期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2018
卷号：96
期号：9
出版社：Journal of Theoretical and Applied
摘要：Feature selection plays a significant role in any data mining research problem. In this research work, comprehensive feature selection is applied for selecting the attributes in the chosen PIMA Indian diabetes dataset. The comprehensive feature selection mechanism makes use of maximum significance pattern for selecting the most edifying features, which effectively distinguish between different classes of samples. By making use of the comprehensive feature selection mechanism, the degree of relevance between the feature variables and target variable is quantified, and the informativeness of features is determined according to the degree of the relevance score. In order to implement this mechanism, novel evidence metric, NEM, is employed to score the relevance degree of a candidate feature variable with respect to the target variable. Once after the feature selection is carried out, SVM classifier is applied for performing the prediction of heart disease among gestational diabetes patients in the dataset. Performance metrics such as sensitivity, specificity, true positive rate, false positive rate, precision, accuracy and time taken for feature selection are taken into account. The results are demonstrated with better performance.
关键词：Feature Selection; Data Mining; Gestational Diabetes; Accuracy; Time Taken; Feature Selection; Risk Prediction
