Choquistic Regression: Generalizing Logistic Regression using the Choquet Integral

Abstract

In this paper, we propose a generalization of logistic regression based on the Choquet integral. The basic idea of our approach, referred to as choquistic regression, is to replace the linear function of predictor variables, which is commonly used in logistic regression to model the log odds of the positive class, by the Choquet integral. Thus, it becomes possible to capture non-linear dependencies and interactions among predictor variables while preserving two important properties of logistic regression, namely the comprehensibility of the model and the possibility to ensure its monotonicity in individual predictors. In experimental studies with real and benchmark data, choquistic regression consistently improves upon standard logistic regression in terms of predictive accuracy.