Open EtienneT opened 5 years ago
Hello, I will release a small package for first-order and second-order ALE plots in a few day, as I have the same needs on this method.
Excellent, update this feature request when you do!
Hello, I started a public repository of ALEPython, which I work on to explain models predictions with ALE plots. Beware, as project is still on dev branch, you might encounter some bugs or issues. I pushed a first release since it is sufficient for my needs at the moment, but I will continue to bring new updates on this package.
So maybe it could be useful for you, but for now, there are no guarantees and potentially mistakes.
Accumulated local effects describe how features influence the prediction of a machine learning model on average. ALE plots are a faster and unbiased alternative to partial dependence plots (PDPs).
https://christophm.github.io/interpretable-ml-book/ale.html
There's R modules supporting them but no Python module.