Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

OptimalBinning2D not working with numerical and categorical predictor

See original GitHub issue

Hello, I enjoy your package very much as it makes feature engineering, in some cases, much simpler. I suspect it is a bug, and you probably didn’t think of 2D binning using categorical predictors. I cannot provide you the data, but I hope it is clear.

from optbinning import OptimalBinning2D

col1, col2 = "NUMERICAL_32", "CATEGORICAL_10"
col_target = "TARGET"
optb = OptimalBinning2D(name_x=col1, name_y=col2, dtype_x="numerical", dtype_y="categorical", solver="cp")
x = data[col1].values
y = data[col2].values
z = data[col_target]
optb.fit(x, y, z)
optb.binning_table.build(show_bin_xy=True)
ifeat_woe = optb.transform(x, y, metric="woe")

I am getting following error.

Traceback (most recent call last):
  File "/Users/danielherman/Documents/projects/ifeat/ifeat_optbinning.py", line 43, in <module>
    ifeat_woe = optb.transform(x, y, metric="woe")
  File "/Users/danielherman/Documents/projects/ifeat/env/lib/python3.9/site-packages/optbinning/binning/multidimensional/binning_2d.py", line 551, in transform
    return transform_binary_target(
  File "/Users/danielherman/Documents/projects/ifeat/env/lib/python3.9/site-packages/optbinning/binning/multidimensional/transformations_2d.py", line 123, in transform_binary_target
    bins_str = bin_xy_str_format(dtype_x, dtype_y, splits_x, splits_y,
  File "/Users/danielherman/Documents/projects/ifeat/env/lib/python3.9/site-packages/optbinning/binning/multidimensional/binning_statistics_2d.py", line 72, in bin_xy_str_format
    _by = _bin_fmt_categorical(by, categories_y)
  File "/Users/danielherman/Documents/projects/ifeat/env/lib/python3.9/site-packages/optbinning/binning/multidimensional/binning_statistics_2d.py", line 42, in _bin_fmt_categorical
    n_categories = len(categories)
TypeError: object of type 'NoneType' has no len()

I am also using the last available version 0.15.1.

Will you add the support of 2D binning with numerical+categorical predictors and categorical+categorical predictors?

Issue Analytics

State:
Created a year ago
Comments:6 (3 by maintainers)

Top GitHub Comments

1reaction

detrincommented, Oct 3, 2022

It didn’t solve the issue, I will try to build reproducible code today.

0reactions

guillermo-navas-palenciacommented, Oct 4, 2022

Perfect. The code must be improved, though.

Top Results From Across the Web

guillermo-navas-palencia/optbinning: Optimal binning - GitHub

OptBinning is a library written in Python implementing a rigorous and flexible mathematical programming formulation to solve the optimal binning problem for ...

Tutorial: Binning process with sklearn Pipeline — optbinning ...

Instantiate a BinningProcess object class with variable names and the list of numerical variables to be considered categorical. Create pipeline object by ...

Using numerical and categorical variables together

We will separate categorical and numerical variables using their data types ... Besides, we use handle_unknown="ignore" to solve the potential issues due to ......

Optimal binning methods for categorical variables

I'm running a multinomial logit to predict the outcome of a categoric response variable ...

Dealing with Categorical Variables in Machine Learning

Real-world data issues and machine learning model. Image by author ... Data set comprises of both numerical and categorical variables.