Interactive Slice Visualization for Exploring Machine Learning Models

Abstract

Machine learning models fit complex algorithms to arbitrarily large datasets. These algorithms are well known to be high on performance and low on interpretability. We use interactive visualization of slices of predictor space to address the interpretability deficit; in effect opening up the black-box of machine learning algorithms, for the purpose of interrogating, explaining, validating and comparing model fits. Slices are specified directly through interaction, or using various touring algorithms designed to visit high-occupancy sections, or regions where the model fits have interesting properties. The methods presented here are implemented in the R package condvis2. Supplementary files for this article are available online.