Machine Learning at Scale with H2O by Gregory Keys | David Whiting

Author:Gregory Keys | David Whiting
Language: eng
Format: epub
Publisher: Packt Publishing Pvt Ltd
Published: 2022-06-29T00:00:00+00:00

Global explanations for multiple models

In determining which model to promote into production, for instance, from an AutoML run, the data scientist could rely purely on predictive model metrics. This could mean simply promoting the model with the best AUC value. However, there is a lot of information that could be used to help in this decision, with predictive power being only one of multiple criteria.

The global and local explain features of H2O provide additional information that is useful for evaluating models in conjunction with predictive attributes. We demonstrate it using the check AutoML object from Chapter 5, Advanced Model Building – Part 1.

The code to launch global explanations for multiple models is simply as follows:

check.explain(test)

This results in a variable importance heatmap, model correlation heatmap, and multiple-model partial dependence plots. We will review each of these in order.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.