Learn in a notebook

Start with classification. Each notebook downloads real research data, prints the table, and explains what happens next. No helper file or testing context is needed. Install the existing Parquet extra and configure your workspace key locally first.

Notebook	What you learn
Classification	Iris → inspect → train/test → fit → predict → accuracy and per-class report
Regression	Yacht → inspect → train/test → fit → predict → MAE, RMSE and R²
Async jobs	The Iris workflow with durable submission and async waiting
Artifacts and usage	Evaluate Iris, reopen the same predictor and inspect operation events
Research datasets	Search, inspect a fixed eligible asset and complete your first classification
Causal discovery	Inspect real Iris measurements and native proposed graph structure

Use a fresh Python3.11–3.13 environment with welt-client[parquet]. Outputs in notebook sources are cleared; your own run shows actual data and measurements. Keep credentials in WELT_API_KEY, never in saved notebook cells. Every modelling call goes to Welt; created private training data/predictors/results persist.

The causal example has no ground-truth graph and makes no causal-accuracy claim. Native scores are not calibrated confidence; approved public Ergodic continuation remains gated. Local release checks patch transport outside the public notebook and execute using authorized SHA-pinned research files. Public CI validates sources/contracts without downloading research bytes or using account credentials. Fixture scores never measure model quality; actual hosted acceptance is separately recorded.