New solution algorithms for the classification problem

Abstract

<p>This thesis proposes a new mixed integer programming model of the two group classification problem. A branch-and-bound procedure is designed to solve the mixed integer programming model. The proposed procedure is more efficient than recently published procedures in the literature. Three heuristics are also designed from the branch-and- bound procedure. The Three Seed and N Seed heuristics provide solutions that are "close" to the optimal solution for substantially less computational effort. This thesis also proposes a quadratic transformation method which provides a solution that is competitive with the classical quadratic discriminant analysis method.</p>