Dev Tools · 1h ago
Random Forest Age Imputation Boosts Kaggle Titanic Score by 0.002
A developer improved the Kaggle Titanic survival prediction model by replacing median age imputation with a RandomForestRegressor. The 5-fold cross-validation score rose from 0.8507 to 0.8519, and the public Kaggle score increased from 0.78708 to 0.78947. The code is available on GitHub.
Meridian48 take
The marginal gain highlights the diminishing returns of feature engineering on a well-worn dataset, but the technique is a solid tutorial for ML beginners.
Read the full reporting
Kaggle Titanic: Improving Survival Prediction with Random Forest Age Imputation →
DEV Community
machine-learningkaggle