文章基本信息

标题：Prediction Models of Diabetes Diseases Based on Heterogeneous Multiple Classifiers
本地全文：下载
作者：I Gede Agus Suwartane ; Mohammad Syafrullah ; Krisna Adiyarta 等
期刊名称：International Journal of Advances in Soft Computing and Its Applications
印刷版ISSN：2074-8523
出版年度：2018
卷号：10
期号：2
出版社：International Center for Scientific Research and Studies
摘要：Diabetes disease is one of the global and most important healthproblems of the 21st century with the number of patients growingevery year. One in two diabetics is undiagnosed patients,consequently many patients who already have severe complications.One way to reduce and slow the complications of diabetes is to makean early diagnosis. With the development of data mining, developedvarious models predicting diabetes by using data mining techniques.The main problem in building predictive models is how to improvethe accuracy of predicted results. In this research, HeterogeneousMultiple Classifiers diabetic prediction model is developed bycombining Support Vector Machine (SVM), K- Nearest Neighbor(KNN) and Decision Tree (C4.5) using Majority Voting. Theprediction model based on Heterogeneous Multiple Classifiers wasconstructed to produce 93.56% accuracy, 97.48% sensitivity, 89.22%specificity, 91.16% precision and 94.13% F-Measure. The resultingperformance value of Heterogeneous Multiple Classifiers basedprediction model is higher than the performance value of SingleClassifier-based prediction model used in building prediction modelbased on Heterogeneous Multiple Classifiers. Optimizationconducted on prediction model based on Heterogeneous MultipleClassifiers in this study also proved to improve the performance ofthe prediction model.
关键词：Diabetes; Heterogeneous Multiple Classifiers; SVM; KNN; C4.5