Fuzzy rough assisted missing value imputation and feature selection

Abstract

Presence of missing values and irrelevant features are commonplace issues that need to be handled effectively. Missing value imputation and feature selection is an efficient technique for redressing such problems. Fuzzy rough set-based approaches provide a handful of solutions for further dealing with vagueness and uncertainty available in the data. The present paper introduces the notion of imputing missing values followed by feature selection utilizing fuzzy rough set-based approaches. The idea of missing value estimation and instance ignorance are combined for fuzzy rough missing value imputation employing only correlated features followed by feature selection with a search heuristic. The experimental evaluation on benchmark datasets demonstrates the applicability and robustness of the proposed work. It significantly reduces data dimensionality after imputing missing values maintaining high performances. A comparative analysis demonstrates the superiority of the proposed methodology. © 2022, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.