-16-16_cnn_kernels=2-2_cnn_strides=2-2_dropout_prob=0$5_l_w=0$08_l_b=0$07\"\n",
    "param_str2 = \"ConvNeuMF_seed=42_e=5_bs=16_lr=0$05_embedding_size=50_cnn_channels=1-16-16_cnn_kernels=2-2_cnn_strides=2-2_dropout_prob=0$5_l_w=0$08_l_b=0$07\"\n",
    "\n",
    "# Use difflib to compare\n",
    "diff = list(difflib.ndiff(param_str1, param_str2))\n",
    "\n",
    "# Extract differences\n",
    "differences = [d for d in diff if d.startswith('- ') or d.startswith('+ ')]\n",
    "\n",
    "# Format the results\n",
    "diff_result = \"\\n\".join(differences)\n",
    "diff_result\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>model</th>\n",
       "      <th>nDCG</th>\n",
       "      <th>experiment</th>\n",
       "      <th>cutoff</th>\n",
       "      <th>model_name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>GeneralizedMF_seed=42_e=10_bs=512_lr=0$5_mffac...</td>\n",
       "      <td>0.001836</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "      <td>GeneralizedMF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...</td>\n",
       "      <td>0.000811</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "      <td>NeuMF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...</td>\n",
       "      <td>0.001753</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "      <td>MultiVAE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.003049</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "      <td>NGCF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>MultiDAE_seed=42_e=10_bs=1024_intermediate_dim...</td>\n",
       "      <td>0.015473</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "      <td>MultiDAE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>ConvMF_seed=42_e=5_bs=32_lr=0$05_embedding_siz...</td>\n",
       "      <td>0.000811</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "      <td>ConvMF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>ConvNeuMF_seed=42_e=5_bs=16_lr=0$05_embedding_...</td>\n",
       "      <td>0.005913</td>\n",
       "      <td>epinions</td>\n",
       "      <td>10</td>\n",
       "      <td>ConvNeuMF</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                model      nDCG experiment  \\\n",
       "8   GeneralizedMF_seed=42_e=10_bs=512_lr=0$5_mffac...  0.001836   epinions   \n",
       "9   NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...  0.000811   epinions   \n",
       "10  MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...  0.001753   epinions   \n",
       "11  NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.003049   epinions   \n",
       "12  MultiDAE_seed=42_e=10_bs=1024_intermediate_dim...  0.015473   epinions   \n",
       "13  ConvMF_seed=42_e=5_bs=32_lr=0$05_embedding_siz...  0.000811   epinions   \n",
       "14  ConvNeuMF_seed=42_e=5_bs=16_lr=0$05_embedding_...  0.005913   epinions   \n",
       "\n",
       "    cutoff     model_name  \n",
       "8       10  GeneralizedMF  \n",
       "9       10          NeuMF  \n",
       "10      10       MultiVAE  \n",
       "11      10           NGCF  \n",
       "12      10       MultiDAE  \n",
       "13      10         ConvMF  \n",
       "14      10      ConvNeuMF  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_epinions_results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test MovieLens\n",
    "\n",
    "The same procedure as before is repeated"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "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>model</th>\n",
       "      <th>nDCG</th>\n",
       "      <th>experiment</th>\n",
       "      <th>run</th>\n",
       "      <th>cutoff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1864</th>\n",
       "      <td>GeneralizedMF_seed=42_e=15_bs=512_lr=0$5_mffac...</td>\n",
       "      <td>0.010426</td>\n",
       "      <td>movielens</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1865</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...</td>\n",
       "      <td>0.011200</td>\n",
       "      <td>movielens</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1866</th>\n",
       "      <td>MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...</td>\n",
       "      <td>0.011165</td>\n",
       "      <td>movielens</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1867</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.009731</td>\n",
