A dynamic rule extraction based on information granularity model for complete data

Rule extraction is an important issue in data mining and knowledge discovery. The effective computation of rule extraction has a direct bearing on the efficiency of knowledge acquisition. In data mining and machine learning tasks, some of the irrelevant attributes not only influence the performance of rule extraction algorithms but also decrease classification accuracy. To acquire brief decision rules, attribute reduction is needed. Since the information granularity is an important approach for attribute measure in rough set theory, in this paper, the effective information granularity based on consistent objects is proposed in complete decision systems, which can effectively measure the discernibility of objects under different value of decision attribute. In addition, when an attribute set may vary dynamically in complete decision system, a dynamic rule extraction approach based on the proposed information granularity is developed in the complete decision system. Finally, the experimental results on different UCI data sets are included to demonstrate the efficiency and effectiveness of the proposed method.