Machine Learning for Business Analytics, 2nd Edition

13 GENERATING, COMPARING, AND COMBINING MULTIPLE MODELS

The previous chapters in this part of the book introduced different supervised methods for prediction and classification. Earlier, in Chapter 5, we learned about evaluating predictive performance and introduced metrics for comparing competing models and selecting the best one. This is facilitated by the Model Comparison platform in JMP Pro.

In this chapter, we look at collections of supervised models. First, we look at an approach for handling multiple models called ensembles, which combines multiple supervised models into a “super‐model.” Instead of choosing a single predictive model, we can combine several models to achieve improved predictive accuracy. We explain the underlying logic of why ensembles can improve predictive accuracy and introduce popular approaches for combining models, including simple averaging, bagging, boosting, and stacking.

Secondly, we introduce the idea of automated machine learning, or AutoML, which allows us to automatically train many supervised models and see their resulting performance. While in previous chapters we have shown how to manually tune parameters for each supervised method, AutoML automates this process by automatically fitting multiple models and generating a rank‐ordered list of candidate models. We illustrate this process and its results using the Model Screening platform in JMP Pro.

Ensembles and AutoML in JMP: The methods introduced in this chapter are only available ...