FAQ & Troubleshooting
Common Issues
ImportError: No module named 'imblearn'
Cause: You are trying to use strategy='smote' but imbalanced-learn
is not installed.
Solution: Install the library with the extra dependency:
pip install ml-audit[balance]
# OR
pip install imbalanced-learn
Where are my visualization files?
By default, ml-audit organizes output into project folders:
- JSON logs go to
audit_trails/ - HTML files go to
visualizations/
If you prefer a custom location, you can pass an absolute path to export_audit_trail.
Can I track custom pandas functions?
Yes! Use the track_pandas method or the specific GenericPandasOp if you are
extending the class.
# Track a dropna operation
auditor.track_pandas("dropna", subset=['col1'])Performance Tips
- For very large datasets, consider disabling
verify_reproducibility()in production loops as it re-runs the entire pipeline. - Use list inputs (e.g.
impute(['a','b'])) instead of calling impute twice for better readability and slightly better performance.