pymc3.traceplot() method documentation error

[AlonSpitz]

The documentation on pymc3.traceplot() documentation says that the 'compact' argument is 'False' by default, but it seems it is in fact 'True'.

Example (from the Getting started with PyMC3 section):

import pymc3 as pm
basic_model = pm.Model()

with basic_model:

`# Priors for unknows model parameters`
`alpha = pm.Normal('alpha', mu=0, sigma=10)`
`beta  = pm.Normal('beta', mu=0, sigma=10, shape=2)`
`sigma = pm.HalfNormal('sigma', sigma=1)`

`# Expected value of outcome`
`mu = alpha + beta[0]*X1 + beta[1]*X2`

`# Likelihood (samplign distribution of observations`
`Y = pm.Normal('Y_obs', mu=mu, sigma=sigma, observed=Y)`

with basic_model:

`trace = pm.sample(500)`

pm.traceplot(trace);

Above code will show the beta parameter (a vector of shape 2) in a single plot, instead of in two separate plots, though according to the documentation we should expect to get the same result as:

pm.traceplot(trace, compact=False)

Versions and main components

• PyMC3 Version: 3.8
• Theano Version: 1.0.4
• Python Version: 3.7.7
• Operating system: macOS Catalina, version 10.15.4
• How did you install PyMC3: conda

[aloctavodia]

Hi @AlonSpitz tanks for reporting this issue. @ColCarroll I guess the problem is that the docstring is read from az.plot_trace, but PyMC3 is changing that default.

[ricardoV94]

This can be closed after #4397 ?

[OriolAbril]

This can be closed already I think, PyMC3 docs already have only links to ArviZ for plots and stats and the docstring is correct on ArviZ's side, and moreover, the latest ArviZ release has compact=True as a default so there are no differences between plot_trace and pm.traceplot (which will also be removed after #4397)