arviz.plot_compare(comp_df, insample_dev=True, plot_standard_error=True, plot_ic_diff=True, order_by_rank=True, figsize=None, textsize=None, labeller=None, plot_kwargs=None, ax=None, backend=None, backend_kwargs=None, show=None)[source]

Summary plot for model comparison.

This plot is in the style of the one used in the book Statistical Rethinking (Chapter 6) by Richard McElreath.


Result of the method

insample_devbool, optional

Plot in-sample deviance, that is the value of the information criteria without the penalization given by the effective number of parameters (pIC). Defaults to True

plot_standard_errorbool, optional

Plot the standard error of the information criteria estimate. Defaults to True

plot_ic_diffbool, optional

Plot standard error of the difference in information criteria between each model and the top-ranked model. Defaults to True


If True (default) ensure the best model is used as reference.

figsizetuple, optional

If None, size is (6, num of models) inches

textsize: float

Text size scaling factor for labels, titles and lines. If None it will be autoscaled based on figsize.

labellerlabeller instance, optional

Class providing the method model_name_to_str to generate the labels in the plot. Read the Label guide for more details and usage examples.

plot_kwargsdict, optional

Optional arguments for plot elements. Currently accepts ‘color_ic’, ‘marker_ic’, ‘color_insample_dev’, ‘marker_insample_dev’, ‘color_dse’, ‘marker_dse’, ‘ls_min_ic’ ‘color_ls_min_ic’, ‘fontsize’

ax: axes, optional

Matplotlib axes or bokeh figures.

backend: str, optional

Select plotting backend {“matplotlib”,”bokeh”}. Default “matplotlib”.

backend_kwargs: bool, optional

These are kwargs specific to the backend being used, passed to matplotlib.pyplot.subplots() or bokeh.plotting.figure().

showbool, optional

Call backend show function.

axesmatplotlib axes or bokeh figures

See also


Plot pointwise elpd differences between two or more models.


Compare models based on PSIS-LOO loo or WAIC waic cross-validation.


Compute Pareto-smoothed importance sampling leave-one-out cross-validation (PSIS-LOO-CV).


Compute the widely applicable information criterion.


Defaults to comparing Leave-one-out (psis-loo) if present in comp_df column, otherwise compares Widely Applicable Information Criterion (WAIC)


Show default compare plot

>>> import arviz as az
>>> model_compare ={'Centered 8 schools': az.load_arviz_data('centered_eight'),
>>>                  'Non-centered 8 schools': az.load_arviz_data('non_centered_eight')})
>>> az.plot_compare(model_compare)

Plot standard error and information criteria difference only

>>> az.plot_compare(model_compare, insample_dev=False)