mcmc-samplers: Combinators for MCMC sampling

[ bsd3, library, machine-learning, math, numeric, statistics ] [ Propose Tags ]

A library of combinators to build transition kernels, proposal distributions, target distributions, and stream operations for MCMC sampling.


[Skip to Readme]

Downloads

Maintainer's Corner

Package maintainers

For package maintainers and hackage trustees

Candidates

  • No Candidates
Versions [RSS] 0.1.0.0, 0.1.1.0, 0.1.1.1
Dependencies base (>=4.6 && <5), containers (>=0.5 && <0.6), hakaru (==0.1.4), hmatrix (>=0.15), mwc-random (>=0.13 && <0.14), primitive (>=0.5 && <0.6), statistics (>=0.11) [details]
License BSD-3-Clause
Author Praveen Narayanan
Maintainer pravnar@indiana.edu
Category Machine Learning, Math, Numeric, Statistics
Uploaded by pravnar at 2014-11-10T17:11:22Z
Distributions
Reverse Dependencies 1 direct, 0 indirect [details]
Downloads 2266 total (9 in the last 30 days)
Rating 2.0 (votes: 1) [estimated by Bayesian average]
Your Rating
  • λ
  • λ
  • λ
Status Docs uploaded by user
Build status unknown [no reports yet]

Readme for mcmc-samplers-0.1.1.1

[back to package description]

Samplers

Here lies a library of combinators for MCMC kernels and proposals

  • The relevant modules are Kernels, Distributions, and Actions
  • See Tests.hs for some examples on how this library can be used
  • Needs the hmatrix package
    • Might need to do cabal install hmatrix
On Gibbs.hs
  • The current implementation is for a Naive Bayes model
  • TODO:
    • Use an existing, "real" dataset instead of randomly generating sentences
    • See which words appear most frequently for each label/class
    • Average over all theta estimates and return top 10 and bottom 10 words according to these averages
    • Implement burn-in and lag (to decrease autocorrelation)