end_level has huge number

[deter3]

I am using Debacl for images features clustering , and the tree results has huge end-level , is there something wrong with my implementation ?

tree = dcl.construct_tree(featlist, k=2,verbose=True)

[bpkent]

Can you post a little more of the code that led to this state? Two things come to mind that could cause this kind of numerical instability: co-located points, where the k-neighbor radius might be 0, and using a high dimension to try to compute an exact density estimate. If it's the former case, you could try adding a very small amount of jitter to the points. If it's the latter, try using a dimension of 1 to get the density estimate order rather than precise estimates - it's all that DeBaCl really needs to compute the tree structure.

If you can post more code, that would be great - without it, it's really hard to know what's going on.

Thanks!
Brian