Visualizing Variable Importance and Variable Interaction Effects in Machine Learning Models

Variable importance, interaction measures, and partial dependence plots are important summaries in the interpretation of statistical machine learning models. In this article, we describe new visualization techniques for exploring these model summaries. We construct heatmap graph-based displays showing variable importance jointly, which carefully designed to highlight aspects fit. a matrix-type layout all single bivariate plots, an alternative based on graph Eulerians focusing key subsets. Our visualizations model-agnostic applicable regression classification supervised settings. They enhance even situations where number variables is large. R package vivid (variable displays) provides implementation. Supplementary files article available online.