Bcn

* Add `n_clusters` argument to `bcn` to specify the number of clusters to be found (for now,
by using k-means clustering)
* Avoid division by zero in scaling the data