#11 · Primary category: Data Quality & Cleaning

Units_of_Measure_Harmonization-intelligence-platform

artificial-intelligence automation data-cleaning data-quality knime machine-learning manufacturing supply-chain

Production-Grade ML System for Automated Unit of Measure Error Detection | 88-92% Accuracy | 94% Autonomy | KNIME Workflow

Project last updated:05/21/26

GitHub Stars

817

Forks

753

Contributors

1

License

Apache-2.0

Why we included this project

Unit of measure errors are a quiet source of real losses in manufacturing and procurement, and this KNIME workflow goes after them directly. It reads CSV or Excel files, runs an XGBoost classifier over 60+ engineered features to flag suspicious values, then checks each one against NIST-compliant physics-based conversion rules before correcting it. A Q-learning agent handles the routine fixes on its own, while the built-in dashboard shows confidence scores, root-cause analytics, and anything that still needs a human reviewer. That pairing of ML detection with rule-based verification is what makes the results defensible, which matters when you need auditable data cleaning rather than a black-box model. It runs in KNIME Analytics Platform 4.5+, so teams already comfortable with visual workflow tooling can pick it up quickly, and the approach carries over to domains beyond manufacturing.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category