Back-propagation DEA

Data Envelopment Analysis (DEA) is one of the most widely used methods in the measurement efficiency and productivity of Decision Making Units (DMUs). DEA for a large dataset with many inputs/outputs would require huge computer resources in terms of memory and CPU time. This paper introduces a neural network back-propagation Data Envelopment Analysis. Neural network requirements of computer memory and CPU time are far less than what is needed by conventional methods DEA and can be a useful tool in measuring efficiency of large datasets. Finally, the back-propagation DEA algorithm is applied to a large dataset to identify the source of inefficiency of DMUs and compare it with the result obtained by conventional DEA. 1 1.. I In nt tr ro od du uc ct ti io on n Data Envelopment Analysis is a linear programming technique for assessing the efficiency and productivity of Decision Making Units (DMUs). Over the last decade DEA has gained considerable attention as a managerial tool for measuring performance of organizations and it has been used widely for assessing the efficiency of public and private sectors such as banks, airlines, hospitals, universities and manufactures (Charnes, Cooper, and Rhodes; 1978). As a result, in the last three decade new applications with more variables and more complicated models are being introduced (Emrouznejad and Podinovski; 2004). In the DEA the assumption is that DMUs typically consume multiple resources in producing multiple products. The technical efficiency of a DMU is measured by weighted sum of its outputs divided by weighted sum of its inputs, that is, the ratio of its virtual output to its virtual input. Within the context of input augmentation, a DMU is considered technically efficient, or Pareto–Koopmans efficient, if the performance of other DMUs does not provide evidence that the same amount of all outputs could be produced while using less of at least one input and no more of any other input. Given an output orientation, a DMU is technically efficient if the performance of the other DMUs does not suggest that the same amount of all inputs could be consumed while producing more of at least one output and no less of any other output. Therefore DEA is a non-parametric method which uses linear programming to construct a piece-wise linear segmented efficiency frontier based on best practice. DEA for a large dataset with many input/output variables would require huge computer resources in terms of …