DEA Modeling Based On Uncertain Variables And Its Application

With the increasing demand for efficiency assessment in the era of big data,the variables in practical problems are not only quantitative data,but also have a great deal of uncertainty,which leads to the problem of efficiency assessment with more uncertain indicators.Uncertain indicators can be divided into qualitative indicators,random indicators and fuzzy indicators.The DEA theoretical models and methods for evaluating the relative efficiency of decision making units(DMUs)with uncertain indicators have been widely used in practice and are in continuous development and improvement.In DEA model,assuming the indicators are certain,the relative efficiency is evaluated by using linear programming to solve the relative maximum output(or minimum input).The DEA model has obvious advantages in dealing with the efficiency evaluation without considering the statistical characteristics of multiple inputs-outputs and has been widely applied in economic,social and management efficiency evaluation.Considering the possibility of multiple DMUs with the same efficiency value,an effective balance index is proposed to solve the ranking problem.This study improves the theoretical basis,improves the validity and accuracy of the efficiency measure,and raises the model from the evaluation method to the decision-making tool.Therefore,this study has certain theoretical significance and application value.This study constructs three types of DEA model with uncertain data to solve the efficiency evaluation problem.One is DEA model with environmental regulation constraints.The utility function is used to quantify the industry's willingness to perform under different intensity of energy efficiency policies.The model is applied to evaluate the energy efficiency problems.The other one is DEA model with random distribution.Considering the distribution characteristics of extreme climate,we solve the energy efficiency evaluation of panel data from the aspect of time series and cross-section.The last one is poverty alleviation efficiency evaluation with fuzzy variables.We quantify the government regulation and satisfaction degree through type-2 fuzzy indicators,use the membership function to defuzzify fuzzy variables.The research provides a reference paradigm for efficiency evaluation with uncertain data to further expand the theoretical model.Empirical research provides suggestions for promoting the sustainable development,and realizes the effective allocation of poverty alleviation resources.