Abstract

Breast cancer is a kind of common malignant tumor of women. It is becoming a leading cause of death among women. But the early detection and diagnosis of this disease can ensure a long survival of patients. Classification plays an increasingly important role in machine learning and data mining. A rough-set classifier based on discretization (RSCBD) is proposed in this paper for breast cancer diagnosis. It is built on fully considering the significance of condition attributes, classification attributes and attribute thresholds. We have tested its effectiveness on Wisconsin Breast Cancer Dataset (WBCD). Experiment results prove the RSCBD can get higher classification accuracy, lower reject rate, breakpoints and rules, which are important for breast cancer diagnosis.

Paper Details