Discovery of Knowledge Patterns in Clinical Data through Data Mining Algorithms: Multi-class Categorization of Breast Tissue Data

This paper highlights the significance of classification in data mining and knowledge discovery. In this paper we investigate the performance of various data mining classification algorithms viz. Rnd Tree, Quinlan decision tree algorithm (C4.5), KNearest Neighbor algorithm etc., on a large dataset from the „Wisconsin Breast tissue dataset‟ (derived from the UCI Machine Learning Repository) that comprises of 11 attributes and 106 instances. The results of this study indicate the level of accuracy and other performance measures of the algorithms in detecting the presence of breast cancer and the associated breast tissue conditions that increase the risk of developing cancer in future. Moreover the importance of feature selection/reduction in improving the performance of classification algorithms is also described. The classification algorithm Rnd Tree produced 100 percent accuracy for classification of all the training data under multiple classes. The classification algorithm was also applied to verify it‟s correctness in classifying test data. General Terms Data Mining, Breast tissue research