Dish Recommendation Engine

A machine learning-powered dish recommendation system that combines collaborative filtering (Truncated SVD) and content-based filtering (TF-IDF) to deliver personalized dish suggestions for a food service platform.

Features

• Hybrid Recommendations: Combines user interaction data (ratings, orders) with dish features (categories, ingredients, etc.).
• Dynamic Weighting: Adjusts model influence based on user activity levels.
• Business Rules: Incorporates dietary restrictions, promotions, and inventory constraints.
• Scalability: Retrains models every 10 minutes using APScheduler for fresh recommendations.

├── categories.csv # Category data ├── dishes.csv # Dish data ├── dish_categories.csv # Dish-to-category mappings ├── dish_feature_mapping.csv # Dish feature mappings ├── user_preferences.csv # User preference data ├── user_ratings.csv # User rating data ├── run.py # Application entry point ├── app/ │ ├── models.py # Database models and engine │ ├── routes.py # API endpoints │ ├── services.py # Recommendation logic │ ├── utils.py # Data extraction and preprocessing │ ├── init.py # Python package initializer ├── config/ │ ├── config.py # Configuration settings ├── csv/ │ ├── dishcategory.py # Dish category processing │ ├── dishes.py # Dish data processing │ ├── dishfeaturemapping.py# Dish feature processing │ ├── ingredients.py # Ingredient data processing │ ├── userpreference.py # User preference processing │ ├── userrating.py # User rating processing

Setup

1. Clone the repository:

   git clone https://github.com/BaraaHazzaa/dish-recommendation-engine.git

2. Install dependencies: pip install -r requirements.txt

3. Configure the database in config/config.py (e.g., set DATABASE_URL).

4. Run the application: python run.py

Technologies:

Python, Pandas, NumPy, Scikit-learn, SQLAlchemy Flask (API routes) APScheduler (periodic model retraining) SQLite/PostgreSQL (database)

Usage:

Access recommendations via API endpoints in routes.py. Update preferences or ratings to trigger model retraining.