Developing a model for determining optimal η in DEA-discriminant analysis for predicting suppliers’ group membership in supply chain

This study aims at developing a model for determining optimal η in data envelopment analysis –discriminant analysis (DEA-DA) in order to predict the suppliers’ group membership in supply chain. Also, this paper improves the DEA-DA model developed by Sueyoshi (Eur. J. Oper. Res. 115(3), 564–582; [11]). In this model, η parameter is used in order to avoid a trivial solution and to create more discrimination. Since the main shortcoming of the DEA-DA is determining η value based on the researcher’s subjective decision, the present paper develops a model to determine optimal η. Consequently, this paper leads to the estimate of an optimal set of weights for producing a hyperplane for determining the discriminant linear function and, hence promotes the prediction precision of the group membership of a set of observations. The efficacy of proposed model is presented using a numerical example as well as a case study.