get_model_data fails when trying to extract model data with monotonized features

[Krzys25]

Hello,

I encountered an issue when trying to use the get_model_data function to extract model data from an ExplainableBoostingClassifier instance with monotonized features.

Steps to Reproduce:

1. Fit an ExplainableBoostingClassifier instance.
2. Apply the monotonize method for one of the features.
3. Attempt to extract the model data using get_model_data.

Expected Behavior:
The function should return the model data successfully.

Actual Behavior:
The function fails with a TypeError in this line, and upon further investigation, I noticed that the entry in ebm.standard_deviations_ for the feature I applied monotonize to is being set to None.

Workaround:
Currently, I'm manually replacing the None value with a placeholder value, but I believe this should be handled gracefully by the library itself.

Please let me know if any further information is required or if there's a known solution to this problem.

Thanks in advance for your help.

Best,
Krzysztof

[HenrikSmith]

Hello,

I noticed a similar behaviour for the ExplainableBoostingRegressor for a model with monotinized features.

get_model_data seems to work fine as long as I do not modify the bins in any way. However, as soon as the bin boundaries move or the overall number of bins is changed, I keep on getting a ValueErrorValueError: operands could not be broadcast together with shapes (10,) (9,). The ebm.standard_deviations_ are returned as zeroes, however the dimension is always one smaller than it's supposed to be.

I can fix that by adding the missing zero to the respective array. However, I'm now facing other issues:

• When calculating a ebm.global_explanation(), get_model_data appears to have updated the scores, but not the bin_weights, so I'm getting a TypeError: Axis must be specified when shapes of a and weights differ from this line.
• During inference with ebm.predict_with_uncertainty(), it seems to be the other way round. I'm getting an IndexError: index 10 is out of bounds for axis 0 with size 10 from this line, because term_scores seems to lack some of the scores requested via bin_indexes.

Is there any way I can work around this?

Thank you very much in advance for your help!

[HenrikSmith]

Hello,

I inspected a little more and I may have found the root causes for the issues. For the variable modified via gamchanger:

• the respective sub-array of ebm.standard_deviations_ is reset to zero but appears to be one entry short (a quick fix may be: replace with array of zeros with correct length)
• the respective sub-array of ebm.bin_weights_ is not updated and hence still has the length (and probably also the content) of the original model (a quick fix may be: replace with array of ones with correct length)
• the respective sub-array of ebm.bagged_scores_ is not updated and hence still has the length and content of the original model (a quick fix may be: replace all bags with the updated ebm.term_scores_ sub-array)

I haven't tested this thoroughly, yet, but maybe this is still of help to you guys or anyone else?

Thank you very much for your work and best regards!