vreyespue/scipy_cut_tree_balanced

Python function that performs a balanced cut-tree of a SciPy linkage matrix

scipy_cut_tree_balanced's Issues

How to draw dendrogram

Hi, I'm trying to draw dendrogram like yours (mark on each cluster id at different levels) by using MLcut, but it doesn't work well. Could you specify a little bit more on how to draw dendrogram at different levels like you did in both pictures of yours?

Thank you for your amaziong code, by the way. It helped me a lot.

pip installable until SciPy integration?

I've been following your thread and looking forward to using it. I'm interested in assisting in a publication as I have lots and lots and lots and lots of data I could benchmark this on. Unfortunately, I'm not going to be able to get to it until the end of the year at the earliest. I'm pursuing a PhD where the majority of my dissertation is based on my publications and I have my keystone publications that I'm currently working on.

I guess my request for this "issue" would be the following:
Generalize ward_cut_tree_balanced to cut_tree_balanced (I think that's what you mentioned the shortened version would be) and make it installable via PyPI (i.e. pip). In the meantime, it can be cited through zenodo to track publications. ANY alternative to dynamicCutTree would be awesome to have at my arsenal.

Discussion: What should the default value be for `max_cluster_size`?

I'm testing out your package and noticed there's no parameter for max_cluster_size. is there any way we can estimate this before and use that estimation as a default?

Here's the Iris dataset and it looks pretty good using max_cluster_size=50 but this is because I know there's 50 of each class.