Performance comparison of various clustering techniques for diagnosis of breast cancer

Abstract

Breast Cancer is a decisive disease when compared to all other cancers in worldwide. Diagnosis of breast cancer is normally clinical and biological in nature. In general we used some of the data mining clustering techniques to predict breast cancer. The objective of this paper is to compare the performance of different Clustering techniques to diagnosis the cancer either benign or malignant. According to the results of our experimental work, we compared five clustering techniques such as DBSCAN, Farthest first, canopy, LVQ and hierarchical clustering in Weka software and comparison results show that Farthest First clustering has higher prediction accuracy i.e. 72% than DBSCAN, Canopy, LVQ and Hierarchical clustering methods.