Comparative study of machine learning algorithms using the breast cancer dataset

Abstract

Breast cancer is the second most common cancer overall and the most common cancer in women worldwide. To better diagnose and predict the development of breast cancer, current medicine uses several techniques and tools based on very powerful and advanced methods such as machine learning algorithms. This work consists to produce a comparative study between 11 machine learning algorithms using the Breast Cancer Wisconsin (Diagnostic) Dataset, and by measuring their classification test accuracy. We have elaborated this study to define the best method to create two classifiers that must define benign from malignant breast lumps based on the features of the dataset which have been extracted from diagnostic images of a fine needle aspirate of a breast mass. The results of the classification experimentation show that the best accuracy in this paper was achieved by the Neural Network algorithm, which had, in its best configuration, 96.49% of accuracy.