Book Image

Hands-On Gradient Boosting with XGBoost and scikit-learn

By : Corey Wade
Book Image

Hands-On Gradient Boosting with XGBoost and scikit-learn

By: Corey Wade

Overview of this book

XGBoost is an industry-proven, open-source software library that provides a gradient boosting framework for scaling billions of data points quickly and efficiently. The book introduces machine learning and XGBoost in scikit-learn before building up to the theory behind gradient boosting. You’ll cover decision trees and analyze bagging in the machine learning context, learning hyperparameters that extend to XGBoost along the way. You’ll build gradient boosting models from scratch and extend gradient boosting to big data while recognizing speed limitations using timers. Details in XGBoost are explored with a focus on speed enhancements and deriving parameters mathematically. With the help of detailed case studies, you’ll practice building and fine-tuning XGBoost classifiers and regressors using scikit-learn and the original Python API. You'll leverage XGBoost hyperparameters to improve scores, correct missing values, scale imbalanced datasets, and fine-tune alternative base learners. Finally, you'll apply advanced XGBoost techniques like building non-correlated ensembles, stacking models, and preparing models for industry deployment using sparse matrices, customized transformers, and pipelines. By the end of the book, you’ll be able to build high-performing machine learning models using XGBoost with minimal errors and maximum speed.
Table of Contents (15 chapters)
1
Section 1: Bagging and Boosting
6
Section 2: XGBoost
10
Section 3: Advanced XGBoost

Modifying gradient boosting hyperparameters

In this section, we will focus on the learning_rate, the most important gradient boosting hyperparameter, with the possible exception of n_estimators, the number of iterations or trees in the model. We will also survey some tree hyperparameters, and subsample, which results in stochastic gradient boosting. In addition, we will use RandomizedSearchCV and compare results with XGBoost.

learning_rate

In the last section, changing the learning_rate value of GradientBoostingRegressor from 1.0 to scikit-learn's default, which is 0.1, resulted in enormous gains.

learning_rate, also known as the shrinkage, shrinks the contribution of individual trees so that no tree has too much influence when building the model. If an entire ensemble is built from the errors of one base learner, without careful adjustment of hyperparameters, early trees in the model can have too much influence on subsequent development. learning_rate limits the influence...