Model Evaluation and Feature Selection
Objectives
By the end, you'll be able to:
- Design reliable train/validation/test workflows.
- Choose metrics that match regression, classification, and ranking tasks.
- Use cross-validation, feature selection, and tuning without leaking information.
Concepts Covered
- Bias-variance trade-off and learning curves
- Regression, classification, and ranking metrics
- k-fold, stratified, repeated, and time-series cross-validation
- Filter, wrapper, and embedded feature selection
- Hyperparameter tuning, early stopping, and pruning
- Imbalanced data, resampling, and threshold selection