Mean-Shift Clustering is a clustering algorithm whereby each data point is made to converge to its cluster representative based on the local density of the dataset. Given a point P, which initializes its cluster representative R, the following two steps are repeated until convergence: (1) a local neighborhood of R is defined based on a 'bandwidth' parameter; (2) R is updated by computing the average (or weighted average based on a preset 'kernel') of the points in the local neighborhood. After all points converge, redundant cluster representatives are eliminated based on proximity. Contrary to K-Means Clustering, in Mean-Shift Clustering the number of clusters does not need to be set in advance.