model lane
Reel Ratings, Predicting IMDb Scores
A team project predicting IMDb film quality from crew metadata, using tree-based models with a Bayesian shrinkage-adjusted rating.
pipeline
- data (covered)
- model (covered)
- eval (covered)
- serve (not covered)
- monitor (not covered)
Pythonscikit-learnXGBoostpandas
linksThe raw IMDb average is noisy for films that only have a few votes, so we first shrink each film’s rating toward a prior. Then we use tree-based models to predict that adjusted rating from the film’s crew metadata.