Rejection threshold optimization using 3D ROC curves: Novel findings on biomedical datasets

dc.authorscopusid55664907400
dc.authorscopusid57222607252
dc.contributor.authorUyar,A.
dc.contributor.authorUyar, Aslı
dc.contributor.otherBilgisayar Mühendisliği / Computer Engineering
dc.date.accessioned2024-05-25T12:34:18Z
dc.date.available2024-05-25T12:34:18Z
dc.date.issued2021
dc.departmentOkan Universityen_US
dc.department-tempUyar A., Computer Eng., Okan University, İstanbul, 34959, Turkey; Sengul Y.A., Industrial Eng., Doğuş University, İstanbul, 34775, Turkeyen_US
dc.description.abstractReject option is introduced in classification tasks to prevent potential misclassifications. Although optimization of error-reject trade-off has been widely investigated, it is shown that error rate itself is not an appropriate performance measure, when misclassification costs are unequal or class distributions are imbalanced. ROC analysis is proposed as an alternative approach to performance evaluation in terms of true positives (TP) and false positives (FP). Considering classification with reject option, we need to represent the tradeoff between TP, FP and rejection rates. In this paper, we propose 3D ROC analysis to determine the optimal rejection threshold as an analogy to decision threshold optimization in 2D ROC curves. We have demonstrated our proposed method with Naive Bayes classifier on Heart Disease dataset and validated the efficiency of the method on multiple datasets from UCI Machine Learning Repository. Our experiments reveal that classification with optimized rejection threshold significantly improves true positive rates in biomedical datasets. Furthermore, false positive rates remain the same with rejection rates below 10% on average. © 2021, Ismail Saritas. All rights reserved.en_US
dc.identifier.citation0
dc.identifier.doi10.18201/ijisae.2021167933
dc.identifier.endpage27en_US
dc.identifier.issn2147-6799
dc.identifier.issue1en_US
dc.identifier.scopus2-s2.0-85103368836
dc.identifier.scopusqualityQ4
dc.identifier.startpage22en_US
dc.identifier.urihttps://doi.org/10.18201/ijisae.2021167933
dc.identifier.urihttps://hdl.handle.net/20.500.14517/2570
dc.identifier.volume9en_US
dc.institutionauthorUyar A.
dc.institutionauthorUyar, Aslı
dc.language.isoen
dc.publisherIsmail Saritasen_US
dc.relation.ispartofInternational Journal of Intelligent Systems and Applications in Engineeringen_US
dc.relation.publicationcategoryMakale - Uluslararası Hakemli Dergi - Kurum Öğretim Elemanıen_US
dc.rightsinfo:eu-repo/semantics/openAccessen_US
dc.subject3D ROC curvesen_US
dc.subjectDecision threshold optimizationen_US
dc.subjectNaive bayesen_US
dc.subjectRejection threshold optimizationen_US
dc.titleRejection threshold optimization using 3D ROC curves: Novel findings on biomedical datasetsen_US
dc.typeArticleen_US
dspace.entity.typePublication
relation.isAuthorOfPublication9cfff5a5-9c60-4015-8119-1add2ac5a936
relation.isAuthorOfPublication.latestForDiscovery9cfff5a5-9c60-4015-8119-1add2ac5a936
relation.isOrgUnitOfPublicationc8741b9b-4455-4984-a245-360ece4aa1d9
relation.isOrgUnitOfPublication.latestForDiscoveryc8741b9b-4455-4984-a245-360ece4aa1d9

Files