Skip to content

README.md

Latest commit

 

History

History
71 lines (55 loc) · 1.91 KB

README.md

File metadata and controls

71 lines (55 loc) · 1.91 KB

Furnisique AI Recommender System

Welcome to the AI component of Furnisique — a smart furniture recommendation system designed to enhance the user's shopping experience by offering tailored furniture suggestions based on similarity.

🧠 Project Overview

This system analyzes a user's last purchase(s) and computes similarity-based furniture recommendations using a hybrid logic of encoded features and cosine similarity. It's deployed as a RESTful Flask API that integrates smoothly with a Laravel backend.

🔧 Tech Stack

Layer Technology
Data Processing Python, Pandas, NumPy
Similarity Logic Scipy (Cosine Similarity)
API Layer Flask
Backend Integration Laravel (calls the Flask API)

🔍 How It Works

  1. Vector Encoding: Each product is converted into a vector based on room type, aesthetic, category, normalized price, and color encoding.
  2. Similarity Computation: Uses cosine similarity to compare the vector of a new item with a user's last purchases.
  3. Color Compatibility Boost: Adds a minor boost to scores when color aesthetics match.
  4. Top-N Ranking: Returns top-N product recommendations (excluding purchased items).

🚀 Running the API

1. Install Requirements

pip install -r requirements.txt

2. Start the Flask Server

python recommender_api.py

3. Example API Request

GET http://localhost:5000/recommendations?user_id=12&top_n=10

✅ Sample Response

{
  "user_id": 12,
  "recommendations": [
    {"id": 5, "score": 0.97},
    {"id": 5, "score": 0.95},
    {"id": 5, "score": 0.92},
    {"id": 14, "score": 0.89},
    {"id": 3, "score": 0.87}
  ]
}

🧪 Testing

Run the local tests to validate the model:

python tests/test_recommender.py

👥 Contributors

  • AI Engineer: Nour Maged