Package for performing Bayesian inference with insufficient and robust statistics
Project description
Insufficient-Gibbs-Sampling
InsufficientGibbs
is a package that enables users to sample from the posterior parameters when only certain robust statistics of the data are accessible. The paper that comprehensively describes the theory behind these methods can be found on arXiv (https://arxiv.org/abs/2307.14973). Additionally, the code responsible for generating all the paper figures can be located in the figures
folder.
Robust Gibbs package
We propose here three main functions named Gibbs_med_MAD
, Gibbs_med_IQR
and Gibbs_Quantile
to cover the case when we observe the pairs (median, MAD) or (median, IQR) or a sequence of quantiles.
Install
Install latest release via pip
pip install InsufficientGibbs
For latest development version clone the repository and install via pip
git clone https://github.com/antoineluciano/Insufficient-Gibbs-Sampling
cd InsufficientGibbs
pip install .
Available models/likelihood
- Normal distribution (
distribution="normal"
) - Cauchy distribution (
distribution="cauchy"
) - Weibull distribution (
distribution="weibull"
) - Translated distribution (
distribution="translated_weibull"
)
Available location priors
- Normal (
par_loc="normal"
) - Cauchy (
par_loc="cauchy"
) - Gamma (
par_loc="gamma"
)
Available scale priors
- Gamma (
par_loc="gamma"
) - Jeffreys (
par_loc="jeffreys"
)
Available shape priors
- Gamma (
par_loc="gamma"
)
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for InsufficientGibbs-1.0.2-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 4711ea4dfee0b51f727c3d1efd2deca77f3d78b40631d1a6f56e58e4890057f0 |
|
MD5 | 68cb9ee4efd05abb75f0abbf59a568fd |
|
BLAKE2b-256 | 0d64cffc6593aa805e9e0df3b35f2b8bea47ce8d847fef39700d22b21465193f |