Fully Automated Density-Based Clustering Method

Cluster analysis is a crucial technique in unsupervised machine learning, pattern recognition, and data analysis. However, current clustering algorithms suffer from the need for manual determination of parameter values, low accuracy, inconsistent performance concerning size structure. To address these challenges, novel algorithm called fully automated density-based method (FADBC) proposed. The FADBC consists two stages: selection cluster extraction. In first stage, proposed extracts optimal parameters dataset, including epsilon minimum number points thresholds. These are then used to scan each point dataset evaluate neighborhood densities find clusters. was evaluated on different benchmark datasets metrics, experimental results demonstrate its competitive without requiring inputs. show that outperforms well-known methods such as agglomerative hierarchical method, k-means, spectral clustering, DBSCAN, FCDCSD, Gaussian mixtures, spatial methods. It can handle any kind set well perform excellently.