Research On Weight Determination Method Of Comprehensive Evaluation Of Panel Data

Comprehensive evaluation has been widely used in economic production,social management,and many other fields.Scholars both at home and abroad increasingly in-depth research on comprehensive evaluation problems at the same time,about its theory and method has been extended from cross section data or sequential data to the panel data,the panel data of comprehensive evaluation problems become the important content of statistical evaluation field of research in the future.In the process of comprehensive evaluation is the key step is to determine the weights,a comprehensive evaluation for the final result has a crucial effect.Since based on panel data of evaluation objects at different points in the object of study has different properties,not only need to empowerment of indicators,and the evaluation value of empowerment to a different time.Empowerment of the research on comprehensive evaluation model of panel data method,to enhance the scientific nature of the comprehensive evaluation,and then improve the effect of evaluation,has important theoretical and practical significance.Empowerment based on panel data in the study of comprehensive evaluation problems,carry out according to the following ideas:first,draw lessons from the research achievements of predecessors,to the problems of empowerment,summarized the research status quo,architecture writing framework;Related to weight the empowerment theory and common methods of comb,strengthen the theoretical basis for the future study;Then according to the characteristics of the panel data,respectively research method to determine the weighing values of index weight and time.In terms of index weight,this article constructed the empowerment method based on the three common-entropy method,correlation coefficient method and the variation coefficient method of assembled combination empowerment method,the combination method is based on three methods of weight empowerment from the ideas of the minimum sum of squared residuals,the minimum corresponding index weight(yi-(mxi + b))calendar year as the moderate function,using genetic algorithm to search the index weight of the weight vectors of eight indicators;In terms of time weight,build time weight empowerment method based on time information,introduction of simulated annealing algorithm is employed for solving nonlinear programming problems with time weight vector,obtained from 2010 to 2015 time weight vector.Finally,from 2010 to 2015 using the empowerment method 16 commercial Banks operating performance to carry on the empirical analysis based on grey correlation model,the empirical results show that this method adopted by the combination of empowerment in discrete degree is lower and has higher optimal benign:based on the evaluation results of cross section data in 2015 and is based on panel data of the evaluation results have differences,but on the whole close to,panel data of comprehensive evaluation results compared with the cross section data can dynamic,more objective real reflects the situation of commercial Banks.