The Use of Convex Least Square Regression to Represent a Fuzzy DEA Model

Abstract

Convex Nonparametric Least Squares (CNLSs) is a nonparametric regression technique to estimate monotonic increasing and convex functions. In addition, CNLS method builds on the same axioms as Data Envelopment Analysis (DEA) and also takes into account noise. This paper is to investigate the use of convex least square regression to represent a fuzzy DEA model. By the results of CNLS, we can repeatedly use the corresponding fuzzy DEA model to assess the performance of unobserved decision making units. Note that DEA results cannot be repeatedly used as the regression results for unobserved entities. The popularity of fuzzy DEA would be enhanced.