The Anatomy of K-means

But another question arises: how do you know the correct value of K, or how many centroids to create? There is no universal answer for this, and although the optimal number of centroids or clusters is not known a priori, different approaches exist to try to estimate it. One commonly used approach is testing different numbers of clusters and measure the resulting sum of squared errors, choosing the K value at which an increase will cause a very small decrease in the error sum, while a decrease will sharply increase the error sum. This point that defines the optimal number of clusters is known as the "elbow point", and can be used as a visual measure to find the best pick for the value of K.