A study of the DEA super-efficiency based on DDF from output

• Main Authors: Su, Sheng-Chieh, 蘇聖傑
• Other Authors: Lee, Hsuan-Shih
• Format: Others
• Language: en_US
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/hdyf4m

碩士 === 國立臺灣海洋大學 === 航運管理學系 === 106 === To solve the infeasibility issue of conventional radial super-efficiency in data envelopment analysis (DEA) models, the Nerlove–Luenberger (N–L) super-efficiency model can provide some approaches but also be failed in two exceptions. The first situation is that when evaluating the database with zero inputs, no feasible solutions can be found. Secondly, the evaluating projection of the output variables under the variable returns to scale (VRS) may be negative. Chen et al. (2013) proposed a new super-efficiency model (CDH model) based on the directional distance function (DDF). Not only the infeasibility issue with zero inputs can be solved, but also their new model can prevents the negative projection of the output variables under the VRS. However, their new model cannot fully eliminate the infeasibility issue when zero data occurring in the outputs. Besides, the evaluation result under their model relies on some predetermined parameters, which are complexly determined. This study develops a new DDF super-efficiency model, which can help both of the N-L super-efficiency model and CDH super-efficiency model solve the infeasible issue when all inputs and outputs are non-negative. Numerical examples are utilized to compare the difference of evaluation results under the different super-efficiency approaches.