A random subspace method that uses different instead of similar models for regression and classification problems. (1st January 2011)

• Record Type: Journal Article
• Authors: Kotsiantis, S.B.
• Is Part Of: International journal of information and decision sciences. Volume 3:Number 2(2011)
• Page Start: 173
• Page End: 188
• Publication Date: 2011-01-01
• Subjects: classifier -- machine learning -- data mining -- regressor; Decision making -- Periodicals; Decision support systems -- Periodicals; Management science -- Periodicals; 658.40305
• Languages: English
• ISSNs: 1756-7017

Abstract

Even though many ensemble techniques have been proposed, there is no clear picture of which method is best. In this study, we propose a technique that uses different subsets of the same feature set with the concurrent usage of a voting (for classification problems) or averaging methodology (for regression problems) for combining different learners instead of similar learners. We performed a comparison of the proposed ensemble with other well-known ensembles that use the same base learners and the proposed technique had better accuracy in most cases.