classifying flexible factors using fuzzy concept

In data envelopment analysis (dea), it is assumed that the role of each factor is known as input or output. however, in some cases, there are shared factors that their input versus output status is not clearly known. these are flexible measures. in such cases, determining whether a factor is input or output is ambiguous. therefore, using fuzzy concept seems to be necessary. in this paper, a two phase procedure is proposed to fuzzy classification of flexible measures.in the first phase, applying the existing classification methods, an orientation of flexible measures to aid in the definition of inputs and outputs is achieved. through defining a membership function in second phase, the input versus output status of a factor is expressed by fuzzy notion. by the proposed method, the efficiency of a decision mating unit is defendedby a membership degree. we illustrate the proposed model in a practical problem setting.