An efficient classification framework for breast cancer using hyper parameter tuned Random Decision Forest Classifier and Bayesian Optimization

Abstract

Decision tree algorithm is one of the algorithm which is easily understandable and interpretable algorithm used in both training and application purpose during breast cancer prognosis. To address this problem, Random Decision Forests are proposed. In this manuscript, the breast cancer classification can be determined by combining the advantages of Feature Weight and Hyper Parameter Tuned Random Decision Forest classifier. Here the Kernel Neutrosophic C-Means Clustering is used as Feature Weight, which allocates greater weights to applicable features and smaller weights to less applicable features. Then Random Decision Forest classifier model are optimized with the help of the Bayesian Optimization algorithm to obtain optimal hyper tuning parameters. By this, the accurate classification of breast cancer is successfully achieved. Then the efficiency of the proposed system is executed in python. The performance analysis are executed in Wisconsin prognostic Breast Cancer (WPBC) dataset, 70 % training and remaining 30 % testing is compared with the Wisconsin Diagnostic Breast Cancer (WDBC) dataset, the accuracy analysis of proposed feature weight and Random Decision Forest Classifier with Bayesian Optimization (FW + BOA-RDF) in Breast Cancer Wisconsin (Prognosis) Data Set is 6.66 %, 12.659 % and 37.618 % higher than existing method like FW + ALO-BPNN, FW + SSA-SVM, FW + GA-SVM respectively. The performance analysis in Wisconsin prognostic Breast Cancer (WPBC) dataset, 75 % training and the remaining 25 % testing is compared at Wisconsin Diagnostic Breast Cancer (WDBC) dataset, the accuracy analysis of FW + BOA-RDF in Breast Cancer Wisconsin (Prognosis) Data Set is 3.7146 %, 5.27398 % and 4.4413 % higher than existing method like FW + ALO-BPNN, FW + SSA-SVM, FW + GA-SVM respectively.