Description
NOTE: We are just about to have a major refactor of the codebase, so this should be done after #381 is merged.
It might be nice to make the api docs (auto build from the docstrings) more engaging by adding some plots. As far as I can tell, it is possible to take the code that we have in the doctests right now and also generate a plot that is rendered in the docs.
This could be relevant for the experiment classes in particular. I don't see any value in adding to the model classes for example.
Here's an example of current docs:
If we add the plot method into that code snippet, then we can get that plot to render in the built docs.
I believe (though have not verified) that we can do this as follows:
.. plot::
:context: close-figs
:include-source: True
:format: doctest
>>> import causalpy as cp
>>> df = cp.load_data("did")
>>> seed = 42
>>> result = cp.pymc_experiments.DifferenceInDifferences(
... df,
... formula="y ~ 1 + group*post_treatment",
... time_variable_name="t",
... group_variable_name="group",
... model=cp.pymc_models.LinearRegression(
... sample_kwargs={
... "target_accept": 0.95,
... "random_seed": seed,
... "progressbar": False,
... }
... )
... )
>>> result.plot()If that works as I imagine, then this will:
- work as a doctest
- render the code in the docs
- include the rendered plot in the docs.
I have not verified that this works, but I imagine there is a way!
Notes
- We might need to call
plt.show()at the end. - At this point I have no idea if the doctest work would be taken over by sphinx, because there is an extension
sphinx.ext.doctest. Would need to look into this. - We'd also need to add
sphinx.ext.plot_directiveto the list of extensions inconf.py.