摘要：针对DEA评价与排序时单独采用相对最优效率模型与相对最差效率模型，存在丢失重要信息的不足，引入DEA模型区间效率的概念，把两种评价模型有机结合，可实现对决策单元更合理的评价与排序。进一步改进了DEA区间效率模型，并对其计算效果进行分析，找出了计算决策单元区间效率的合理模型。在此基础上引入决策者的偏好系数β来计算区间效率的评价指标，分析得出当0≤β≤05时采用相对最差效率模型，05≤β≤1时采用相对最优效率模型来计算区间效率这一结论。通过具体的数值算例，对决策者偏好不同的情况下决策单元区间效率的评价指标进行计算和敏感性分析，计算结果表明，改进的DEA区间效率模型对决策单元排序更为合理。

Abstract： In allusion to the fact when best relative efficiency model and worst relative efficiency model are separately adopted for the Data Envelopment Analysis(DEA) evaluation and ranking, important information will be losed. The definition of interval efficiencies is introduced, which combines two models organically and realizes more reasonable evaluating and ranking on Decision Making Units(DMUs).The DEA models with interval efficiencies are further improved, and the performance of results is analyzed, then reasonable models for calculating interval efficiencies of DMUs are found. In addition, an optimism coefficient of decision maker is introduced for calculating the evaluation index of interval efficiencies. And the analysis results show that worst relative efficiency model is adopted when 0≤β≤05 and best relative efficiency model is adopted when 05≤β≤1. Combining with the numerical example, the evaluation index of interval efficiencies are computed and sensitivity analysis are conducted, and the ranking results indicate that the improved DEA model based interval efficiencies for DMUs ranking is more reasonable.