Code
9 articles
XGBoost for Classification · Binary, Multi-Class, and Ranking
How XGBoost handles classification problems — the objective functions, probability calibration, class imbalance strategies, and multi-class extensions.
XGBoost Distributed Training · Multi-Node and Dask Integration
How to train XGBoost across multiple machines — the distributed framework, Dask integration, Spark integration, and practical deployment patterns.
XGBoost for Regression · Objective Functions and Metrics
XGBoost regression guide — squared error, absolute error, pseudo-Huber, Tweedie, Gamma, and quantile regression objectives and when to use each.
XGBoost Cross-Validation · cv() vs Manual K-Fold
How XGBoost's built-in cross-validation works — the cv() function, stratified folds, and how to use it for both hyperparameter tuning and final model evaluation.
XGBoost Custom Objective Functions · Beyond Built-in Losses
How to implement custom objective functions and evaluation metrics in XGBoost — gradient and hessian derivation, worked examples for quantile loss and Huber loss.
XGBoost Early Stopping · Prevent Overfitting Automatically
How XGBoost's early stopping mechanism works — setting the right eval_metric, choosing evaluation sets, and practical tips for robust early stopping without overfitting.
XGBoost GPU Acceleration · Training Speed and Memory
How to enable GPU training in XGBoost — CUDA vs CPU benchmarks, the tree_method parameter, multi-GPU setup, and when GPU acceleration is worth the cost.
XGBoost Hyperparameter Tuning · A Systematic Guide
How to tune XGBoost hyperparameters effectively — the ordering of parameters, the most impactful knobs, and practical search strategies from grid search to Bayesian optimization.
XGBoost Hyperparameters · A Practical Tuning Guide
A field guide to XGBoost's most important hyperparameters — what each one does, how they interact, and a recommended tuning sequence.