Return SignedRankTest objects, allow to use the built-in plot() methods in baycomp

[Daniel300000]

I added another attribute 'sample_matrix' for the SignedRankTest objects from the package 'baycomp'.

One can then use them to draw some plots, e.g.:

# do Bayesian testing with autorank
result_bayesian = autorank(data, alpha=0.05, verbose=False, approach='bayesian')

# plot the simplex with baycomp
fig = result_bayesian.sample_matrix.loc['Sample1', 'Sample2'].plot(names=('Sample1', 'Sample2'))

[sherbold]

Thanks for this contribution @Daniel300000!

Great idea to expose this.

The change itself looks good. However, before accepting two things need to be done:

• The change breaks the test case test_plot_stats_invalid because the number of fields in RankResult changed. This is easily fixed with another None.
• The additional parameter needs to be documented in the docstring of the autorank method.

[Daniel300000]

Thanks for merging the pull request!

It seems that I was a little bit too fast with my commit, sorry. Should we also update the description? For example:

    sample_matrix (DataFrame):
        Matrix with SignedRankTest objects from package baycomp. Can be used to do further analysis, e.g. to generate
        plots using the built-in plot() method of baycomp. For a detailed description of methods and parameters, see
        the documentation of baycomp.
        _(New in Version 1.2.0)_

What would be the new version on PyPI? Version 1.2.0?

[sherbold]

Yes, next will be 1.2.0. If you want to update the documentation, add a link directly to the relevant part of baycomp.