qq_postpred: Quantile-quantile plot from posterior predictive distribution

Description

Produces a quantile-quantile plot, calculated from the quantiles of a vector of data (most likely a time series), with respect to the matrix of associated posterior predictive distributions.

While not an omnibus posterior predictive check, this plot can be useful for detecting an overparameterized model, or else improper specification of observation error. Like a traditional Q-Q plot, a well-specified model will have points that lie close to the x=y line. In the case of this function, an overparametrized model will typically produce a plot with a much shallower slope, possibly with many associated posterior predictive quantiles close to 0.5.

It should be noted that this function will only produce meaningful results with a vector of data, as opposed to a single value.

The posterior predictive distribution can be specified in two possible ways: either a single output object from jagsUI with an associated parameter name, or as a matrix or data.frame of posterior samples.

Usage

qq_postpred(ypp, y, p = NULL, add = FALSE, ...)

Value

NULL

Arguments

ypp: Either a matrix or data.frame of posterior samples, or an output object returned from jagsUI and a supplied parameter name
y: The associated data vector
p: A character name, if a jagsUI object is passed to ypp
add: Whether to add the plot to an existing plot. Defaults to FALSE.
...: Optional plotting arguments

Author

Matt Tyers

Examples

Run this code

# first, a quick look at the example data...
str(SS_data)
str(SS_out$sims.list$ypp)

# plotting the example posterior predictive distribution with the data
# points overlayed.  Note the overdispersion in the posterior predictive.
caterpillar(SS_out, p="ypp")
points(SS_data$y)

# using a jagsUI object as ypp input
qq_postpred(ypp=SS_out, p="ypp", y=SS_data$y)

# using a matrix as ypp input
qq_postpred(ypp=SS_out$sims.list$ypp, y=SS_data$y)