Research On Methodology And Application Of Efficiency Evaluation For Decision Making Units With Network Structure

Efficiency evaluation is a classical problem in management and economics.The research on network system is lively and heated since the network DEA can identify the cause of any inefficiencies in the system and give more accurate evaluation results as comparing to the conventional DEA.On the basis of the previous researches on the network system,this paper focuses on measuring efficiencies of decision making units(DMUs)with considering their network inner structure.The research result can provide effective decision supports for decision makers.This paper mainly divided into six chapters,and the corresponding content is summarized as follows:The first chapter introduces the basic theory of the DEA and the network DEA,then the relative literature review and summary on main network DEA structure have been carried out.Moreover,the significance of this paper has been analyzed in this chapter.The second chapter studies the efficiency evaluation on the system with nonhomogeneous outputs in subsystems.This chapter incorporates the characteristic of nonhomogeneous outputs in subsystems of Hong Kong hospitals and extends performance evaluation models of nonhomogeneous DMUs.The proposed approach is verified by an empirical study of Hong Kong hospitals.This is the first time in hospitals research to consider the nonhomogeneity of outputs in subsystems and it helps to make an extension on the application of DEA.The third chapter tackles the variable return to scale problem in the two-stage relational DEA model to promote the improvement in theory,and the extended two-stage relational model is applied to measure achievements of participants at the Olympics.The extended two-stage relational model is non-linear and a heuristic search procedure is employed to solve the non-linear model.This chapter considers the performance of participants at the Olympics as a two-stage process(the athlete preparation stage and the athlete competition stage).The evaluation result shows that the efficiency of the entire two-stage Olympic process is more significantly related to that of the athlete competition stage.The fourth chapter aims at studying the uniqueness problem of subsystems'efficiencies in the parallel system,and the proposed approach has been used to measure the performance of participants at the Olympics.The efficiency ranking for each parallel subsystem has been obtained since this chapter takes all possible efficiencies of subsystems into consideration.Moreover,this chapter considers the Summer Olympic Games and the Winter Olympic Games as two parallel subsystems to evaluate the performance of participants at the Olympic Games.Thus,the unfair problem in two situations has been solved.One of the situations is using the performance at the Summer Olympic Games to represent the sports levels of participants that are located on the high altitude and cold regions,and another one is using the performance at the Winter Olympic Games to represent the sports levels of participants that are located on the low altitude and tropical regions.The evaluation results find that the performance of participants at the Summer and Winter Olympic Games accords with their geographical features.The fifth chapter focuses on studying dominance relations and efficiency ranking intervals of DMUs with considering their complex network structure.That is,this chapter measures dominance relations among DMUs with considering their inner parallel structure over all possible input and output weights.The proposed efficiency dominance relation model for the parallel structure can figure out dominance relations among DMUs.Based on the efficiency dominance between DMUs,the efficiency ranking interval for each DMU can be obtained.Therefore,the influence of the uncertainty in weights on the performance evaluation can be effectively avoided.The sixth chapter makes a summary on the main content and the main innovations of this paper.On the basis of the analysis on the deficiency of this paper,directions for further research on this topic have been suggested.The main contributions of this dissertation includes:(1)based on the idea of using all possible input and output weights to make performance evaluation,this dissertation studies dominance relations and efficiency ranking intervals of parallel production systems;(2)it is the first time in the performance evaluation of the hospital system with considering the nonhomogeneity of outputs in subsystems,and models for evaluating the performance of nonhomogeneous DMUs have been extended;(3)considering the performance of participants at the Olympic Games as a two-stage system and as a parallel system to measure Olympics achievements,extensions on the two-stage DEA as well as the parallel DEA have been carried out.