Finding Outlier DMUs in Data Envelopment Analysis

Data Envelopment Analysis (DEA) is a mathematical programming for evaluating efficiency of a set of Decision Making Units (DMUs). One of the problems in DEA, is distinguishing outlier DMUs which have a different behavior in contrast to the general prevailing behavior of the population. The important issue is that the outlier DMUs, which are caused by the incorrect way of collecting data or other unknown factors which can be social, political and etc. , can affect the efficiency of other DMUs. Thus, recognizing and excluding them from the population or reducing their effect and proportioning their status with the population can influence the improvement of total efficiency of population. Therefore, as a result, it prevented the incorrect deduction about the population. In this paper, it is assumed that the efficiency of population must have a unimodal symmetric distribution, and a method based on the skewness of efficiency and inefficiency presented. The important contribution of this method is that it can recognize all the outlier DMUs, in different layers.