| After summarizing the achievements in the Multiple-Objective-Comprehensive-Evaluation (MOCE) theory research as well as its application during the past decades, the author systematically studies the theory and methodology on MOCE. In the author's opinion, MOCE technique is a frontier science, which is a branch of decision making theory as well as that of statistics. For the sake of systematicness, this paper almost covers every important aspects in MOCE technique.1) In chapter one, the author put forwards the basic theories in MOCE,such as its position,its process, its contents and methods structure.2) In chapter two, the author deals with the indicator-system theory of MOCE, introducing how to set up a set of scientific and optimized evaluation indicator-system, how to transfer a negative indicator into a positive one or versa vise, and how to make a qualitative varible be quantitative.3) In chapter three.the author studies the principal model of Effective-Function-Comprehensive Evaluation(EFCE).According to this model, he thoroughly studies the theory of how to transfer indicators that have different measurements into those having the same measurement (so called non-measurement), the rules of how to combining effective-value, particularly the mathematical characteristics of the Power-Mean-Function.4) In chapter four and chapter five, the author studies the methodology of weights.He points out that the various methods of weight making can be divided into two categories: "Basic Weight Making Method" (BWMM) and "Expanded Weight Making Method"(EWMM).The latter is an advanced idea for weight making.In chapter four, the author studies AHP weight making method and invents three sets of approximated solution to deriving AHP weight vector.He deals with AHP scaling theory as well as methods for consistency test. Moreover, the author invents a new method for weights making based on the distribution-judgement-matrix. Finally, the author deals with other BWMM.In chapter five,the author studies the EWMM, explains their process and advantages as well.5) In chapter six, the author systematically studies the application of Multivariable-statistics in MOCE, such as Discrimination Analysis (DA), Principal Component Analysis (PCA), Factor Analysis (FA), Cluster Analysis (CA). etc. He discusses that only the first PC can be used for ranking, other PCs stands for quite different meaning, that a negative coefficient for a positive indicator in the first PC is improper for evaluation, that too high correlation amongst variables (indicators) results in unfair evaluation conclusion. He puts forward some remedied methods for PCA, such as weighted PCA, robust PCA.B-matrix-typed PCA. The author also deals with FA as a method for optimizing the evaluation indicator system, putting forwrjd a new MOCE idea named FA-PCA. He discusses DA and points out some missing of it in the past debates.6) In chapter seven, the author studies the Fuzzy Comprehensive-Evaluation (FCE). Hefocuses on the fuzzy combination, pointing out some defects in FCE process, such as defects in maximun-minimun fuzzy operator, fuzzy functions and fuzzy matrixes. He suggests to apply the K-power-mean fuzzy operator in FCE. He also discovers some shortcomings in fuzzy model identification principals. At last, he compares the difference between EFCE and FCE, arriving at the conclusion that FCE is not superior to EFCE, which just as many researchers think so.7) In chapter eight, the author discusses briefly other important MOCE methods and theories which are not touched in previous chapters. He studies the Grey System MOCE (GSCE), AHP (Analysis of Hierarchical Process) MOCE, DEA (Data Envelopment Analysis) MOCE, ANN (Artificial neural net system) MOCE, studies experts MOCE, combining MOCE and multilevel mixed MOCE, suggesting several new viewpoints.
|
PDF Full Text Request