Skip to main content

Build an ML model

Train classification or regression from gold data or a sample.

Fastest path

Never used the lake? Start with sample quick start.

Before you start​

PathYou need
Your dataGold published for ML, status ready
SampleNothing — Housing or Iris in the wizard

Train from gold​

  1. Machine Learning → New model
  2. Published dataset (Gold) → select artifact
  3. Pick target column
  4. Name model → optional Advanced search effort
  5. Train → watch Training queue
No upload step

Gold already passed ML checks. The wizard loads schema from the artifact.

Train from sample​

  1. Wizard → Sample datasets
  2. Housing or Iris → target → Train

After training​

Troubleshooting​

Dataset missing from list
Open Data → Gold. Confirm ML publish is on and status is ready.

File upload removed on purpose

Direct uploads skip governance. Use the Data lake; keep samples for demos only.

Used Forge?
Forge outputs bronze. Publish gold for ML — avoid deprecated “import to ML” shortcuts.

Target column wrong?

Training fails or scores look nonsense. Re-check gold metadata and republish if the target changed.