BayesianStanRegression

This repository provides the stan files used in the article:

Stafford, P.J. (2018), Continuous integration of data into ground-motion models using Bayesian updating. Journal of Seismology, under review

The repository contains just two *.stan files:

• PJSstanModel_reES_AS.stan is the file for performing regression analysis on a traditional dataset. The data passed to this code contains observations from a number of earthquake events, and the code specifies informative univariate priors for the fixed effects. Note that the priors are defined for the peak ground acceleration as used in the above article.
• PJSstanModel_reES_update_AS.stan is the file used for performing Bayesian updating. The data passed to this code contains just observations from a single new event. The priors are specified as a multivariate prior for the fixed effects and univariate priors for the variance components and random effects.

Note that in both cases the file name contains the string _reES_ to indicate that random effects for Event and Station are considered in these scripts.

To actually run the regressions, ground-motion data must be prepared and passed through calls to the Stan program. While the rstan interface was used in this particular study, it is possible to prepare the data and interact with Stan in a number of ways. See http://mc-stan.org for more details.