Analysis of Machine Learning Techniques for Breast Cancer Prediction

Abstract

Of all the deaths worldwide, the second most common cause of death is breast cancer. In 2011, the number of deaths was estimated to be 5.8 millions because of breast cancer. So, early diagnosis of cancer is very important.  In this paper, machine learning techniques are explored in order to increase the accuracy of diagnosis. Methods such as CART, Random Forest, K-Nearest Neighbors are compared. The dataset used is obtained from UC Irvine Machine Learning Repository. The obtained accuracy prediction performances are proportionate to existing methods. However it is found that KNN algorithm has much better performance than the other techniques used in comparison.