PR Summary

This PR enhances the diagnostic functions in the test suite by ensuring that only numeric and categorical feature columns are processed. In the OverfitDiagnosis function, the code now concatenates numeric and categorical feature columns and raises an error if no valid columns are found. Similarly, the WeakspotsDiagnosis function has been modified to filter the feature columns based on the numeric and categorical columns available in the dataset, and to raise a ValueError if none remain. These changes help avoid issues when non-relevant feature columns are present and improve the accuracy of the analysis conducted by these functions.

Test Suggestions

• Write tests to ensure that OverfitDiagnosis raises a ValueError when datasets[0] contains no numeric or categorical columns.
• Verify that WeakspotsDiagnosis filters out non-numeric and non-categorical columns correctly and raises the appropriate error when the filtered list is empty.
• Test the functions with a mix of valid and invalid columns to confirm that only the expected features are processed.
• Ensure that the diagnostic functions continue to produce the expected outputs (e.g., figures, headers) when valid data is provided.