Diagnosis for Early Stage of Breast Cancer using Outlier Detection Algorithm Combined with Classification Technique

Abstract

Breast cancer is the most dangerous cancers that lead to women in death. Particularly in the developed countries it takes second leading place that increase the chance of death in women. It can be not easily diagnosed by the lab. It has difficult to identifying at the beginning stage. This cancer begins from breast and disseminate to other body parts. It has cured easily if it is identified at beginning stage. The correct classification of benign cancer can prevent from superfluous treatment for patients. This paper focused on diagnosis early stage of the breast cancer based on data mining algorithms. The automatic diagnosis process plays on important role in data mining. The proposed method has a process of three stages. First, data objects are grouped into clusters using k-means clustering algorithm. Size of the dataset has to shrink gently the computation time also reduced. The second stage, the outlier detection (OD) algorithm has used to detect the outliers from the cancer dataset. Finally, diagnose the cancer is either benign or malignant using decision tree classification algorithm. The breast cancer dataset has been used to test the efficiency of the proposed method. The experiments were conducted in breast cancer dataset before and after removal of outliers. Comparison results prove that the proposed method as serves as the better one with high accuracy. This breast cancer research will help with a medical practitioner to diagnose the breast cancer and so that it helps to recover the patients.