FDEA-DA: Discriminant analysis method for grouping observations with Fuzzy Data Based on DEA-DA

DEA discriminant analysis, denoted as DEA-DA, is a non-parametric approach which combines the methodological strength of DEA with the Discriminant Analysis. To overcome the limitation of only dealing with crisp data by DEA-DA, many researchers focus on how to group observations with fuzzy data. However, because of the difficulty of identification on fuzzy data, fuzzy DEA-DA canpsilat rank peer of DMUs fully. In this paper, a two-boundary FDEA-DA was proposed. This model has two stages, which dispose fuzzy records and maintain its discriminant capability in fuzzy system. With the characteristics of fuzzy data, the upper bound and lower bound were classified by interval discriminant function with respectively -level sets in the two stages. The values are selected by the expertspsila preference, and a critical value d and c are found as the optimal results to classify all of the observations in the two stages. The FDEA-DA model is demonstrated with numerical examples.