Closed josp70 closed 3 months ago
Hi, @josp70.
For this case, you can use the parameters user_splits
and user_splits_fixed
. There is an example in this tutorial: https://gnpalencia.org/optbinning/tutorials/tutorial_binary.html#User-defined-split-points.
If the solution turns out to be infeasible (check information()
), disallow all constraints. If you are willing to provide a reproducible example, I can have a look.
@guillermo-navas-palencia that did it:
user_splits = np.array([[x] for x in data[variable].unique()], dtype=object)
user_splits_fixed = [True for x in data[variable].unique()]
data is my dataframe and variable, the name of the categorical variable to plot
Thanks, great package!!!
I have a dataset with a binary target an a categorical variable taking 3 different values. I want to report the statistics for the binning table with the number of bins equal to 3 (without optimization). For that I'm setting the parameter min_n_bins=3 but the result is a table with a single bin. If I set min_n_bins to the default value None then I get a result with 2 bins.
Is there a way to build a binning with each category as a bin (no category grouping)?