DEA-Discriminant Analysis Using Goal Programming for More than Two Groups that Expand for Interval Data: Apply to Stock Companies

One of the interesting subjects that engages the minds of researchers is hypothesizing the correct classification of a new sample by using available data. Data Envelopment Analysis and Discriminant Analysis can be used to classify data independently of each other. DEA classifies them as efficient and inefficient groups, and DA classifies the observations by their historical data. Merging these two methods is a powerful tool for classifying the data. Most of the methods represented are just useful for classifying observations into two groups. In this paper, we represent our new DEA-DA method by using Goal Programming to classify data into more than two groups. Since, in the real world, in many cases we do not have certain data, we present a method that can be used for certain and interval data. Then, we present an empirical example of our purpose method on the Iranian stock companies' data. We divided stock companies into four groups with certain and interval data. Since most of the classical DA models are used for two groups, the advantage of the proposed model is highlighted. The result shows that the model can predict and classify more than two groups with certain and interval data completely correctly.