Interpretability-by-Design with Accurate Locally Additive Models and Conditional Feature Effects

TL;DR

CALMs introduce a novel model class that balances interpretability and accuracy by allowing feature effects to vary across regions, capturing interactions while maintaining interpretability.

Contribution

The paper proposes CALMs, a new model class that enables region-specific univariate effects, combining the interpretability of GAMs with the accuracy of GA²Ms.

Findings

CALMs outperform GAMs in accuracy across diverse tasks.

CALMs achieve comparable accuracy to GA²Ms while maintaining interpretability.

The training pipeline effectively identifies homogeneous regions for modeling.

Abstract

Generalized additive models (GAMs) offer interpretability through independent univariate feature effects but underfit when interactions are present in data. GA$^2$Ms add selected pairwise interactions which improves accuracy, but sacrifices interpretability and limits model auditing. We propose \emph{Conditionally Additive Local Models} (CALMs), a new model class, that balances the interpretability of GAMs with the accuracy of GA$^2$Ms. CALMs allow multiple univariate shape functions per feature, each active in different regions of the input space. These regions are defined independently for each feature as simple logical conditions (thresholds) on the features it interacts with. As a result, effects remain locally additive while varying across subregions to capture interactions. We further propose a principled distillation-based training pipeline that identifies homogeneous regions with limited interactions and fits interpretable shape functions via region-aware backfitting. Experiments on diverse classification and regression tasks show that CALMs consistently outperform GAMs and achieve accuracy comparable with GA$^2$Ms. Overall, CALMs offer a compelling trade-off between predictive accuracy and interpretability.