Assign missing class to a bin & BinningProcess with user splits about optbinning HOT 2 CLOSED

Hi @billy-odera,

1. It is not possible to combine the missing bin with other bins. A missing bin is automatically generated if the dataset includes missing data. However, you can change the WoE value from -0.418329 to -0.0854013 when transforming using transform function in OptimalGrouping: http://gnpalencia.org/optbinning/binning_binary.html#optbinning.OptimalBinning.transform; use metric_missing= -0.0854013.

2. You need to pass a dictionary with particular options via parameter binning_fit_params. Example:

binning_fit_params = {
    "variable": {"dtype": "numerical", "solver": "mip", "user_splits": [2,6,14,24,34]}
}

binning_process = BinningProcess(variable_names, categorical_variables=categorical_variables, 
                                 selection_criteria=selection_criteria, binning_fit_params=binning_fit_params)

from optbinning.

Thanks for the feedback

from optbinning.