defaultPriorList: Create list of prior parameters for 'mcmc.aggregate(...)'...
In NMML/agTrend: Estimate Linear Trends for Aggregated Abundance Data

Description Usage Arguments Details Value

This function takes user input to create the list of prior paremters necessary for Bayesian inference of aggregated trends. The prior parameter list is a necessary argument for the mcmc.aggregate function.

1	defaultPriorList(trend.limit = NULL, model.data, gamma.mean, gamma.prec)

`trend.limit`	Bounds on the individual site trends.
`model.data`	A data set which provides the individual site models used in `mcmc.aggregate`.
`gamma.mean`	Prior mean of the (multivariate) normal prior distribution for gamma
`gamma.prec`	Precision (inverse of variance) of the gamma prior distribution

Using the site model data set model.data and a soft bound on the individual site trends (trend.limit) this function creates a sensible default for the list of prior parameters necessary for Bayesian inference of aggregated trends. The value of trend.limit should be in terms of of 20 percent growth at each individual site). The lower bound is calculated as 1/(1+trend.limit). This is the ower bound that is symmetric about zero on the log-scale. For trend.limit=0.2, the lower bound is approximately -0.17.

Specifically, the prior distribution for the linear trend parameters (beta) in log abundance at each site is set to a bivariate normal with mean m.i=c(0,0) and precision matrix Q.i=diag(c(0,q)), where q is chosen such that Pr[1/(1+trend.limit) < exp(beta[2])-1 < trend.limit] = 0.95.

A named list with entries for each of the parameters for which prior specification is necessary.

NMML/agTrend documentation built on May 7, 2019, 6:02 p.m.