Heimdall ML
Heimdall ML is the no-code machine learning product on Heimdall by RejiCo — build and deploy classification and regression models from gold datasets or explore with Housing/Iris samples.
| Path | Requirement |
|---|---|
| Your data | Gold published for ML, status ready, 100+ rows |
| Samples | Nothing — use wizard Sample datasets |
Open ML in the app
Sidebar → Data Intelligence → Machine Learning (/ml)
- New model — opens training wizard
- Training queue — in-flight jobs
- Model list — completed models with deploy status (green dot = API key active)
ML vs Forecast vs Loop
| ML | Forecast | Loop | |
|---|---|---|---|
| Predicts | Category or number from features | Future values over time | Category/number that improves live |
| Gold required? | Yes (or samples) | Yes (or AirPassengers sample) | No — separate wizard |
| Sidebar | Machine Learning | Forecast | Loop |
Full table: What is Heimdall?
What you can build
Classification — churn, fraud, quality grades, species
Regression — pricing, demand scores, risk values
Typical path
Common mistakes
- Uploading CSV in the wizard — disabled except samples; use Lake → gold
- Empty gold picker — publish with Machine learning use case, status ready
FAQ
Do I need Lake for every ML model?
No — Housing and Iris samples skip Lake. Production data requires Gold publish.
What's the difference between ML and Forecast?
ML predicts a category or number from feature columns. Forecast predicts future values over a time axis. See product chooser.
Can I upload CSV in the ML wizard?
Only for built-in samples. Your spreadsheets go through Lake → Gold first.