Shortest Reducts Versus Shortest Constructs

This paper investigates the comparative performance of shortest reducts and shortest constructs in supervised classification tasks using real-life datasets from various domains, including both original datasets and noise-distorted variations. The study evaluates the effectiveness of the shortest reducts and shortest constructs, particularly in noisy environments. Experimental results provide insights into the relative performance of these attribute subsets and their impact on classification accuracy. The findings contribute to the understanding of the use of shortest reducts versus shortest constructs in supervised classification problems.