       "      <td>movielens</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1868</th>\n",
       "      <td>MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...</td>\n",
       "      <td>0.016083</td>\n",
       "      <td>movielens</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1173</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.008221</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1174</th>\n",
       "      <td>MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...</td>\n",
       "      <td>0.012410</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1175</th>\n",
       "      <td>ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...</td>\n",
       "      <td>0.012007</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1176</th>\n",
       "      <td>ConvMF_seed=42_e=10_bs=32_lr=0$05_embedding_si...</td>\n",
       "      <td>0.012007</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1177</th>\n",
       "      <td>EASER_neighborhood=1017_l2_norm=3$72</td>\n",
       "      <td>0.020207</td>\n",
       "      <td>movielens</td>\n",
       "      <td>30</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>240 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                  model      nDCG experiment  \\\n",
       "1864  GeneralizedMF_seed=42_e=15_bs=512_lr=0$5_mffac...  0.010426  movielens   \n",
       "1865  NeuMF_seed=42_e=10_bs=256_lr=0$5_mffactors=256...  0.011200  movielens   \n",
       "1866  MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...  0.011165  movielens   \n",
       "1867  NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.009731  movielens   \n",
       "1868  MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...  0.016083  movielens   \n",
       "...                                                 ...       ...        ...   \n",
       "1173  NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.008221  movielens   \n",
       "1174  MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...  0.012410  movielens   \n",
       "1175  ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...  0.012007  movielens   \n",
       "1176  ConvMF_seed=42_e=10_bs=32_lr=0$05_embedding_si...  0.012007  movielens   \n",
       "1177               EASER_neighborhood=1017_l2_norm=3$72  0.020207  movielens   \n",
       "\n",
       "      run  cutoff  \n",
       "1864    1      10  \n",
       "1865    1      10  \n",
       "1866    1      10  \n",
       "1867    1      10  \n",
       "1868    1      10  \n",
       "...   ...     ...  \n",
       "1173   30      10  \n",
       "1174   30      10  \n",
       "1175   30      10  \n",
       "1176   30      10  \n",
       "1177   30      10  \n",
       "\n",
       "[240 rows x 5 columns]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_movielens_experiments = df_experiments[df_experiments[\"experiment\"] == \"movielens\"].copy()\n",
    "df_movielens_results = df_results[df_results[\"experiment\"] == \"movielens\"]\n",
    "df_movielens_experiments"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Remove EASER"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_movielens_experiments = df_movielens_experiments[~df_movielens_experiments[\"model\"].str.startswith(\"EASER\")] \n",
    "df_movielens_results = df_movielens_results[~df_movielens_results[\"model\"].str.startswith(\"EASER\")]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check distributions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1500x1200 with 7 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Disclaimer: This cell was partly generated with the support of a LLM (ChatGPT)\n",
    "\n",
    "\n",
    "models = df_movielens_experiments[\"model\"].unique()\n",
    "num_models = len(models)\n",
    "\n",
    "# Define subplot grid (adjust for readability)\n",
    "cols = 3  # Number of columns in the subplot grid\n",
    "rows = int(np.ceil(num_models / cols))  # Calculate rows dynamically\n",
    "\n",
    "# Create a figure with subplots\n",
    "fig, axes = plt.subplots(rows, cols, figsize=(15, rows * 4))  \n",
    "axes = axes.flatten()  # Flatten axes for easy iteration\n",
    "\n",
    "# Loop through each model and plot the histogram in the corresponding subplot\n",
    "for i, model in enumerate(models):\n",
    "    df_plot = df_movielens_experiments[df_movielens_experiments[\"model\"] == model]\n",
