reno.viz.compare_posterior

reno.viz.compare_posterior(varname, traces, prior_trace=None, smoothing=0.1, ax=None, legend=True, title=None, per_dim=None, **figargs)

Plot the sampled distribution densities for the specified variable from each of the passed traces + prior trace.

Parameters:

• varname (str) – The name of the variable in the xr.Datasets to plot

• traces (list[az.InferenceData | xr.Dataset] | dict[str, az.InferenceData | xr.Dataset) – A list or dictionary of traces or dataset to plot the variable from. Passed traces will plot from the posterior, pass the specific dataset if you need the priors (pymc_trace.prior). If a dictionary is used, the legend will use the specified keys.

• prior_trace (az.InferenceData | xr.Dataset) – A trace or dataset to plot with the ‘prior’ key.

• smoothing (float) – What degree of smoothing to apply to the density plot. Lower = more bumpy.

• ax – Optionally pass an axis if one already exists, otherwise this function will create a new one, using any additional figargs passed.

• legend (bool) – Whether to render a legend or not. If True and traces is a dictionary, the keys will be used as the legend labels.

• title (str) – Optional title to set on the axis.

• per_dim (str) – TODO

• **figargs – Parameters to pass to plt.subplots(**figargs) if no axis passed in.