Using regression model for eliminating the puzzlement of Non-Archimedean number in Data Envelopment Analysis

Since Data Envelopment Analysis (DEA) had been developed, scholars have proposed solutions of the multiplier problem. Even though some of these methods have solved the problem of the value setting for multiplier, it causes several new problems. Some scholars even suggested the compromise solution, which would violate the essence of DEA. This research proposes the regression DEA model. It not only solves the puzzlement that the multiplier of every item in each Decision Making Unit (DMU) may be zero, also may avoid the puzzlement of setting Non-Archimedean small number(ε), which simultaneously has not violated the spirit of DEA. To compares with the result of Cross-Efficiency (CE) DEA model, the illustration demonstrated that the distribution tendency of each DMU´s efficiency value extremely closes to each other; and it is able to confirm the discrimination of the Regression DEA method mentioned here.