A Density-based Clustering Algorithm Using Adaptive Parameter K-Reverse Nearest Neighbor

Density-based clustering method DBSCAN is one of the most widely used clustering methods, this algorithm does not need to input the number of clustering, and can find clusters of any shape. However, DBSCAN algorithm requires two parameters, with different parameters, the clustering results are quite different. This paper proposes a density-based clustering algorithm using adaptive parameter k-reverse nearest neighbor: ARKNN-DBSCAN, which can effectively identify and separate clusters with different densities by analyzing the correlation through the reverse k-nearest neighbor of the observation in the dataset without input parameters. Our results show that this algorithm is more accurate than other density-based clustering algorithms using adaptive parameter.