Experiments
Movie recommendation experiments
An exploration of three movie recommendation approaches: weighted ratings, TF-IDF description similarity and SVD collaborative filtering. Not executed for this review, so no accuracy figures are published.
Patrich's role: Author of all three scripts.
Problem and context
Exploring different ways to recommend movies from ratings and description data.
Solution
Three independent scripts (main.py, main2.py, main3.py) implement: weighted ratings, TF-IDF description similarity, and SVD collaborative filtering (with RMSE/MAE cross-validation) using pandas, scikit-learn and Surprise.
How I built it
No evidence of AI assistance was observed in this code.
Technologies
Pythonpandasscikit-learn
Lessons learned
main.py has a formula error (expresses v/v+m instead of v/(v+m)) that skews the weighted rating. main2.py references a file (movies_small.csv) missing from the reviewed tree. The scripts were not executed for this review, so no accuracy figures are published and running them without adjustments isn't guaranteed.