Conceptdata versioning
Data Version Control
An engineering practice for managing machine learning training datasets in a versioned manner, tracking changes to datasets over time. Common tools include DVC and LakeFS. The practice addresses the temporal degradation of training data relevance and supports reproducibility and auditability of machine learning experiments.
Timeline (last 90 days)
Oct 9
Some teams have begun adopting data version control tools such as DVC and LakeFS, and are experimenting with attaching business rule version numbers to each training sample.
Unverified50%