Enhancing density-based clustering: Parameter reduction and outlier detection

► We stratify the input space using a local density of a point together with a function, INFLO. ► We remove outliers, we improve the DBSCAN performances. ► We cluster dataset with different densities. ► The method needs only one parameter (with respect to DBSCAN that has two dataset-dependent parameters). ► It is not sensitive to the clustering starting point.