Bolukbas, OrhanHaber, ZeynepUguz, Harun2025-03-222025-03-2220252193-567X2191-4281https://doi.org/10.1007/s13369-025-10015-1https://hdl.handle.net/20.500.13091/9908Feature selection is the process of determining which k features, out of n characteristics, best represent a dataset by evaluating the features in accordance with the method used. Selecting the fewest features without compromising the accuracy of the results is another optimization challenge in feature subset selection. Problems with a binary search space can be solved immediately via the very effective global search optimization technique known as scatter search. However, it might become stuck in local optimum solutions and be unable to locate the worldwide optimum answer. This work aims to achieve a balance between the local and global searches of the scatter search algorithm. By combining the concepts of scatter search and snake optimizer algorithms, a novel technique known as scatter search snake optimization (SSSO) is presented for this purpose. We evaluate the proposed method against well-known optimization methods on two different datasets, one with respect to epileptic disease and the other with respect to well-known machine learning datasets from the UCI Machine Learning Repository during the feature selection phase. The comparative results show that the proposed SSSO technique is an effective metaheuristic for feature selection problems.eninfo:eu-repo/semantics/openAccessFeature SelectionMachine LearningScatter SearchSnake OptimizationNature-Inspired AlgorithmsArticle10.1007/s13369-025-10015-12-s2.0-85218842183