Towards a Better Model for Predicting Cancer Recurrence in Breast Cancer Patients

Abstract

Breast cancer is the most common type of cancer in Egyptian women. According to the International Agency for Research on Cancer, the causes of breast cancer are not yet fully known but some of the main risk factors have been identified and taken into consideration in researches performed on patients with high risk of breast cancer. In this paper various classification techniques are used to classify whether breast cancer is recurrent or non-recurrent for a number of patients. Classification techniques used are K-Nearest Neighbor (KNN), Decision Trees (DT), Naive Bayes (NB), Support Vector Machines (SVM) and ensemble techniques Bagging, Voting and Random Forest (RF). The dataset is taken from the University of California Irvine (UCI) machine learning repository and experiments are conducted with Waikato Environment for Knowledge Analysis (WEKA) data mining tool. The research conducted goes through two phases, in the first phase the Random Forest classifier produced the best results with (84.3%) accuracy and the second phase, voting ensemble classifier produced the best results of 89.9% accuracy. The system model show an improvement in the overall accuracy compared to other researches done on the same dataset.