Adding quantile dot plots to PPC distributions

[behramulukir]

Right now, bayesplot offers various functions for posterior predictive check such as ppc_dens, ppc_hist, ppc_freqpoly, ppc_boxplot, and a couple of others. One of the plots that bayesplot currently doesn't have a function for is quantile dot plots.

In their preprint, Säilynoja et al compared KDE plots, histograms, and quantile dot plots, finding that "compared to KDEs, quantile dot plots have the added benefit of allowing for fast visual probability estimation in the tails of the distribution" and "in case of discontinuities and outliers, quantile dot plots also offer a better visualization than kernel density plots and histograms."

Based on that, I suggest adding quantile dot plots. It has certain benefits that are not currently offered by the package and gives an alternative visualisation method to users.

For this implementation, I am thinking of using geom_dotplot function, which should be fit for the task.

[jgabry]

Sounds like a good idea! My guess is that you can probably use most of the same internal code that we use for the histograms and KDE plots but then pass it to geom_dotplot instead.

[behramulukir]

Thanks, I'll work on this now.