Tabu search approaches for solving the two-group classification problem

摘要: The two-group classification problem consists in constructing a classifier that can distinguish between the two groups. In this paper, we consider which determining hyperplane minimizes number of misclassified points. We assume data set is numeric and with no missing data. develop tabu search (TS) heuristic for solving NP-hard problem. TS approach based on more convenient equivalent formulation also propose supplementary new intensification phases surrogate constraints. results conducted computational experiments show our algorithms produce solutions very close to optimum require significantly lower effort, so it valuable alternative MIP approaches. Moreover procedures showed paper be extended natural way general problem, generating than one separating hyperplanes.