Harris’ hawks optimization-tuned density-based clustering

Clustering is a machine learning technique that groups data samples based on similarity and identifies outliers with distinct features. Density-based clustering outperforms other methods because it can handle arbitrary shapes of clustering distributions. However, it has a limitation of requiring empirical values for the cluster center and the nominal distance between the cluster center and other data points. These values affect the accuracy and the number of clusters obtained by the algorithm. This paper proposes a solution to optimize these parameters using Harris’ hawks optimization (HHO), an efficient optimization technique that balances exploration and exploitation and avoids stagnation in later iterations. The proposed HHO-tuned density-based clustering achieves better performance as compared to other optimizers used in this work. This research also provides a reference for designing efficient clustering techniques for complex shaped datasets.