Interaction Forests: Identifying and exploiting interpretable quantitative and qualitative interaction effects

Although interaction effects can be exploited to improve predictions and allow for valuable insights into covariate interplay, they are given limited attention in analysis. Interaction forests a variant of random categorical, continuous, survival outcomes that explicitly models quantitative qualitative bivariable splits performed by the trees constituting forests. The new effect importance measure (EIM) associated with allows ranking pairs respect their effects' prediction. Using EIM, separate value lists univariable effects, obtained. In spirit interpretable machine learning, split types target easily communicable effects. To learn about nature interplay between covariates identified as interacting it is convenient visualise estimated influence. Functions perform this task provided R package diversityForest, which implements large-scale empirical study using 220 data sets, tended deliver better than conventional competing forest variants use multivariable splitting. simulation study, EIM delivered considerably rankings relevant approaches. These results indicate suitable tools challenging identifying making predictive modelling.