    "    title = model.split(\"_\")[0]  # Extract part of the model name\n",
    "\n",
    "    axes[i].hist(df_plot[\"nDCG\"], bins=10, alpha=0.7, edgecolor=\"black\")  # Improved histogram\n",
    "    axes[i].set_title(f\"Histogram {title}\", fontsize=12)\n",
    "    axes[i].set_xlabel(\"nDCG\")\n",
    "    axes[i].set_ylabel(\"Frequency\")\n",
    "    axes[i].grid(True)\n",
    "    axes[i].tick_params(axis = \"x\",rotation = 90)\n",
    "\n",
    "# Hide any unused subplots (if total models < grid size)\n",
    "for j in range(i + 1, len(axes)):\n",
    "    fig.delaxes(axes[j])\n",
    "\n",
    "plt.tight_layout()  # Adjust layout to prevent overlap\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The results look mostly acceptable again. Again *ConvMF* and *NeuMF* have very similiar distributions.\n",
    "\n",
    "Perform ttest: \n",
    "\n",
    "$$\n",
    "H_0: \\ \\mu_{sample} = \\mu_0\n",
    "$$\n",
    "\n",
    "$$\n",
    "H_1: \\ \\mu_{sample} \\neq \\mu_0\n",
    "$$\n",
    "\n",
    "We perform the test at a significance level of $\\alpha = 0.05$\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>Model</th>\n",
       "      <th>Mean (Sample)</th>\n",
       "      <th>SD (Sample)</th>\n",
       "      <th>Pop Mean (nDCG)</th>\n",
       "      <th>T-Statistic</th>\n",
       "      <th>p-Value</th>\n",
       "      <th>Degrees of Freedom</th>\n",
       "      <th>Significant (p &lt; 0.05)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>MultiDAE</td>\n",
       "      <td>0.01726</td>\n",
       "      <td>0.00260</td>\n",
       "      <td>0.07148</td>\n",
       "      <td>-112.35716</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>ConvNeuMF</td>\n",
       "      <td>0.01103</td>\n",
       "      <td>0.00091</td>\n",
       "      <td>0.03713</td>\n",
       "      <td>-154.20034</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NeuMF</td>\n",
       "      <td>0.01098</td>\n",
       "      <td>0.00099</td>\n",
       "      <td>0.02196</td>\n",
       "      <td>-59.63629</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>ConvMF</td>\n",
       "      <td>0.01098</td>\n",
       "      <td>0.00099</td>\n",
       "      <td>0.02196</td>\n",
       "      <td>-59.63629</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>MultiVAE</td>\n",
       "      <td>0.01085</td>\n",
       "      <td>0.00100</td>\n",
       "      <td>0.00645</td>\n",
       "      <td>23.80201</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GeneralizedMF</td>\n",
       "      <td>0.01033</td>\n",
       "      <td>0.00077</td>\n",
       "      <td>0.01595</td>\n",
       "      <td>-39.17640</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>NGCF</td>\n",
       "      <td>0.00910</td>\n",
       "      <td>0.00047</td>\n",
       "      <td>0.01351</td>\n",
       "      <td>-50.35295</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           Model  Mean (Sample)  SD (Sample)  Pop Mean (nDCG)  T-Statistic  \\\n",
       "4       MultiDAE        0.01726      0.00260          0.07148   -112.35716   \n",
       "5      ConvNeuMF        0.01103      0.00091          0.03713   -154.20034   \n",
       "1          NeuMF        0.01098      0.00099          0.02196    -59.63629   \n",
       "6         ConvMF        0.01098      0.00099          0.02196    -59.63629   \n",
       "2       MultiVAE        0.01085      0.00100          0.00645     23.80201   \n",
       "0  GeneralizedMF        0.01033      0.00077          0.01595    -39.17640   \n",
       "3           NGCF        0.00910      0.00047          0.01351    -50.35295   \n",
       "\n",
       "   p-Value  Degrees of Freedom Significant (p < 0.05)  \n",
       "4      0.0                  29                      ✅  \n",
       "5      0.0                  29                      ✅  \n",
       "1      0.0                  29                      ✅  \n",
       "6      0.0                  29                      ✅  \n",
       "2      0.0                  29                      ✅  \n",
       "0      0.0                  29                      ✅  \n",
       "3      0.0                  29                      ✅  "
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "# Disclaimer: This Notebook Cell was partly created with an LLM (ChatGPT). It was used to improve the visual output of the test.\n",
