Add more estimators to Silhouette Visualizer

[gshigin]

trafficstars

At the moment Silhouette Visualizer can handle KMeans and MiniBatchKMeans. But as far as I know, the construction of silhouette plot isn't tied to centroid clustering in any way, but simply counts the score according to the specified metric. The only problem I can see here is that some algorithms, like OPTICS and DBSCAN, have a separate label for outliers.

Maybe I could help with this, however I'm new on GitHub, so I might need some help :)

[bbengfort]

@gshigin it's a great idea and it would be great to see a PR for this!

[rpauli]

I came across the same problem, I think it might be as easy as replacing all self.estimator.n_clusters with len(np.unique(self.estimator.labels_)).

Is anyone working on this? Otherwise I'd give it a try

[lwgray]

@rpauli You are more than welcome to submit a PR for this.