On the performance of linear programming heuristics applied on a quadratic transformation in the classification problem

Abstract We have applied LP heuristics to a quadratic transformation of the observation data in the classification problem. Four ‘real world’ research data sets in the literature have been used to test the method. These data sets include continuous, integer and binary attributes for the observations to be classified. In all four research data sets, the proposed quadratic transformation method outperformed quadratic discriminant analysis.