Beginners guide to k-Means Clustering - Analytics Vidhya

The very first clustering algorithm that most people get exposed to is k-Means clustering. This is probably because it is very simple to understand, however, it has several disadvantages which I will mention later. Clustering is generally viewed as an unsupervised method, so it is difficult to establish a good performance metric. However, a lot of useful information can be extrapolated from this algorithm. The problem is how to assign semantics to each cluster and thus measure the "performance" of your algorithm.