Associative Classification in Multi-label Classification: an Investigative Study
Raed Alazaidah, Mohammed Amin Almaiah, Mo’ath Alluwaici
Abstract
Multi-label Classification (MLC) is a very interesting and important domain that has attracted many researchers in the last two decades. Several single label classification algorithms that belong to different learning strategies have been adapted to handle the problem of MLC. Surprisingly, no Associative Classification (AC) algorithm has been adapted to handle MLC problem, where AC algorithms have shown a high predictive performance comparing with other learning strategies in single label classification. In this paper, a deep investigation regarding utilizing AC in MLC is presented. An evaluation of several AC algorithms on three multi-label datasets with respect to five discretization techniques reveals that utilizing AC algorithms in MLC is very promising, comparing with other algorithms from different learning strategies. [JJCIT 2021; 7(2.000): 166-179]