Final Year Project · Applied Machine Learning
Personalized Movie Recommendation System
University of London · BSc Computer Science, Machine Learning and AI · CM3070
A solo final-year project comparing five recommendation approaches, from a classic SVD baseline to a PyTorch autoencoder and a from-scratch Restricted Boltzmann Machine, then combining the two strongest signals into a weighted hybrid, evaluated with both offline metrics and a 10-person user test.
01
At a glance
02
03
Engineering & research trace
Follow the research from data preparation to user testing. Expand a stage to inspect the method.
04
Preparation
05
Model Explorer
Compare each model’s approach, test metrics and observed tradeoffs.
06
07
Methodology caveats
08
What worked
What users challenged
What changed
09
Limitations
10
What I would explore next
Report-supported future work, distinct from what was actually built for this project.
11