Feature selection is a critical preprocessing step in machine learning that significantly impacts model performance, computational efficiency, and interpretability. This paper presents an optimized feature selection approach using the Firefly Algorithm (FA), a nature-inspired metaheuristic based on the flashing behavior of fireflies. The proposed method leverages FA’s global optimization capabilities to identify optimal feature subsets while maintaining or improving classification accuracy. We empirically evaluate our approach on four benchmark datasets from the UCI repository, comparing it against well-established feature selection techniques including Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and traditional statistical methods. Experimental results demonstrate that the Proposed Enhanced Firefly Algorithm achieves an average accuracy of 89.14% and feature reduction ratios averaging 73.4% across all datasets. The algorithm shows consistent convergence behavior and maintains reasonable computational efficiency with an average execution time of 181.76 s. Statistical analysis reveals no significant performance differences between methods at the 0.05 significance level, indicating that algorithm choice may depend on specific application requirements and computational constraints.