Breast Cancer Diagnosis Using Optimized Machine Learning Algorithms

Abstract

This paper presents an investigation study of seven Machine Learning Algorithms (MLAs) for Breast Cancer (BC) diagnosis. These algorithms are: Decision Tree (DT), Discriminated Analysis (DA), Naive Bayes (NB), Support Vector Machine (SVM), K Nearest Neighbor (KNN), Ensemble Methods (EMs) and Multi-Layer Perceptron (MLP) classifier. All of these algorithms are applied to the Wisconsin Diagnostic Breast Cancer (Diagnostic) (WDBC) dataset.The main objective of the study is to optimize the hyperparameters of each MLA in order to achieve the best BC classification. This process can also help to reduce the effort and time required for classification. For this reason, Bayesian optimization method is used in MATLAB software to select the hyperparameters values of the six first algorithms. In Python language, Grid search method is used to optimize the MLP hyperparameters. To demonstrate the effect of the optimization process, several predefined models with a corresponding optimized model are evaluated for each algorithm to diagnose the category of BC, whether benign or malignant. The maximum accuracy reported in this study is 96.52%, offered by SVM and MLP algorithms.