Create 10_Benchmark_1.ipynb

10_Benchmark_1.ipynb

@@ -0,0 +1,501 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "id": "11777db4-14dd-4991-87e4-a8e6ec0c7e89",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "124033\n"
+     ]
+    }
+   ],
+   "source": [
+    "import pandas as pd\n",
+    "import numpy as np\n",
+    "import networkx as nx\n",
+    "from sklearn.metrics.pairwise import cosine_similarity\n",
+    "\n",
+    "def generate_graph_modality(file_path, threshold=0.2):\n",
+    "    # Read the uploaded file containing user-item ratings\n",
+    "    ratings_df = pd.read_csv(file_path)  # Assuming CSV format, adjust accordingly if different\n",
+    "\n",
+    "    # Compute user-item matrix\n",
+    "    user_item_matrix = pd.pivot_table(ratings_df, values='rating', index='user_id', columns='business_id', fill_value=0)\n",
+    "\n",
+    "    # Compute cosine similarity between users\n",
+    "    user_similarity_matrix = cosine_similarity(user_item_matrix)\n",
+    "\n",
+    "    # Convert similarity matrix to binary adjacency matrix\n",
+    "    binary_adjacency_matrix = np.where(user_similarity_matrix > threshold, 1, 0)\n",
+    "\n",
+    "    # Convert binary adjacency matrix to a list of tuples for graph modality\n",
+    "    graph_modality_list = []\n",
+    "    for i in range(len(user_item_matrix.index)):\n",
+    "        for j in range(i + 1, len(user_item_matrix.index)):\n",
+    "            if binary_adjacency_matrix[i][j] == 1:\n",
+    "                graph_modality_list.append((user_item_matrix.index[i], user_item_matrix.index[j], 1.0))\n",
+    "\n",
+    "    return graph_modality_list\n",
+    "\n",
+    "# Example usage:\n",
+    "file_path = \"../data/rating_final.csv\"  # Update with the actual file path\n",
+    "graph_modality_list = generate_graph_modality(file_path)\n",
+    "trust_graph_df = pd.DataFrame(graph_modality_list)\n",
+    "# print(\"Graph Modality List:\")\n",
+    "# print(graph_modality_list)\n",
+    "print(len(trust_graph_df))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "b877dbe6-7175-4de9-ba89-37d43661500e",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "rating_threshold = 1.0\n",
+      "exclude_unknowns = True\n",
+      "---\n",
+      "Training data:\n",
+      "Number of users = 10999\n",
+      "Number of items = 4922\n",
+      "Number of ratings = 176857\n",
+      "Max rating = 5.0\n",
+      "Min rating = 1.0\n",
+      "Global mean = 3.8\n",
+      "---\n",
+      "Test data:\n",
+      "Number of users = 10999\n",
+      "Number of items = 4922\n",
+      "Number of ratings = 58885\n",
+      "Number of unknown users = 0\n",
+      "Number of unknown items = 0\n",
+      "---\n",
+      "Validation data:\n",
+      "Number of users = 10999\n",
+      "Number of items = 4922\n",
+      "Number of ratings = 58902\n",
+      "---\n",
+      "Total users = 10999\n",
+      "Total items = 4922\n",
+      "Total number of users: 11000\n",
+      "Total number of restaurants: 4963\n",
+      "Total possible ratings: 54593000\n",
+      "Actual number of ratings: 294763\n",
+      "Sparsity of the data: 99.46007180407744\n",
+      "\n",
+      "[BPR] Training started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 100/100 [00:01<00:00, 51.29it/s, correct=84.93%, skipped=0.81%]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Optimization finished!\n",
+      "\n",
+      "[BPR] Evaluation started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10561/10561 [00:02<00:00, 4182.97it/s]\n",
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10534/10534 [00:02<00:00, 4405.11it/s]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "[WBPR] Training started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 100/100 [00:01<00:00, 50.09it/s, correct=50.72%, skipped=3.02%]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Optimization finished!\n",
+      "\n",
+      "[WBPR] Evaluation started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10561/10561 [00:02<00:00, 4141.18it/s]\n",
