Breast cancer prediction using ensemble voting classifiers in next-generation sequences

Abstract

Breast cancer has become the greatest frequent cancer among worldwide. Machine learning techniques contribute much tocancer prognosis. The prime focus of thework is to enhance the prognosis of breast cancer at an earlier stage using an ensemble of machine learning classi ers. Next generation genetic sequences of homo sapiens,BRCA1and BRCA2from National Centre for Biotechnology Information were derived for prediction of breast cancer. The proposed ensembled classi ers by hard voting and soft voting,combinedmodelslike Decision Tree technique, SVMalgorithm, LR statistical model, Linear Discriminant analysis model, Naive Bayes classi er and k-nearest neighbours' algorithm.Five ensembled models from 6 machine learning classi ers were concatenated for the prediction purpose. Classi cation accuracy of ensemble hard voting and soft voting classi ers were evaluated statistically.Soft voting classi er for model 1(DT & SVM) and model2(DT, SVM&LR) achieved greatest value for classi cation performance metrics. Among all ensembled models, model 1 as well as model 2 achieved maximum classi cation precision of 94%.