Comparative Research Of Comprehensive Evaluation Methods Developed For Exploring The Relations Among Samples On The Basis Of Multivariate Quantitative Data

Objective We aim to solve two important critical questions raised from current comprehensive evaluation research conducted by worldwide scientists.The 1st goal is to develop a synthesis evaluation method which is more appropriate than the current ones.The 2nd goal is to evaluate which one of the methods is the most efficient and accurate.Methods(1)By thoroughly literature search and reviewing of worldwide research on this topic,we summary all the possible data structures suitable to manage by the method of synthesis evaluation.(2)On the basis research digging,we elucidate the difference and connections between the method of sample cluster analysis and traditional comprehensive evaluation.In the meanwhile,we compare the method of traditional comprehensive evaluation method with multivariate statistical analysis method,one type of indirect synthesis evaluation method.(3)Firstly,we extract the structural principle of project index functions from the method of projection pursuit cluster analysis,and then combine this theory with the simple idea of determining the sequence between any pair of samples.As such,we establish a new method of synthesis evaluation and give a standard criterion for screening a better method.(4)Secondly,we retrieve the primary data from 5 literatures which were originally dealing with projection pursuit cluster analysis.And then we use 8 types of synthesis evaluation method to deal with these data and make comparison between results.(5)Thirdly,we provide SAS code for two types of synthesis evaluation method on the basis of fuzzy comprehensive evaluation and grey relational analysis,which are challenging and may be potential and practical.Results(1)Introduce four types of data structure;(2)Establish a new method of synthesis evaluation method,namely Summation method based on standardized transformation;(3)Provide two criteria for distinguishing a better quality of synthesis evaluation methods;(4)Verify eight synthesis evaluation methods in five data sets and make comparison between results generated by the eight methods.Conclusion(1)To the best of our knowledge,we,for the first time,establish an entire new method of Comprehensive Evaluation,namely Summation method based on standardized transformation.This method is designed for the Type I data structure with a good performance in its flexibility.(2)On the basis of the methods of constructing the project index function in the method of projection pursuit cluster analysis and how to easily order between any two samples,we bring up a standard for screening the methods of comprehensive evaluation method.This method provides a way to the unsolved problem remained in the current literatures.(3)In the current situation,the software for synthesis evaluation method is not that sophisticated and mature,we develop a series of SAS codes of fuzzy comprehensive evaluation and grey relational analysis.(4)It is found that the results generated by the traditional comprehensive evaluation methods,including the one derived from Summation method based on standardized transformation,is better than a series of multivariate statistical analysis methods which may be applied in synthesis evaluation indirectly.Among all the aforementioned eight methods,the one developed in this study is one of a kind.