An in-Depth Analysis of Eps and MinPts Selection in DBSCAN with Genetic Algorithm Optimization

DBSCAN (Density-Based Spatial Clustering of Applications with Noise) has been recognized as a powerful clustering algorithm for detecting non-linearly separable clusters but has faced challenges in selecting optimal MinPts (the minimum number of points required to form a cluster) and Eps (the radius defining the neighborhood of a point) values, which have typically been chosen experimentally. To overcome this, we have introduced a Genetic algorithm (GA) driven optimization approach that has dynamically adjusts these parameters using a well-defined fitness function. Comparative evaluations with K-means and hierarchical clustering across diverse datasets have confirmed that our method has enhanced clustering effectiveness, making it a reliable choice for complex data analysis.