PCA-Based Model to Enhance The Performance of DataClustering Using A Metaheuristic Algorithm

This paper explores the enhancement of clustering techniques through the integration of preprocessing methods and nature-inspired algorithms. Achieving perfection in making clusters from the given dataset is a crucial task. Many algorithms have ‎been proposed in the literature to make useful clusters of raw data. But due to the day-by-day changing properties and ‎complexity of data, like multidimensional data values, increasing size, and many more parameters, they still require lots of ‎enhancements in clustering techniques. By addressing the limitations of traditional clustering approaches, the research ‎emphasizes the importance of pre-processing methodologies like PCA and KNN to improve clustering outcomes. ‎The limitation of handling high-dimensional data by an optimizer algorithm is addressed by PCA, which improves feature examination of ‎the dataset at a low level for further processes. Additionally, it examines various nature-inspired algorithms that can be applied to ‎clustering tasks, demonstrating their efficacy in optimizing and refining clustering results. Overall, the study presents a ‎framework combining KNN+PCA+EGWO that combines these elements to achieve superior clustering performance, i.e., ‎increasing the result performance by 45-50 % and the effects of preprocessing steps are also shown with low error and other ‎great parameters present in the paper‎.