Comparative Study of Eight Classification Models for Diagnosis and Prediction of Breast Cancer

Abstract

AbstractNowadays, breast cancer is creating wea big problem for women all over the world. Correct and early prediction of disease is very much important for the treatment of curing the disease. Women identified at the stage of benign have high chances of getting curable but identifying at the stage of malignant is regarded as a dangerous state of cancer. Many machine learning algorithms are used for the diagnosis of breast cancer effectively. In this article, eight classification models such as Logistic Regression (LR), K-Nearest Neighborhood (K-NN), Decision Tree (DT), Random Forest (RF), Artificial Neural Network (ANN), Gaussian Naïve Bayes (NB), Support Vector Machine (SVM), and AdaBoost classifier are used for predicting two classes, i.e., benign and malignant. To choose the best fit classification model for prediction, a confusion matrix is used to evaluate the performance of each model. Also, parameters such as accuracy, precision, recall, specificity, F-measure, and Matthews correlation coefficient (MCC) are discussed for each model. For experimental results, the Wisconsin Breast Cancer Diagnosis dataset and Coimbra Breast cancer datasets are used, and at last, a comparison is being done for all of these models. KeywordsDecision Tree (DT)K-Nearest Neighborhood (K-NN)Logistic Regression (LR)Random Forest (RF)Support Vector Machine (SVM)Artificial Neural Network (ANN)Gaussian Naïve Bayes (NB)AdaBoost Classifier