Question $6$

Recall the properties of the k$-$means clustering algorithm. Select all the statements below that are TRUE about k$-$means clustering.

$1$ point

k$-$means clustering works especially well for data having complex, irregular structure.

Like hierarchical clustering, k$-$means clustering assumes a tree$-$like structure for the data.

The biggest advantage of k$-$means clustering is that it automatically finds the best number of clusters $($k$)$ during the estimation process.

k$-$means clustering works well when the data forms very elongated chains in feature space.

k$-$means has the advantage that it is robust to the choice of initial cluster centers $-$ that is$,$ you will get the same final clustering no matter which random initialization you use for cluster centers.

One heuristic for finding an optimal k$-$means clustering is to run the algorithm multiple times with different start configurations, and then pick the highest quality result from this set.