PhuHuynh07/OOP-Project---Demo- ? reverse-engineered prompt

Build me a Python movie recommendation app from scratch using clean object oriented design. I want it to use the MovieLens 100K data, let me train a few simple recommenders like popularity, collaborative filtering, and matrix factorization, then show top movie recommendations for a user in a simple web app.

Please keep the code modular and easy to test, with separate layers for data loading, models, evaluation, and a small service layer that ties everything together. It should handle basic edge cases nicely, like unknown users, empty data, or trying to recommend before training. I’d also like a few ranking metrics so I can compare models, and a demo page where I can pick a user and see recommended movies with scores and some helpful details.

If you need to, look up any current docs online. Also include a simple way to run the app locally and a basic test suite so I can verify the main pieces work.