Building Ensemble Models with Python

• Published in: Machine Learning in Python pp. 255 - 317
• Main Author: Bowles, Michael
• Format: Book Chapter
• Language: English
• Published: United States John Wiley & Sons, Incorporated 2015; John Wiley & Sons, Inc
• Subjects: bagging; binary classification; boosting; multiclass classification; penalized regression methods; Python ensemble packages; random forest; random forest model
• ISBN: 1118961749; 9781118961742
• DOI: 10.1002/9781119183600.ch7

This chapter uses several available Python packages to build predictive models using the ensemble algorithms. The examples show these methods at work building models on a variety of different types of problems. The chapter also covers regression, binary classification, and multiclass classification problems, and discusses variations on these themes such as the workings of coding categorical variables for input to Python ensemble methods, such as bagging, boosting and random forest. Ensemble methods are easy to use as they do not have many parameters to tune. The chapter then demonstrates the use of available Python packages. Seeing them exercised in the example code can help you get started using these packages. The comparisons given at the end of the chapter demonstrate how these algorithms compare. The ensemble methods frequently give the best performance. The penalized regression methods are blindingly much faster than ensemble methods and in some cases yield similar performance.