A NOVEL INTERPRETABLE WEB-BASED TOOL ON THE ASSOCIATIVE CLASSIFICATION METHODS: AN APPLICATION ON BREAST CANCER DATASET

Authors : Ahmet Kadir ARSLAN, Zeynep TUNÇ, İpek BALIKÇI ÇİÇEK, Cemil ÇOLAK

Pages : 33-40

View : 15 | Download : 6

Publication Date : 2020-06-30

Article Type : Research Paper

Abstract :Aim: The second-largest cause of cancer mortality for women is breast cancer. The main techniques for diagnosing breast cancer are mammography and tumor biopsy accompanied by histopathological studies. The mammograms are not detective of all subtypes of breast tumors, particularly those which arise and are more aggressive in young women or women with dense breast tissue. Circulating prognostic molecules and liquid biopsy approaches to detect breast cancer and the death risk are desperately essential. The purpose of this study is to develop a web-based tool for the use of the associative classification method that can classify breast cancer using the association rules method. Materials and Methods: In this study, an open-access dataset named “Breast Cancer Wisconsin insert ignore into journalissuearticles values(Diagnostic); Data Set” was used for the classification. To create this web-based application, the Shiny library is used, which allows the design of interactive web-based applications based on the R programming language. Classification based on association rules insert ignore into journalissuearticles values(CBAR); and regularized class association rules insert ignore into journalissuearticles values(RCAR); are utilized to classify breast cancer insert ignore into journalissuearticles values(malignant/benign); based on the generated rules. Results: Based on the classification results of breast cancer, accuracy, balanced accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and F1-score values obtained from the CBAR model are 0.954, 0.951, 0.939, 0.964, 0.939, 0.964, and 0.939 respectively. Conclusion: In the analysis of the open-access dataset, the proposed model has a distinctive feature in classifying breast cancer based on the performance metrics. The associative classification software developed based on CBAR produces successful predictions in the classification of breast cancer. The hypothesis established within the scope of the purpose of this study has been confirmed as the similar estimates are achieved with the results of other papers in the classification of breast cancer.
Keywords : Artificial intelligence, association rules, associative classification, web based software, breast cancer