The k-means clustering algorithm classifies given points into k groups in such a way that a distance between the members of the same group is minimized.

The k-means clustering algorithm determines the initial k-centroids (points to be in a cluster center) – one for each cluster. Then each feature is classified into the cluster whose centroid is closest to that feature. After classifying all the features, we have formed initial k clusters.

For each cluster we recompute the centroid to be the average of the points in that cluster. After we have moved the centroids, we recompute the classes again. Features may change the classes. Then we will have to recompute the centroids again. If the centroids do not move anymore, then the k-means clustering algorithm terminates.