Breast Cancer Prediction and Detection Using Data Mining Classification Algorithms: A Comparative Study

Abstract

Today, cancer has become a common disease that can afflict the life of one of every three people. Breast cancer is also one of the cancer types for which early diagnosis and detection is especially important. The earlier breast cancer is detected, the higher the chances of the patient being treated. Therefore, many early detection or prediction methods are being investigated and used in the fight against breast cancer. In this paper, the aim was to predict and detect breast cancer early with non-invasive and painless methods that use data mining algorithms. All the data mining classification algorithms in Weka were run and compared against a data set obtained from the measurements of an antenna consisting of frequency bandwidth, dielectric constant of the antenna's substrate, electric field and tumor information for breast cancer detection and prediction. Results indicate that Bagging, IBk, Random Committee, Random Forest, and SimpleCART algorithms were the most successful algorithms, with over 90% accuracy in detection. This comparative study of several classification algorithms for breast cancer diagnosis using a data set from the measurements of an antenna with a 10-fold cross-validation method provided a perspective into the data mining methods' ability of relative prediction. From data obtained in this study it can be said that if a patient has a breast cancer tumor, detection of the tumor is possible.