Discussion On Improved DEA Models And Its Sensitivity Analysis

Data envelopment analysis (DEA) is a new branch of Operations Research. It was first put forward by A.Charnes, W. W. Cooper and Rhodes in 1978. It is a performance measurement technique which, as we shall see, can be used for evaluating the relative efficiency of decision-making units (DMU's) in organizations. Examples of such units to which DEA has been applied are banks, police stations, hospitals, tax offices, prisons, defense bases (army, navy, air force), schools and university departments.The whole thesis consists of four chapters. Chapter one tells us the history of DEA, the basic concept of DEA and sensitivity analysis in linear programming problem. The work of this paper focuses on chapter two and three.Chapter two developed an improved DEA model. We can see one variation in one input or output data may result in variation of two data in classic DEA models. It increases the difficulty of Sensitivity Analysis. In order to reduce the difficulty, I developed an improved DEA model as effective as the classic DEA model.Chapter three made sensitivity analysis about this improved DEA model. It only analyzed the change of DEA efficiency on the base of variation in date in previous papers. But in this paper I Derived the range of variation in one data while remaining the efficiency of a specific DMU. This results is useful for DMU analysis the change of input or output date and meaningful in real production.Chapter four gave the practical analysis related to chapter three.