Machine LearningOnline
Housing Price Prediction
Kaggle 4-model weighted ensemble
A straight modeling exercise done properly: feature engineering first, then an ensemble whose members fail differently — gradient boosting alongside regularized linear models — so their errors partially cancel.
What it does
- 4-model weighted ensemble: XGBoost, LightGBM, ElasticNet, Ridge
- Log-transformed targets to handle price skew
- 65 engineered features
- Stacking architectures explored