Versioning Data

Created: 2023-02-13 11:50
#note

Managing data with a data versioning system provides the ability to change data continuously while ensuring consistent training results and collaboration.

Data versioning is a similar concept as code versioning -> it helps track of changes.
Versions can be at a feature, record, or dataset level.
Benefits:

• traceability;
• reproducibility;
• change log capture alternative;
• rollback to last know good state;
• multiple user can reference different versions.

Tags

#mlops #course #data