min_cluster_size = 1

[cpvandehey]

When I try to set the min_cluster_size to 1, the hdbscan lib throws this exception: ValueError: Min cluster size must be greater than one.

I do not understand why this is enforced. Could this simply be a warning instead? How can I use HDBSCAN assuming that none of my data points are noise?

This is supported with regular DBSCAN. DBSCAN(eps=1.2, min_samples=1).fit(some_matrix)

[nanodan]

Also having this same error, it's a hard check in the code and I also don't see why it's necessary.

[nanodan]

@Querela this is a parameter for the minimum number of samples per cluster, not the minimum number of clusters.

[Querela]

Oh, you are right. Sorry, I switch the parameter names. Then, I guess, my previous statement is no that correct.

[sergeyf]

Any updates here? I also have a clustering problem where individual points are eligible to be clusters.

[algorithmvisualize]

There is possible one way to solve it --- double your every input