Fair Multivariate Adaptive Regression Splines for Ensuring Equity and Transparency
Parian Haghighat, Denisa G'andara, Lulu Kang, Hadis Anahideh
TL;DR
This paper tackles inequity and opacity in predictive analytics by introducing fairMARS, a fairness-aware extension of multivariate adaptive regression splines (MARS). It integrates fairness directly into the learning process through two mechanisms: fairness-aware knot optimization in the forward/backward MARS procedure and fairness-aware, weighted coefficient estimation via a weighted least-squares objective. By measuring subgroup disparities in prediction errors and incorporating a tunable trade-off parameter $\lambda$, fairMARS balances accuracy with equity while preserving MARS’s interpretability and nonparametric flexibility. Empirical results on education- and crime-related datasets show improved fairness with competitive accuracy and significant computational efficiency relative to existing fair regression baselines, highlighting the method’s practical value for responsible analytics in social settings.
Abstract
Predictive analytics is widely used in various domains, including education, to inform decision-making and improve outcomes. However, many predictive models are proprietary and inaccessible for evaluation or modification by researchers and practitioners, limiting their accountability and ethical design. Moreover, predictive models are often opaque and incomprehensible to the officials who use them, reducing their trust and utility. Furthermore, predictive models may introduce or exacerbate bias and inequity, as they have done in many sectors of society. Therefore, there is a need for transparent, interpretable, and fair predictive models that can be easily adopted and adapted by different stakeholders. In this paper, we propose a fair predictive model based on multivariate adaptive regression splines(MARS) that incorporates fairness measures in the learning process. MARS is a non-parametric regression model that performs feature selection, handles non-linear relationships, generates interpretable decision rules, and derives optimal splitting criteria on the variables. Specifically, we integrate fairness into the knot optimization algorithm and provide theoretical and empirical evidence of how it results in a fair knot placement. We apply our fairMARS model to real-world data and demonstrate its effectiveness in terms of accuracy and equity. Our paper contributes to the advancement of responsible and ethical predictive analytics for social good.