    "\n",
    "\n",
    "# To get rid of pandas warnings\n",
    "df_movielens_experiments = df_movielens_experiments.copy()\n",
    "df_movielens_results = df_movielens_results.copy()\n",
    "df_movielens_experiments.loc[:,\"model_name\"] = df_movielens_experiments.loc[:,\"model\"].str.split(\"_\").str[0]\n",
    "df_movielens_results.loc[:,\"model_name\"] = df_movielens_results.loc[:,\"model\"].str.split(\"_\").str[0]\n",
    "\n",
    "# Create a list to store results\n",
    "ttest_results = []\n",
    "\n",
    "for model in df_movielens_experiments[\"model_name\"].unique():\n",
    "    result = df_movielens_results.loc[df_movielens_results[\"model_name\"] == model, \"nDCG\"].item() \n",
    "    samples = df_movielens_experiments.loc[df_movielens_experiments[\"model_name\"] == model, \"nDCG\"].to_numpy()\n",
    "    \n",
    "    # Perform t-test\n",
    "    ttest_result = ttest_1samp(a=samples, popmean=result)\n",
    "    \n",
    "    # Store formatted results\n",
    "    ttest_results.append({\n",
    "        \"Model\": model,\n",
    "        \"Mean (Sample)\": np.round(np.mean(samples), 5),\n",
    "        \"SD (Sample)\" : np.round(np.std(samples),5),\n",
    "        \"Pop Mean (nDCG)\": np.round(result, 5),\n",
    "        \"T-Statistic\": np.round(ttest_result.statistic, 5),\n",
    "        \"p-Value\": np.round(ttest_result.pvalue, 5),\n",
    "        \"Degrees of Freedom\":(ttest_result.df),\n",
    "        \"Significant (p < 0.05)\": \"✅\" if ttest_result.pvalue < 0.05 else \"❌\"\n",
    "    })\n",
    "\n",
    "# Convert to DataFrame for better display\n",
    "df_ttest_results = pd.DataFrame(ttest_results)\n",
    "\n",
    "df_ttest_results.sort_values(by = \"Mean (Sample)\", ascending=False)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Again we can reject $H_0$ for all models, meaning our results are significantly different than the ones reported in the paper. \n",
    "\n",
    "\n",
    "Compare possible differences between the parameters of the models:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==============================\n",
      "GeneralizedMF\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "NeuMF\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "MultiVAE\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "NGCF\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "MultiDAE\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "ConvNeuMF\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "ConvMF\n",
      "No Differences Found Between Models\n"
     ]
    }
   ],
   "source": [
    "# Disclaimer: This cell was partly generated with the support of a LLM (ChatGPT)\n",
    "\n",
    "\n",
    "for model in df_movielens_experiments[\"model_name\"].unique():\n",
    "    \n",
    "    print(\"=\" * 30)\n",
    "    print(model)\n",
    "\n",
    "    model1 = df_movielens_experiments.loc[df_movielens_experiments[\"model_name\"] == model, \"model\"].unique().item()\n",
    "    model2 = df_movielens_results.loc[df_movielens_results[\"model_name\"] == model, \"model\"].unique().item()\n",
    "\n",
    "    # Use difflib to compare\n",
    "    diff = list(difflib.ndiff(model1.split(\"_\"), model2.split(\"_\")))\n",
    "\n",
    "    # Extract and print differences\n",
    "    differences = []\n",
    "    for d in diff:\n",
    "        if d.startswith('- ') or d.startswith('+ '):\n",
    "            key = d[2:].split(\"=\")[0] if \"=\" in d else \"Unknown\"\n",
    "            value = d[2:].split(\"=\")[1] if \"=\" in d else d[2:]\n",
    "            model_source = \"Config File\" if d.startswith(\"- \") else \"Results File\"\n",
    "            differences.append((key, value, model_source))\n",
    "\n",
    "    # Print results\n",
    "    if differences:\n",
    "        print(\"Differences found:\\n\")\n",
    "        for param, value, model_source in differences:\n",
    "            print(f\"Parameter: {param}\")\n",
    "            print(f\"   - {model_source}: {value}\\n\")\n",
    "    else:\n",
    "        print(\"No Differences Found Between Models\")\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test Amazon\n",
    "\n",
    "\n",
    "Here we have less samples, as the runs failed multipled times and took longer than for the other datasets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "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>model</th>\n",
