Explainable robust DB-predictive models

E. Boj del Val, A. Grané

Distance-based (DB) predictive models are powerful tools which can be applied to any kind of data whenever a distance measure can be computed among units. However, this advantage can be their main drawback if one is interested in assessing the predictors’ influence on the response, since the relationship between predictors and response depends on the distance measure. In this work, we focus on the robust DB generalized linear model (DB-GLM) and explore the use of explainable artificial intelligence (XAI) methodologies of type LIME (local interpretable model-agnostic explanations) or SHAP (Shapley additive explanations) to help the visualization of predictors’ influence.

Palabras clave: DB prediction, robust distances, explainable AI, dbglm, R

AMC2 Métodos Robustos
10 de junio de 2025  17:10
Auditorio 1. Ricard Vinyes

S. Díaz Aranda, J. M. Ramírez, J. Aguilar, A. Fernández Anta, R. E. Lillo

L. A. García-Escudero, G. Greselin, A. Mayo-Iscar, G. Zaccaria

M. Durbán, V. Guerrero, C. Molero-Río, J. Secilla Martínez