Closed hang-wu closed 9 years ago
(I forgot about this one!)
I'm glad you like the code!
The observation-specific special model DiagGaussGMMHSMMPossibleChangepointsSeparateTrans would only need to be implemented if there were some specialized low-level code for that combination of observations and model features. (In the past I've wanted extra-fast inference for some specialized models, hence those bespoke classes.) You probably just want an HSMMPossibleChangepointsSeparateTrans as your model (if it exists, it should just be a combination of mixins), and pass in a list of DiagonalGaussian instances as the obs_distns.
The plotting code is not very general or polished and pretty much only works for the examples and animations I've made, which usually just use Gaussian emissions. You'll have to write your own plotting code for the most part.
Hi Matthew,
Thanks a lot for your awesome codes. However, when I try to run examples/ hsmm-possiblechangepoints.py, I encountered some issues.
I am wondering why this happened. I am using a MAC OSX, with python version Python 2.7.8 (default, Oct 18 2014, 19:36:29) [GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.54)] on darwin
Thanks again. Best