user@devops:~$ cat README.md
Recommendation Systems
# Description
Complete recommendation systems project with three classic approaches: Collaborative Filtering (User-User and Item-Item based on cosine similarity), SVD Matrix Factorization (latent factor decomposition with scikit-surprise, 5-fold cross-validation), and Content-Based Filtering (category-based item recommendation). Realistic synthetic dataset with 300 users, 60 items, and 5,000 ratings (72% sparse). Includes user-item matrix visualization, latent factors, content similarity, and per-item error analysis.
# Key features
$ Collaborative Filtering User-User and Item-Item with cosine similarity
$ SVD Matrix Factorization with 5-fold cross-validation
$ Content-Based Filtering based on item categories
$ Top-5 personalized recommendations per user
$ RMSE and MAE evaluation metrics for each approach
$ Latent factor and content similarity visualization
$ 3-way comparison across 5,000 synthetic ratings
# Gallery
# Technologies used
