ML Experimentation
How we run and evaluate ML forecasting experiments. Unlike the roadmap (which holds forward-looking design for work not yet built), this area documents methodology that is implemented and in use — the durable home for ML experimentation docs once they leave the roadmap.
Documents
- Our approach to MLops — why we automate experiments, and why the model that wins the evaluation is deployed bit-for-bit, with nothing rewritten on the way to production.
- Running an ML experiment end-to-end — step-by-step recipe for going
from raw data to a trained, MLflow-tracked model using the Dagster pipeline; explains why
trained_cv_modelreads config from MLflow rather than YAML. - Model configuration — how to set hyperparameters and choose features; the full feature vocabulary and the lookahead-bias guardrails.
- Cross-validation folds — the expanding-window CV protocol, the current single fold and why the data constrains us to it, and the target multiple-yearly-fold protocol.