Practical Gradient Boosting: An deep dive into Gradient Boosting in Python

This book on Gradient Boosting methods is intended for students, academics, engineers, and data scientists who wish to master XGBoost, LightGBM or CatBoost. They will discover in depth the functioning of Gradient Boosting used to build decision tree ensembles. All the concepts are illustrated by examples of application code. They allow the reader to rebuild from scratch their own XGBoost like library. In parallel, the book presents the best practices of Data Science and provides the reader with a solid technical background to build Machine Learning models. After a presentation of the principles of Gradient Boosting citing the application cases, advantages and limitations, the reader is introduced to the details of the mathematical theory. A simple implementation is given to illustrate how it works. The reader is then armed to tackle the application and configuration of these methods. Data preparation, training, explanation of a model, management of Hyper Parameter Tuning and use of objective functions are covered in detail! The last chapters of the book extend the subject to the application of Gradient Boosting for time series, the presentation of the emblematic libraries XGBoost, CatBoost and LightGBM as well as the concept of multi-resolution models.