Nearest-Neighbour-Induced Isolation Similarity and its Impact on Density-Based Clustering