exPrior

Recent papers have made the case that statistics in subsurface hydrology is still falling short on available software tools (Rubin, 2018) and that Bayesian inference is the most relevant framework for subsurface statistics (Hesse, 2019). The exPrior package, providing tools for Bayesian inference, is therefore a timely addition to the R ecosystem of statistical tools.

The aim of this package is to provide practitioners of Bayesian statistics a tool to derive priors for Bayesian inference in subsurface hydrology. Priors summarize hydro-geological knowledge from studies at similar sites (Cucchi, 2019). The main features of the package are:

• generate prior distributions based on external data only, therefore meeting the likelihood principle,
• account for possible autocorrelation in these data,
• account for available soft data, say, in the form of expert information on bounds and moments,
• comes bundled with the largest open-source database on hydro-geological parameters, etc.

Installation

exPrior is not yet on CRAN. You can install the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("GeoStat-Bayesian/exPrior")

Example

Examples on how to use exPrior can be found in the vignettes in the /vignettes folder.

Documentation

A publication detailing the algorithm can be found at Cucchi (2019). A publication detailing the package and how to use it is in preparation.