       "      <th>nDCG</th>\n",
       "      <th>experiment</th>\n",
       "      <th>cutoff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>EASER_neighborhood=10025_l2_norm=2$72</td>\n",
       "      <td>0.076518</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...</td>\n",
       "      <td>0.003260</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...</td>\n",
       "      <td>0.002641</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...</td>\n",
       "      <td>0.001620</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>GeneralizedMF_seed=42_e=15_bs=512_lr=0$07_mffa...</td>\n",
       "      <td>0.000719</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...</td>\n",
       "      <td>0.000612</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...</td>\n",
       "      <td>0.000437</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>ConvMF_seed=42_e=10_bs=32_lr=0$07_embedding_si...</td>\n",
       "      <td>0.000437</td>\n",
       "      <td>amazon</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                model      nDCG experiment  \\\n",
       "47              EASER_neighborhood=10025_l2_norm=2$72  0.076518     amazon   \n",
       "43  MultiDAE_seed=42_e=5_bs=1024_intermediate_dim=...  0.003260     amazon   \n",
       "44  ConvNeuMF_seed=42_e=10_bs=16_lr=0$06_embedding...  0.002641     amazon   \n",
       "42  MultiVAE_seed=42_e=10_bs=256_intermediate_dim=...  0.001620     amazon   \n",
       "40  GeneralizedMF_seed=42_e=15_bs=512_lr=0$07_mffa...  0.000719     amazon   \n",
       "46  NGCF_seed=42_e=15_bs=512_lr=0$09_factors=64_l_...  0.000612     amazon   \n",
       "41  NeuMF_seed=42_e=10_bs=256_lr=0$08_mffactors=25...  0.000437     amazon   \n",
       "45  ConvMF_seed=42_e=10_bs=32_lr=0$07_embedding_si...  0.000437     amazon   \n",
       "\n",
       "    cutoff  \n",
       "47      10  \n",
       "43      10  \n",
       "44      10  \n",
       "42      10  \n",
       "40      10  \n",
       "46      10  \n",
       "41      10  \n",
       "45      10  "
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_amazon_experiments = df_experiments[df_experiments[\"experiment\"] == \"amazon\"].copy()\n",
    "df_amazon_results = df_results[df_results[\"experiment\"] == \"amazon\"]\n",
    "df_amazon_results.sort_values(\"nDCG\", ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Remove EASER"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_amazon_experiments = df_amazon_experiments[~df_amazon_experiments[\"model\"].str.startswith(\"EASER\")] \n",
    "df_amazon_results = df_amazon_results[~df_amazon_results[\"model\"].str.startswith(\"EASER\")]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Look at distributions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1500x1200 with 7 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Disclaimer: This cell was partly generated with the support of a LLM (ChatGPT)\n",
    "\n",
    "\n",
    "models = df_amazon_experiments[\"model\"].unique()\n",
    "num_models = len(models)\n",
    "\n",
    "# Define subplot grid (adjust for readability)\n",
    "cols = 3  # Number of columns in the subplot grid\n",
    "rows = int(np.ceil(num_models / cols))  # Calculate rows dynamically\n",
    "\n",
    "# Create a figure with subplots\n",
    "fig, axes = plt.subplots(rows, cols, figsize=(15, rows * 4))  \n",
    "axes = axes.flatten()  # Flatten axes for easy iteration\n",
    "\n",
    "# Loop through each model and plot the histogram in the corresponding subplot\n",
    "for i, model in enumerate(models):\n",
    "    df_plot = df_amazon_experiments[df_amazon_experiments[\"model\"] == model]\n",
    "    title = model.split(\"_\")[0]  # Extract part of the model name\n",
    "\n",
    "    axes[i].hist(df_plot[\"nDCG\"], bins=10, alpha=0.7, edgecolor=\"black\")  # Improved histogram\n",
    "    axes[i].set_title(f\"Histogram {title}\", fontsize=12)\n",
    "    axes[i].set_xlabel(\"nDCG\")\n",
    "    axes[i].set_ylabel(\"Frequency\")\n",
    "    axes[i].grid(True)\n",
    "    axes[i].tick_params(axis = \"x\",rotation = 90)\n",
    "\n",
    "# Hide any unused subplots (if total models < grid size)\n",
    "for j in range(i + 1, len(axes)):\n",
    "    fig.delaxes(axes[j])\n",
