<![CDATA[This research focuses on improving the quality of high-dimensional data clustering results through the integration of Uniform Manifold Approximation and Projection (UMAP) and the K-Means algorithm. The main objective is to evaluate how UMAP, when used as a preprocessing stage, enhances cluster compactness and separation produced by K-Means. The experiment compares two approaches—standard K-Means and the UMAP + K-Means combination—using the Davies–Bouldin Index (DBI) as the primary evaluation metric. Empirical findings indicate that UMAP integration significantly reduces the DBI value from 0.704 to 0.094, representing an 86.6% improvement in clustering quality. Furthermore, visual analysis shows that UMAP enables K-Means to form more compact and clearly separated clusters. These results confirm that manifold-based embeddings like UMAP effectively overcome K-Means limitations in handling nonlinear, high-dimensional data. This study contributes to the development of more accurate and efficient clustering approaches applicable to various domains, including bioinformatics, medical imaging, and socio-economic data analysis.]]>
Copyrights © 2025
