Hybrid Product Recommendation System

A hybrid product recommendation system that analyzes customer behavior and shopping habits to deliver personalized product suggestions for e-commerce. The system combines content-based, collaborative filtering, and machine learning approaches to maximize personalization and recommendation accuracy.

---

📝 Description & System Overview

This project leverages the Consumer Behavior and Shopping Habits Dataset (~3900 rows, 18 columns), containing key customer demographics, product interactions, ratings, and purchase histories.

Key Features:

• Unified Streamlit web interface: view product list (list/grid), filter, and receive personalized product recommendations.
• Hybrid recommendation engine: content-based, collaborative filtering, ensemble ML (LightGBM, CatBoost, XGBoost), and hybrid weighted scoring.
• Advanced feature engineering and preprocessing for categorical and numerical attributes.
• Model evaluation using ranking metrics (Precision@N, Recall@N, NDCG, Coverage, RMSE, etc.).
• Modular, maintainable code structure for easy extension.

Video Demo:
👉 Watch the demo video here

Website Demo:
👉 website demo here

---

📁 Project Directory Structure

Product-Recommendation-System/
│
├── app.py                                              
├── requirements.txt                                    
├── README.md                                           
│
├── data/                                               
│   ├── shopping_behavior_updated.csv
│   ├── shopping_behavior_processed.csv
│   ├── shopping_behavior_final_features.csv
│   └── user_item_matrix.csv
│
├── src/                                                
│   ├── __init__.py
│   ├── eda.py                                          
│   ├── preprocessing.py                                
│   ├── feature_engineering.py                          
│   ├── training.py                                     
│   ├── recommendation_models.py                        
│   └── logger.py                                       
│
├── Models/                                             
│   ├── best_model.pkl
│   ├── user_profiles.pkl
│   ├── item_features.pkl
│   └── user_item_matrix.pkl
│   └── train_df.pkl
│
├── logs/                                               
│   └── *.log
│
├── Figures/                                            
│   ├── EDA/
│   ├── Preprocessing/
│   ├── Feature_Engineering/
│   └── Training/
│
└── notebooks/                                          
    └── *.ipynb

---

🚀 Local Setup & Running Instructions

1. Clone the repository

git clone https://github.com/KhoiBui16/Product-Recommendation-System.git
cd Product-Recommendation-System

2. (Optional) Set up a Python virtual environment

Windows:

python -m venv venv
venv\Scripts\activate

Linux/MacOS:

python3 -m venv venv
source venv/bin/activate

3. Install dependencies

pip install -r requirements.txt

4. Prepare the dataset

• Make sure the Data/ folder contains at least shopping_behavior_updated.csv.
• Download from Kaggle if needed.

5. Run the application

streamlit run app.py

• The app will be available at: http://localhost:8501

---

🤝 Contact & Contributions

• Main Author: [KhoiBui16] (khoib1601@gmail.com)
• Contributions, Issues, Ideas:
  Please open an issue or pull request on this GitHub repository, or contact via email above.