SVM-Kmeans: Support Vector Machine based on Kmeans Clustering for Breast Cancer Diagnosis

Breast cancer is the most common cancer in women, and is considered one of the most common causes of death. It increases by an alarming rate globally. Earlier detection and diagnosis could save lives and improve quality of life. In this paper, a new method for breast cancer diagnosis is presented. The proposed method, SVM-Kmeans, combines Kmeans, an unsupervised learning clustering technique, with Support Vector Machine (SVM), a supervised learning classifier. SVM-Kmeans determines number of clusters which achieves the best performance. Moreover, SVM-Kmeans removes irrelevant features using Chi-square feature selection method. This step speeds up SVM-Kmeans and solves curse dimensionality problem. We use Precision, Recall, and Accuracy performance measures to evaluate SVM-Kmeans using two breast cancer datasets. Experimental results show that SVM-Kmeans has a competitive performance compared to other methods in literature. Results show an accuracy rate achievement of 99.8%. Keywords-component; Support Vector Machine; Kmeans; Breast cancer diagnosis; Classification