A very successful global optimization technique is particle swarm optimization (PSO). PSO has some drawbacks, including slow convergence, early convergence, and getting stranded at local optima. In order to enhance the search process, the chaotic map and dynamic-weight are introduced into PSO in this paper. By modifying the position update formula and the chaotic dynamic-weight Particle Swarm Optimization (CPSO) inertia weight, they effectively balance the local and global PSO feature selection processes. The effectiveness of CPSO was compared to three metaheuristic methods: Differential Evolution (DE), Genetic Algorithm (GA), and PSO, using eight numerical functions. Four datasets are used to evaluate this approach. The findings show that by balancing the exploration and exploitation search processes, the CPSO is an effective feature selection strategy that produces high-quality results. The suggested CPSO approach successfully classified features using the KNN Classifier for the four datasets.