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10534/10534 [00:02<00:00, 4526.76it/s]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "[MF] Training started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 100/100 [00:00<00:00, 228.21it/s, loss=1346.61]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Optimization finished!\n",
+      "\n",
+      "[MF] Evaluation started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10561/10561 [00:02<00:00, 4127.18it/s]\n",
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10534/10534 [00:02<00:00, 4469.11it/s]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "[WMF] Training started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 155/155 [01:10<00:00,  2.20it/s, loss=301]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Learning completed!\n",
+      "\n",
+      "[WMF] Evaluation started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Ranking: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10561/10561 [00:12<00:00, 848.89it/s]\n",
+      "Ranking: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10534/10534 [00:12<00:00, 876.97it/s]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "[NeuMF] Training started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:46<00:00,  9.23s/it, loss=0.501]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "[NeuMF] Evaluation started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Ranking: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10561/10561 [00:19<00:00, 542.52it/s]\n",
+      "Ranking: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10534/10534 [00:18<00:00, 566.23it/s]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "[VAECF] Training started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:06<00:00,  1.22s/it, loss=1.17]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "[VAECF] Evaluation started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10561/10561 [00:03<00:00, 2717.43it/s]\n",
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10534/10534 [00:03<00:00, 2860.03it/s]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "[CVAECF] Training started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 50/50 [01:31<00:00,  1.84s/it, loss=0.93]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "[CVAECF] Evaluation started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10561/10561 [00:05<00:00, 1832.24it/s]\n",
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10534/10534 [00:05<00:00, 1897.65it/s]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "[BiVAECF] Training started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 50/50 [01:42<00:00,  2.06s/it, loss_i=0.441, loss_u=0.191]\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "[BiVAECF] Evaluation started!\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10561/10561 [00:02<00:00, 4216.96it/s]\n",
+      "Ranking: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10534/10534 [00:02<00:00, 4669.25it/s]"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "VALIDATION:\n",
+      "...\n",
+      "        | NCRR@10 | NDCG@10 | Recall@10 | Time (s)\n",
+      "------- + ------- + ------- + --------- + --------\n",
+      "BPR     |  0.0377 |  0.0413 |    0.0468 |   2.3963\n",
+      "WBPR    |  0.0297 |  0.0333 |    0.0399 |   2.3315\n",
+      "MF      |  0.0040 |  0.0043 |    0.0042 |   2.3616\n",
+      "WMF     |  0.0489 |  0.0541 |    0.0632 |  12.0190\n",
+      "NeuMF   |  0.0013 |  0.0014 |    0.0015 |  18.6082\n",
+      "VAECF   |  0.0347 |  0.0383 |    0.0445 |   3.6877\n",
+      "CVAECF  |  0.0545 |  0.0615 |    0.0739 |   5.5564\n",
+      "BiVAECF |  0.0002 |  0.0002 |    0.0002 |   2.2606\n",
+      "\n",
+      "TEST:\n",
+      "...\n",
+      "        | NCRR@10 | NDCG@10 | Recall@10 | Train (s) | Test (s)\n",
+      "------- + ------- + ------- + --------- + --------- + --------\n",
+      "BPR     |  0.0425 |  0.0456 |    0.0502 |    1.9605 |   2.5325\n",
+      "WBPR    |  0.0332 |  0.0365 |    0.0422 |    2.0041 |   2.5546\n",
+      "MF      |  0.0033 |  0.0035 |    0.0034 |    0.4536 |   2.5634\n",
+      "WMF     |  0.0533 |  0.0583 |    0.0669 |   70.6555 |  12.4469\n",
+      "NeuMF   |  0.0009 |  0.0011 |    0.0014 |   46.3940 |  19.4710\n",
+      "VAECF   |  0.0401 |  0.0427 |    0.0469 |    6.0933 |   3.8909\n",
