Binary Classification with XGBoost (Done Right)

For tabular binary classification, default to XGBoost. It dominates Kaggle and real-world benchmarks for tabular data, handles missing values and mixed feature types essentially for free, and gives you SHAP-based explanations as a side effect. This skill covers the four things that separate "ROC-AUC on a notebook" from "a model you can deploy and trust."

When to use this skill