Abstract
In the conventional discriminant analysis, the attribute values of each sample are represented by a point in the multi-dimensional feature space. But they should be represented by an interval vector in the feature space when the observed values are not constant but fluctuating with time. Thus we intend to propose a discriminant method for multi-dimensional interval data. The discriminant method for the interval data is based on the interval linear discriminant function which maps an interval vector in the multi-dimensional feature space to an interval in the one-dimensional decision space. Therefore our discriminant problem is to determine the interval linear discriminant function and group intervals which represent the given groups in the one-dimensional decision space. In this paper, two mathematical programming problems are formulated for such a discriminant problem and computational algorithms using the linear programming are shown.