    "\n",
    "plt.tight_layout()  # Adjust layout to prevent overlap\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In these distributions, we can detect some potential outliers, such as in *ConvMF* or *NeuMF* (equal distribution again), which would be another violation of the t-test assumptions. However, as mentioned in the slides, the experiments are robust, and we decided to violate/ignore this assumption in our decision-making process.  \n",
    "\n",
    "Perform t-test:\n",
    "\n",
    "$$\n",
    "H_0: \\ \\mu_{sample} = \\mu_0\n",
    "$$\n",
    "\n",
    "$$\n",
    "H_1: \\ \\mu_{sample} \\neq \\mu_0\n",
    "$$\n",
    "\n",
    "We perform the test at a significance level of $\\alpha = 0.05$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "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>Model</th>\n",
       "      <th>Mean (Sample)</th>\n",
       "      <th>SD (Sample)</th>\n",
       "      <th>Pop Mean (nDCG)</th>\n",
       "      <th>T-Statistic</th>\n",
       "      <th>p-Value</th>\n",
       "      <th>Degrees of Freedom</th>\n",
       "      <th>Significant (p &lt; 0.05)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>MultiDAE</td>\n",
       "      <td>0.00610</td>\n",
       "      <td>0.00178</td>\n",
       "      <td>0.00326</td>\n",
       "      <td>5.53362</td>\n",
       "      <td>0.00013</td>\n",
       "      <td>12</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GeneralizedMF</td>\n",
       "      <td>0.00438</td>\n",
       "      <td>0.00050</td>\n",
       "      <td>0.00072</td>\n",
       "      <td>25.23671</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>12</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>MultiVAE</td>\n",
       "      <td>0.00402</td>\n",
       "      <td>0.00092</td>\n",
       "      <td>0.00162</td>\n",
       "      <td>9.08045</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>12</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ConvNeuMF</td>\n",
       "      <td>0.00239</td>\n",
       "      <td>0.00052</td>\n",
       "      <td>0.00264</td>\n",
       "      <td>-1.69575</td>\n",
       "      <td>0.11570</td>\n",
       "      <td>12</td>\n",
       "      <td>❌</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>NGCF</td>\n",
       "      <td>0.00180</td>\n",
       "      <td>0.00026</td>\n",
       "      <td>0.00061</td>\n",
       "      <td>16.11010</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>12</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NeuMF</td>\n",
       "      <td>0.00105</td>\n",
       "      <td>0.00013</td>\n",
       "      <td>0.00044</td>\n",
       "      <td>16.08625</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>12</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>ConvMF</td>\n",
       "      <td>0.00105</td>\n",
       "      <td>0.00013</td>\n",
       "      <td>0.00044</td>\n",
       "      <td>16.08625</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>12</td>\n",
       "      <td>✅</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           Model  Mean (Sample)  SD (Sample)  Pop Mean (nDCG)  T-Statistic  \\\n",
       "3       MultiDAE        0.00610      0.00178          0.00326      5.53362   \n",
       "0  GeneralizedMF        0.00438      0.00050          0.00072     25.23671   \n",
       "2       MultiVAE        0.00402      0.00092          0.00162      9.08045   \n",
       "4      ConvNeuMF        0.00239      0.00052          0.00264     -1.69575   \n",
       "6           NGCF        0.00180      0.00026          0.00061     16.11010   \n",
       "1          NeuMF        0.00105      0.00013          0.00044     16.08625   \n",
       "5         ConvMF        0.00105      0.00013          0.00044     16.08625   \n",
       "\n",
       "   p-Value  Degrees of Freedom Significant (p < 0.05)  \n",
       "3  0.00013                  12                      ✅  \n",
       "0  0.00000                  12                      ✅  \n",
       "2  0.00000                  12                      ✅  \n",
       "4  0.11570                  12                      ❌  \n",
       "6  0.00000                  12                      ✅  \n",
       "1  0.00000                  12                      ✅  \n",
       "5  0.00000                  12                      ✅  "
