A Hybrid Particle Swarm Optimization (PSO) and Sine Cosine Algorithm (SCA) for Feature Selection in Lung Cancer Detection

Abstract

Feature selection is known to significantly improve the performance of the machine learning approach in cases of lung cancer. This research presents a PSO-SCA fusion feature selection mechanism. The proposed approach combines the advantage of the global search capacity of SCA with the local optimization capabilities of PSO to avoid premature convergence. The performance of the proposed hybrid PSO-SCA was tested using public lung cancer datasets with SVM and k-NN as benchmark classifiers. The experiments also showed a marked enhancement in the classification rate against the separate approaches as PSO, SCA, GA, as well as ACO and other feature selection methodologies. Using the SVM classifier, the proposed hybrid model of PSO-SCA yielded an overall classification accuracy of 92.8% and was found to be superior to the algorithms in terms of accuracy, precision, recall, and F1-score. It was also observed that the method proposed in this paper has also offered better computational efficiency with fewer iterations to achieve convergence. These findings show that the PSO-SCA technique can be applied in the practical diagnostic domain.