Heuristic Solutions for the (α, β) -k Feature Set Problem

Feature selection aims to choose a subset of features, out of a set of candidate features, such that the selected set best represents the whole in a particular aspect. The (α, β) -k feature set problem (FSP) is a combinatorial optimization-based approach for selecting features. On a dataset with two groups of data, the (α, β) -k FSP aims to select a set of features such that the set maximizes the similarities between entities of the same group and the differences between entities of different groups. This study develops a matheuristic algorithm for the (α, β) -k FSP. We test the algorithm on 11 real-world instances ranging from medium to large. The computational results demonstrate that the proposed matheuristic competes well with the standard solver CPLEX