A hybrid approach for improving the accuracy of classification algorithms in data mining

摘要

Classification and rule induction are two important methods/processes to extract knowledge from data. In rule induction, the representation of knowledge is defined as IF-THEN rules which are easily understandable and applicable by problem-domain experts. Classification is to organize a large data set objects into predefined classes, described by a set of attributes, using supervised learning methods. The objective of this study is to present a new classification algorithm, RES (Rule Extraction System), for automatic knowledge acquisition in data mining. It aims at eliminating the pitfalls and the disadvantages of the techniques and algorithms currently in use. The proposed algorithm makes use of the direct rule extraction approach, rather than the decision tree. For this purpose, it uses a set of examples to induce general rules. In this study, the rule base is created through the knowledge discovery by employing RES algorithm, a data mining technique, on the sample sets of the Wisconsin Breast Cancer, Ljubljana Breast Cancer, Dermatology, Hepatitis, Iris, Tic-Tac-Toe, Nursery, Lympograph, CRX and Diabetes, which are real life data and commonly used in the machine learning. In terms of the accuracy rate, the results of this study were compared to the results of the algorithms widely used in this field, such as C4.5, NavieBayes, PART, CN2, CORE, GA-SVM. The proposed algorithm showed promising results.