interface to arbitrary data about ssm HOT 4 CLOSED

Maybe my problem is I don't understand the documentation ;-)

from ssm.

Hey Kam,

Glad you found your way to SSM! Real data examples are on the horizon, don't worry. In the meantime, my advice would be: take the examples for simulating data and fitting another model to it and strip out the simulating step. The interface is pretty simple.

hmm = HMM(num_states, data_dim, observations="gaussian")
lls = hmm.fit([data_array1, data_array2, ...], method="em")

where data_array1 is a time x data_dim array.

Hope that gets you by for now. Thanks for your patience while I beef up the docs!

Scott

from ssm.

Thanks I see how it works now, cheers. (Doing a comparison with my optimization-based method.)

from ssm.

Great, which part of SSM are you using in particular? If you come up with some new optimization-based fitting methods let me know -- it'd be great to a) learn about them, and b) incorporate them into the package!

from ssm.