A implementation of a Hybrid Matrix Factorization engine designed to solve the Cold-Start Problem using the MovieLens 100k dataset.

Model: Hybrid Matrix Factorization via LightFM

Cold-Start Strategy: Integrated item-feature metadata (genres) into latent space embeddings

Loss Function: Optimized using WARP (Weighted Approximate-Rank Pairwise) loss

Data Engineering: Leveraged COO Sparse Matrices for memory efficiency

Primary Metric: Precision@10 - 0.1253

Validation: 80/20 random train-test split

Create venv (Python 3.10 recommended).

Install requirements: pip install -r requirements.txt.

Run pipeline: python main.py.

Note: Utilizes a source-patched version of LightFM for compatibility with modern build backends.