Performance analysis of unsupervised feature selection methods

TL;DR

This paper evaluates various unsupervised feature selection methods like PCA, Rough PCA, USQR, and EDR to identify the most relevant features for classification, aiming to improve accuracy and reduce computational costs.

Contribution

It provides a comparative analysis of multiple unsupervised feature selection techniques using standard classification metrics.

Findings

PCA and EDR effectively select discriminative features.

Unsupervised methods can improve classification accuracy.

Efficiency varies among different feature selection approaches.

Abstract

Feature selection (FS) is a process which attempts to select more informative features. In some cases, too many redundant or irrelevant features may overpower main features for classification. Feature selection can remedy this problem and therefore improve the prediction accuracy and reduce the computational overhead of classification algorithms. The main aim of feature selection is to determine a minimal feature subset from a problem domain while retaining a suitably high accuracy in representing the original features. In this paper, Principal Component Analysis (PCA), Rough PCA, Unsupervised Quick Reduct (USQR) algorithm and Empirical Distribution Ranking (EDR) approaches are applied to discover discriminative features that will be the most adequate ones for classification. Efficiency of the approaches is evaluated using standard classification metrics.