Kevin Sheppard

We can't reproduce on CI: https://dev.azure.com/statsmodels/statsmodels-testing/_build/results?buildId=5869&view=logs&j=ff02c0bf-a6da-574b-cd48-e6a8a9d1f05d Configuration: ``` build 1.2.1 certifi 2024.7.4 colorama 0.4.6 contourpy 1.2.1 coverage 7.6.1 cycler 0.12.1 Cython 3.0.11 execnet 2.1.1 flake8 7.1.1 fonttools 4.53.1 iniconfig 2.0.0...

There seems to be a decent bug here that isn't easy to fix. It appears in the 0.14.2 code even if built against SciPy 1.14.1.

The source of the issue has been found. It is due to some changes in the arch used by the GitHub MaxOS runners, which were previously x86_64 and have moved...

Appears to be fixed now. Will need a release note and then likely to release early next week. https://github.com/MacPython/statsmodels-wheels/actions/runs/10848683287

Wheels are on PyPi (excluding a couple that are slow building). Closing. Please let us know if there continues to be a problem.

Citation of what? Or a method to cite statsmdoels?

https://www.statsmodels.org/stable/index.html#citation

There are related test failures. Need to figure out what is going on before we can merge. Possibly these are swapped somewhere else in the code.

Can start literally anywhere. statsmodels will be pretty hard to types in general since many parts poke attributes onto classes. But simple functions are a good place to start.

The biggest challenge with statsmodels is there are too many attributes that are totally missing when not used. These should all be replaced with code like `self.attr = None` and...