Processing Methods For Incomplete Information Systems Based On Rough Sets

The main work of this paper is to give an in-depth study on the processing method of incomplete data problem using rough set theory,and to improve the quality and efficiency of KDD.Firstly,this paper analyzes the defects of the ROUSTIDA algorithm and their causes.With improvement of quantization of tolerance relation and considering the decision-making conflicts,a new algorithm is proposed to improve the imperfect data based on a completion strategy of "decision-making attribute first and condition attribute last".In order to show the effectiveness of this algorithm,experiment data are selected from UCI machine learning database. Experiment results clearly demonstrate that it not only increases complement chances but decreases decision-making conflicts.Secondly,this paper discusses the limitation of existing knowledge reduction algorithm,without changing the original incomplete information system.By combining the knowledge granularity with incomplete entropy,the importance of the attributes in incomplete information system is defined and then based on this definition;an improved reduction algorithm is put forward for incomplete information system,which is much more effective in finding minimum reduction than the existing methods.This effectiveness is proved by the following theory and example analysis.