Applied Predictive Modeling by Thomas Kuhn

A practical, example-driven guide to building and evaluating predictive models with a focus on real-world workflows, covering data preprocessing (imputation, encoding, transformations, PCA), exploratory data analysis, feature selection, resampling and model assessment (cross-validation, ROC and confusion matrices), and model tuning; it compares common algorithms (linear and penalized regression, trees, random forests, boosting, SVMs), addresses class imbalance and overfitting, and emphasizes variable importance, interpretability, reproducible code (notably using the caret framework), and best practices for producing robust predictive models.