Intermediate · Data

Data Versioning

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. Tracking changes to datasets the way Git tracks changes to code.

Technical Definition

Tracking changes to datasets the way Git tracks changes to code.

How it works

Reproducible ML requires knowing exactly which data trained which model. Tools like DVC, LakeFS, and Delta Lake snapshot datasets, enabling rollbacks, audits, and side-by-side experiments. Without versioning, regressions become impossible to diagnose.

Related Concepts

• Data Pipeline — An automated workflow that ingests, transforms, and delivers data to models.
• Experiment Tracking — Systematically recording the inputs, code, parameters, and outputs of every ML experiment.
• Reproducibility — The ability for others to re-run an experiment and obtain the same results.