Title: An analysis of parallel ensemble diabetes decision support system based on voting classifier for classification problem
Authors: S. Sathurthi; K. Saruladha
Addresses: Department of Computer Science and Engineering, Pondicherry Engineering College, Pondicherry-605014, India ' Department of Computer Science and Engineering, Pondicherry Engineering College, Pondicherry-605014, India
Abstract: Diabetes mellitus is one of the prominent health challenges in the world. Diabetes is a dangerous, metabolic disease that caused by human blood sugar level and progresses throughout life. In supervised learning-based systems have been proposed that incorporate ensemble learning techniques for diabetes prediction depends upon the diagnostic measurement of the diabetes patient. In this paper, voting classifier were used for combining the various ensemble and base classifiers for designing diabetes disease prediction. Voting mechanism helps to build the multiple ensemble and base classifier model. The accuracy of ensemble of ensemble classifiers has resulted in high rate of accuracy (79%) when compared to the ensemble of base classifiers (77%) with majority rule voting (MRV) and weighted majority voting (WMV) models. Hence, ensemble of ensemble classifier was chosen as the best model for diabetes healthcare prediction. This system has been experimented with Pima Indian diabetes UCI dataset and its implemented in python language.
Keywords: base classifiers; ensemble classifiers; cross validation; bagging; boosting; decision tree; majority rule voting; MRV; weighted majority voting; WMV.
Electronic Government, an International Journal, 2020 Vol.16 No.1/2, pp.25 - 38
Received: 31 Mar 2019
Accepted: 23 May 2019
Published online: 22 Feb 2020 *