Feature Relevance Bounds for Ordinal Regression

TL;DR

This paper introduces a method to determine feature relevance bounds in ordinal regression, helping to distinguish between strongly and weakly relevant features, especially in high-dimensional or correlated data.

Contribution

It proposes a novel approach for identifying feature relevance bounds that differentiate between strongly and weakly relevant features in ordinal regression models.

Findings

Identifies relevant features in ordinal regression

Differentiates between strongly and weakly relevant features

Addresses issues with high-dimensional and correlated data

Abstract

The increasing occurrence of ordinal data, mainly sociodemographic, led to a renewed research interest in ordinal regression, i.e. the prediction of ordered classes. Besides model accuracy, the interpretation of these models itself is of high relevance, and existing approaches therefore enforce e.g. model sparsity. For high dimensional or highly correlated data, however, this might be misleading due to strong variable dependencies. In this contribution, we aim for an identification of feature relevance bounds which - besides identifying all relevant features - explicitly differentiates between strongly and weakly relevant features.