Finding an appropriate set of features from data of high dimensionality for building an accurate classification model is a well-known NP-hard computational problem. Unfortunately in data mining, some big data are not only big in volume but they are described by a large number of features. Many feature subset selection algorithms have been proposed in the past, they are nevertheless far from perfect. Since using brute-force in exhaustively trying every possible combination of features takes seemingly forever, stochastic optimization may be a solution. In this paper, we propose a new feature selection algorithm for finding an optimal feature set by using metaheuristic, called Swarm Search. The advantage of Swarm Search is its flexibility in integrating any classifier as its fitness function, and installing in any metaheuristic algorithm for facilitating heuristic search. Simulation experiments are carried out by testing the Swarm Search over a high-dimensional dataset, with different classification algorithms and various metaheuristic algorithms. Swarm search is observed to achieve satisfactory results.