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "# Disclaimer: This Notebook Cell was partly created with an LLM (ChatGPT). It was used to improve the visual output of the test.\n",
    "\n",
    "\n",
    "# To get rid of pandas warnings\n",
    "df_amazon_experiments = df_amazon_experiments.copy()\n",
    "df_amazon_results = df_amazon_results.copy()\n",
    "df_amazon_experiments.loc[:,\"model_name\"] = df_amazon_experiments.loc[:,\"model\"].str.split(\"_\").str[0]\n",
    "df_amazon_results.loc[:,\"model_name\"] = df_amazon_results.loc[:,\"model\"].str.split(\"_\").str[0]\n",
    "\n",
    "# Create a list to store results\n",
    "ttest_results = []\n",
    "\n",
    "for model in df_amazon_experiments[\"model_name\"].unique():\n",
    "    result = df_amazon_results.loc[df_amazon_results[\"model_name\"] == model, \"nDCG\"].item() \n",
    "    samples = df_amazon_experiments.loc[df_amazon_experiments[\"model_name\"] == model, \"nDCG\"].to_numpy()\n",
    "    \n",
    "    # Perform t-test\n",
    "    ttest_result = ttest_1samp(a=samples, popmean=result)\n",
    "    \n",
    "    # Store formatted results\n",
    "    ttest_results.append({\n",
    "        \"Model\": model,\n",
    "        \"Mean (Sample)\": np.round(np.mean(samples), 5),\n",
    "        \"SD (Sample)\" : np.round(np.std(samples),5),\n",
    "        \"Pop Mean (nDCG)\": np.round(result, 5),\n",
    "        \"T-Statistic\": np.round(ttest_result.statistic, 5),\n",
    "        \"p-Value\": np.round(ttest_result.pvalue, 5),\n",
    "        \"Degrees of Freedom\":(ttest_result.df),\n",
    "        \"Significant (p < 0.05)\": \"✅\" if ttest_result.pvalue < 0.05 else \"❌\"\n",
    "    })\n",
    "\n",
    "# Convert to DataFrame for better display\n",
    "df_ttest_results = pd.DataFrame(ttest_results)\n",
    "\n",
    "df_ttest_results.sort_values(\"Mean (Sample)\", ascending = False)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can reject $H_0$ except for ConvNeuMF where the difference is not siginificant.\n",
    "\n",
    "Check possible differences between parameters of the models:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==============================\n",
      "GeneralizedMF\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "NeuMF\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "MultiVAE\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "MultiDAE\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "ConvNeuMF\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "ConvMF\n",
      "No Differences Found Between Models\n",
      "==============================\n",
      "NGCF\n",
      "No Differences Found Between Models\n"
     ]
    }
   ],
   "source": [
    "# Disclaimer: This cell was partly generated with the support of a LLM (ChatGPT)\n",
    "\n",
    "\n",
    "for model in df_amazon_experiments[\"model_name\"].unique():\n",
    "    \n",
    "    print(\"=\" * 30)\n",
    "    print(model)\n",
    "\n",
    "    model1 = df_amazon_experiments.loc[df_amazon_experiments[\"model_name\"] == model, \"model\"].unique().item()\n",
    "    model2 = df_amazon_results.loc[df_amazon_results[\"model_name\"] == model, \"model\"].unique().item()\n",
    "\n",
    "    # Use difflib to compare\n",
    "    diff = list(difflib.ndiff(model1.split(\"_\"), model2.split(\"_\")))\n",
    "\n",
    "    # Extract and print differences\n",
    "    differences = []\n",
    "    for d in diff:\n",
    "        if d.startswith('- ') or d.startswith('+ '):\n",
    "            key = d[2:].split(\"=\")[0] if \"=\" in d else \"Unknown\"\n",
    "            value = d[2:].split(\"=\")[1] if \"=\" in d else d[2:]\n",
    "            model_source = \"Config File\" if d.startswith(\"- \") else \"Results File\"\n",
    "            differences.append((key, value, model_source))\n",
    "\n",
    "    # Print results\n",
    "    if differences:\n",
    "        print(\"Differences found:\\n\")\n",
    "        for param, value, model_source in differences:\n",
    "            print(f\"Parameter: {param}\")\n",
    "            print(f\"   - {model_source}: {value}\\n\")\n",
    "    else:\n",
    "        print(\"No Differences Found Between Models\")\n",
    "\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".venv",
   "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": "3.12.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}