A Study on Synthetic Evaluation Based on Data Mining and Evidence Theory
Han Yuan-jie · Modern Electronic Technique · 2005
There are two problems in synthetic evaluation.Firstly,for the largescale evaluated objectives,the time and space complexity of the operation process expands explosively through classifying,optimizing and searching,massive memory to be stored.They need not only to be stored in database or data ware,but also the results are to be put on the desktop as production,and this leads unreasonable and unscientific results with the existing evaluating methods.Secondly,the data of same class are approximate,and the evaluating objects are accurately for undistinguished difference.From the evaluating results,synthetic evaluation is classified as classified evaluation,ordering evaluation,ordering and classified evaluation.For solving the above problems,the technique of data mining with appropriate arithmetic and information fusion with DS theory is integrated into the new synthetic evaluating method.The method implements classified evaluation with data mining,and the following process performs ordering evaluation based DS evidential reasoning.The data of operation in the example demonstrate that the objects are separated to leveling sorts with data mining when facing the evaluating database objects.The accurate evaluating results are obtained through information fusion.The evaluating objects are more distinctly,and the uncertainty declines greatly.The degree of belief increases,and the results of assessment are improved.