PPD-overview: Plots of posterior or prior predictive distributions

In bayesplot: Plotting for Bayesian Models

Plots of posterior or prior predictive distributions

Description

The bayesplot PPD module provides various plotting functions for creating graphical displays of simulated data from the posterior or prior predictive distribution. These plots are essentially the same as the corresponding PPC plots but without showing any observed data. Because these are not "checks" compared to data we use PPD (for prior/posterior predictive distribution) instead of PPC (for prior/posterior predictive check).

PPD plotting functions

The functions for plotting prior and posterior predictive distributions without observed data each have the prefix ppd_ and all have a required argument ypred (a matrix of predictions). The plots are organized into several categories, each with its own documentation:

• PPD-distributions: Histograms, kernel density estimates, boxplots, and other plots of multiple simulated datasets (rows) in ypred. These are the same as the plots in PPC-distributions but without including any comparison to y.

• PPD-intervals: Interval estimates for each predicted observations (columns) in ypred. The x-axis variable can be optionally specified by the user (e.g. to plot against against a predictor variable or over time).These are the same as the plots in PPC-intervals but without including any comparison to y.

• PPD-test-statistics: The distribution of a statistic, or a pair of statistics, over the simulated datasets (rows) in ypred. These are the same as the plots in PPC-test-statistics but without including any comparison to y.

References

Gabry, J. , Simpson, D. , Vehtari, A. , Betancourt, M. and Gelman, A. (2019), Visualization in Bayesian workflow. J. R. Stat. Soc. A, 182: 389-402. doi:10.1111/rssa.12378. (journal version, arXiv preprint, code on GitHub)

See Also

Other PPDs: PPD-distributions, PPD-intervals, PPD-test-statistics

bayesplot documentation built on May 29, 2024, 2:27 a.m.