Skip to content

Latest commit

 

History

History
20 lines (12 loc) · 641 Bytes

NEWS.md

File metadata and controls

20 lines (12 loc) · 641 Bytes

Bayesian-HMM 0.0.3

  • README expanded to include more usage (plotting of posteriors from mcmc results)
  • Multiprocessing pool now closed after parallel chain resampling
  • MCMC steps reordered
  • Latent states only updated at the start of each MCMC step, not within the count update step
  • Improve code hygeine, moving non-class code to separate module
  • Store aggregate None state within the parameter dictionaries

Bayesian-HMM 0.0.2

  • All HDPHMM parameters and parameter priors now stored in dict objects HDPHMM.parameters and HDPHMM.priors
  • HDPHMM.mcmc now returns a dict object

Bayesian-HMM 0.0.1

  • Initial release