plot_density function doesn't work for predictive prior unless credible interval is specified as 1

Describe the bug When attempt to plot prior predictive distributions using plot_density function, plots show nothin unless credible interval is defined as 1

To Reproduce az.plot_density(prior,credible_interval=1)

az.plot_density(prior)

Expected behavior Expected that the plot should only contain the 0.89 credible interval, instead of blank plots

Additional context arviz 0.7.0 pymc3 3.8 matplotlib 3.2.1

also using %matplotlib inline

[aloctavodia]

Thanks for the example. Now I can confirm the issue. I will work on this and keep you posted. In the meantime you can do this:

data = az.from_pymc3(posterior_1, prior=prior, posterior_predictive=posterior_pred_1)

data will be an ArviZ’s InferenceData object

you can then pass it to arviz functions, for example.

az.summary(data)

or az.plot_density(data.prior) or equivalently az.plot_density(data, group=prior) the default group is posterior thus az.plot_density(data) will give you a plot of the posterior.

One tip, you can globally set a few arviz parameters using az.rcParams, so for example you can add az.rcParams["stats.credible_interval"] = 0.89 at the beginning, and then all plots and summaries will use this value instead of the default one (0.94).

My guess about statistical rethinking is because the credible interval of .89 is used by the book even when the most common default value is 0.95. ArviZ and Statistical rethinking both uses non common values to remind user of the arbitrary nature of any default value.

Your report was very good. You clearly stated the problem and the excepted outcome. A reproducible (simple) example always helps.

[aloctavodia]

Closing as the recommended workflow is to generate the idata object.