Christian Schwantes
Christian Schwantes
This happens when Cluster.py has already written the Assignments.h5 file. When you cluster with a stride of 1 it uses all of your data, which means you do not need...
This has caused quite a few people coming to me with a failed run. It's actually more than just a lack of characters, because it's a feature of argparse, since...
thanks for checking this out, I'm going to make these changes (hopefully later today)
In practice, I've found that the hybrid k-medoids that msmb implements doesn't change things drastically. If you wanted to do k-means, however, you could probably gain a lot. I don't...
We don't have k-means in msmbuilder. I think I tried it once on my own with scikit-learn though On Thu, Feb 6, 2014 at 11:12 AM, kyleabeauchamp [email protected]: > Thanks....
Does it work? I didn't know that On Thu, Feb 6, 2014 at 11:17 AM, kyleabeauchamp [email protected]: > We do have a commented-out KMeans class in clustering.py... > > ##...
Yea, though it's not a streaming clusterer, but it can load / project things streaming so you can gain a lot On Thu, Feb 6, 2014 at 12:44 PM, kyleabeauchamp...
Looks cool. What is the 'thin' kwarg supposed to do? Will this only work for dense matrices? Do you think this is going to be possible for a 10,000 state...
How slow do you mean? Waiting a day, probably isn't a big deal. Can you make a parallel version?
I spoke too soon, it looks like it's just stochastic