K-means is pretty poor when the clusters are not linearly separated, but it is the basis of a lot of more modern clustering techniques (kernel K-means if you h

K-means is pretty poor when the clusters are not linearly separated, but it is the basis of a lot of more modern clustering techniques (kernel K-means if you have prior knoweledge, spectral clustering...)