+      "CVAECF  |  0.0601 |  0.0661 |    0.0770 |   91.9570 |   5.7691\n",
+      "BiVAECF |  0.0005 |  0.0005 |    0.0005 |  103.3335 |   2.5094\n",
+      "\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "import cornac\n",
+    "from cornac.eval_methods import RatioSplit\n",
+    "from cornac.models import BPR, MF, NeuMF, VAECF, CVAECF, BiVAECF, LightGCN, WBPR, WMF\n",
+    "from cornac.metrics import NCRR\n",
+    "from cornac.data import GraphModality\n",
+    "import pandas as pd\n",
+    "\n",
+    "# Assume data is a Cornac dataset object\n",
+    "# data = cornac.data.Dataset.from_uir(your_data)\n",
+    "\n",
+    "# Model parameters\n",
+    "LATENT_DIM = 50\n",
+    "ENCODER_DIMS = [20]\n",
+    "ACT_FUNC = \"tanh\"\n",
+    "LIKELIHOOD = \"gaus\"\n",
+    "NUM_EPOCHS = 5\n",
+    "BATCH_SIZE = 128\n",
+    "LEARNING_RATE = 0.01\n",
+    "\n",
+    "SEED=4567\n",
+    "VERBOSE=True\n",
+    "\n",
+    "df = pd.read_csv('../data/rating_final.csv')\n",
+    "data_list = df.values.tolist()\n",
+    "\n",
+    "eval_metrics = [\n",
+    "  cornac.metrics.Recall(k=10),\n",
+    "  cornac.metrics.NDCG(k=10),\n",
+    "  cornac.metrics.NCRR(k=10),\n",
+    "]\n",
+    "\n",
+    "user_graph_modality = GraphModality(data=graph_modality_list)\n",
+    "\n",
+    "# Split the data\n",
+    "ratio_split = RatioSplit(data=data_list, val_size=0.2, test_size=0.2, \n",
+    "                        user_graph=user_graph_modality,\n",
+    "                        exclude_unknowns=True, seed=SEED, verbose=True)\n",
+    "\n",
+    "# Define models\n",
+    "models = [\n",
+    "    BPR(k=50, learning_rate=0.01, lambda_reg=0.01, max_iter=100),\n",
+    "    WBPR(k=50, max_iter=100, learning_rate=0.001, lambda_reg=0.01, verbose=True),\n",
+    "    MF(k=50, learning_rate=0.01, lambda_reg=0.01, max_iter=100),\n",
+    "    WMF(k=50, max_iter=155, a=1.0, b=0.1, learning_rate=0.00555, lambda_u=0.0155, lambda_v=0.0155,\n",
+    "          verbose=VERBOSE, seed=SEED),\n",
+    "    NeuMF(num_factors=50, layers=(64, 64, 32, 16), act_fn='relu', reg=0.01, num_epochs=5, \n",
+    "              batch_size=128, num_neg=4, lr=0.01, learner='adam', trainable=True, verbose=True, seed=SEED),\n",
+    "    VAECF(k=50, autoencoder_structure=[20], act_fn='tanh', likelihood='pois', n_epochs=5, batch_size=128),\n",
+    "    # LightGCN(seed=SEED,emb_size=64,num_epochs=5,num_layers=3,early_stopping={\"min_delta\": 1e-4, \"patience\": 50},batch_size=128,\n",
+    "    #     learning_rate=0.01,lambda_reg=0.01,verbose=True),\n",
+    "    CVAECF(z_dim=50,h_dim=20,autoencoder_structure=[40],learning_rate=0.01,n_epochs = 50,batch_size = 128,seed = SEED),\n",
+    "    BiVAECF(k=LATENT_DIM,encoder_structure=ENCODER_DIMS,act_fn=ACT_FUNC,likelihood=LIKELIHOOD,n_epochs=50,batch_size=BATCH_SIZE,\n",
+    "        learning_rate=LEARNING_RATE,seed=SEED,trainable = True,use_gpu=True,verbose=True)\n",
+    "]\n",
+    "\n",
+    "# Count the total number of unique users and unique businesses\n",
+    "num_users = df['user_id'].nunique()\n",
+    "num_businesses = df['business_id'].nunique()\n",
+    "\n",
+    "# Calculate the total number of possible ratings\n",
+    "total_possible_ratings = num_users * num_businesses\n",
+    "\n",
+    "# Count the actual number of ratings in the dataset\n",
+    "num_ratings = len(df)\n",
+    "\n",
+    "# Calculate the sparsity of the data\n",
+    "sparsity = 1 - (num_ratings / total_possible_ratings)\n",
+    "\n",
+    "print(\"Total number of users:\", num_users)\n",
+    "print(\"Total number of restaurants:\", num_businesses)\n",
+    "print(\"Total possible ratings:\", total_possible_ratings)\n",
+    "print(\"Actual number of ratings:\", num_ratings)\n",
+    "print(\"Sparsity of the data:\", sparsity * 100)\n",
+    "\n",
+    "\n",
+    "# Evaluate models\n",
+    "cornac.Experiment(eval_method=ratio_split, models=models, metrics=eval_metrics, verbose=True).run()\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "e44e15b4-1127-4048-b429-895a1382ddfb",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "91efdbb7-6e6c-491a-8ca3-5812429807fe",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.0"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}