The Research Of Attribute Reduction And Minimum Rule Sets Acquisition In Decision Rough Set Theory

Rough Set theory is presented by the professor pawlak who is from the Warsaw University of technology in the 1980s. And it is an effective data analysis theory which can deal with all kinds of incomplete information with the feature of imprecise, inconsistent and incomplete. The main advantage of Rough set is that it does not need prior knowledge. The introduction of probability inclusion relation into decision rough set theory, which is an extension of the classical rough set model, increases the tolerance of rough set model.The paper conducted a deep research on the relevant theories and methods of attribute reduction and rule extraction in the decision rough set theory. The main work is embodied in the following two aspects.Firstly, in the decision rough set model, the positive region does not increase with the increase of attribute, thus the method of maintain constant positive region reduction in decision rough set is no longer appropriate. In order to efficiently obtain reduction set, the paper designed a heuristic function, important degree of decision making, according to the definition of remain the positive decision of all objects unchanged. The heuristic function can define the importance of every attribute based on the size of its positive decision object collection. The bigger the positive decision object collection, the higher the attribute importance degree. On the basis of this we put forward the heuristic attribute reduction algorithm based on decision important degree. The advantage of this algorithm is that we can determine the search direction based on the attribute decision importance sorting, we can avoid the attribute combination of calculation, reduce the amount of calculation, and then find out a smaller reduction set. The experimental results show that the algorithm is effective and can obtain good reduction effect.Secondly, we search decision rules from the data sets that have been reducted. Using the minimum rule set as a classifier, we can better predict new data. In this paper, we use the frequency of rule occurrence as heuristic information, and then use absorption law to obtain a reduction of rule set from conjunctive normal form. In the end a heuristic algorithm for acquiring minimum rule sets is given. Experiment results show that the algorithm has higher efficiency and better accuracy, and can play a good application value in practical application.Finally, we summarized all the research work done in this paper, and put forward the future research direction in this paper.