Recommendation engine software for HustleSasa's e-commerce platform.

• Fast and Asynchronous: Built with Python's asyncio for high performance.
• Built-in Validation: Request and response validation out of the box using Pydantic.
• Scalable: Designed for both small and large-scale applications.

• Python: Make sure you have Python 3.7 or higher installed.
• Pip: Ensure pip is installed for dependency management.
• Docker: Ensure docker and docker compose is installed for container management.
• Optional: virtualenv for isolated environments.

1. Clone the repository:

   git clone https://github.com/dakbill/hustlesasa-recommendations.git
   cd /path/to/hustlesasa-recommendations
   

2. Create a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   

3. Install dependencies:

   pip install -r requirements.txt
   

4. Run the application:

   docker compose up -d --force-recreate
   fastapi dev main.py
   

.
├── app
├── main.py                 # Entry point of the application
├── tests                   # Unit tests
├── models                  # Pydantic models
├── repositories            # Neo4j graph interface
├── routers                 # rest routes
├── internal.py             # startup and shutdown logic, core_api_mocks (follow, purchase, rating)
├── requirements.txt        # Project dependencies
├── docs                    # Project documentation assets
├── README.md               # Project documentation
└── .gitignore              # Files and directories to ignore in Git

Once the application is running, navigate to the following URLs:

pytest

1. Product Recommendations

This type of recommendations is based on products that frequently included in the same cart as the product of reference, sorted by the weight determined by the frequency of purchase and average rating.

2. Social Recommendations

This type of recommendations based on products that purchased by other buyers a user of reference follows, sorted by the weight determined by the frequency of purchase and average rating.