[SC-7456] intro_for_model_developers.ipynb throws SHAPGlobalImportance.py error

[juanmleng]

Internal Notes for Reviewers

When running the 102-start_development_process.ipynb notebook the SHAPGlobalImportance test fails with:

ERROR(validmind.vm_models.test_suite.test): Failed to run test 'validmind.model_validation.sklearn.SHAPGlobalImportance': (TypeError) loop of ufunc does not support argument 0 of type float which has no callable rint method

There seems to be an issue with the Logistic Regression setup in the notebook causing some SHAP values to have object data type instead of floats, which creates problems with the SHAP plots. The fix here is to convert these values to proper float arrays.

External Release Notes

[github-actions]

PR Summary

This pull request enhances the logic for selecting SHAP values in the select_shap_values function within the validmind/tests/model_validation/sklearn/SHAPGlobalImportance.py file. The changes include:

• Improved handling of SHAP values for both regression and classification tasks.
• For regression tasks, the SHAP values are returned as-is if they are not a list.
• For binary classification, the function defaults to using SHAP values for class 1 if no specific class is specified.
• For multi-class classification, the function now checks if the class_of_interest is valid and raises a ValueError if it is not.
• Added type conversion to ensure the selected SHAP values are returned as a float array.

These changes aim to improve the robustness and clarity of the SHAP value selection process.

Test Suggestions

• Test with regression SHAP values to ensure they are returned as-is.
• Test binary classification SHAP values without specifying class_of_interest to ensure class 1 values are returned.
• Test multi-class classification with valid class_of_interest to ensure correct SHAP values are selected.
• Test multi-class classification with invalid class_of_interest to ensure a ValueError is raised.
• Test the type conversion to ensure the returned SHAP values are of type float array.