classifications of linking activities based on their inefficiencies in network dea

Network dea models deal with measurements of relative efficiency of decision-making units (dmus) when the insight of their internal structures is available. in network models, sub-processes are connected by links or intermediate products. links have the dual role of output from one division or sub-process and input to another one. therefore, improving the efficiency score of one division by increasing its output may reduce the score of another division because of increasing its input. to address this conflict, in the present paper we proposed a new approach in slack-based measure (sbm) framework which provides deeper insights regarding the sources of inefficiency. the proposed approach is a two-phase procedure in which phase-i determine the role of intermediate measures by solving a linear program and partitions the intermediate measures into three groups of “input type”, “output type” and “fixed-flows” and phase-ii measures the scores of the dmus under evaluation. providing a classification for intermediate products and account their excesses or shortfalls in efficiency calculation while the continuity of link flows between subunits are kept, are the advantages of the proposed approach.