Performance Analysis of Accuracy of Machine Learning Classifier in the Classification of Breast Cancer Disease

Abstract

The Breast cancer is recognized as a highly prevalent and life-threatening condition affecting women. In various domains, ML techniques have shown to be reliable predictors. In this study, medical data from Wisconsin breast cancer dataset were used to examine and compare supervised machine learning algorithms for predicting Breast diseases. There are several classifiers with varying level of accuracy, which is the research challenge. This paper suggests a method for enhancing the efficiency and precision of four distinct classifiers: Random Forest Tree, K- nearest neighbor (KNN), Logistic Regression, and SVM. To assess the efficacy of different algorithms, the evaluation employed both the AUC score and confusion matrix. The decision to utilize an algorithm is determined by an AUC score exceeding 0.5, which serves as an assessment metric validating the algorithm's efficacy. The employment of a machine learning technique hinges on achieving this threshold AUC score. Among all the machine learning algorithms tested, the Random forest tree attained the highest AUC score in the trial, reaching 1.0.