ajpierce / bnpcr_toolbox

The original Bayesian Nonparametric Covariance Regression Toolbox by Emily Fox and David Dunson

Matlab Software for Bayesian Nonparametric Covariance Regression

Copyright (C) 2011, Emily B. Fox (fox[at]stat[dot]duke[dot]edu)

This software package includes Matlab scripts that implement the Gibbs sampling algorithm for the model described in: Bayesian Nonparametric Covariance Regression E. B. Fox and D. B. Dunson arXiv:1101.2017, January 2011 (revised February 2011). http://arxiv.org/abs/1101.2017 Please cite this paper in any publications using the HDP-AR-HMM or HDP-SLDS package.

Package Organization and Documentation

Summary of BNP Covariance Regression package contents:

• BNP_covreg.m: Main inference script for running the sampler without missing data or imputing the missing values.

• BNP_covreg_varinds.m: Main inference script for running the sampler with analytic marginalization of missing values.

• runstuff_BNPcovreg.m and runstuff_var_inds_flu.m: See below for explanation

• /utilities: Script SIMplots.m contains an example of how to process and visualize the results from the stored samples.

Setup and Usage Examples

For an example of running the sampler without missing data or imputing the missing values, see runstuff_BNPcovreg.m. This script examines a synthetic data example.

For an example of running the sampler with analytic marginalization of missing values, see runstuff_varinds_flu.m. This script examines the Google Flu Trends dataset.

Copyright & License

Copyright (C) 2011, Emily B. Fox.

http://web.mit.edu/ebfox/www/

Permission is granted for anyone to copy, use, or modify these programs and accompanying documents for purposes of research or education, provided this copyright notice is retained, and note is made of any changes that have been made.

These programs and documents are distributed without any warranty, express or implied. As the programs were written for research purposes only, they have not been tested to the degree that would be advisable in any important application. All use of these programs is entirely at the user's own risk.