Module automl 

AutoML — the framework, pointed at itself.

Everything AutoML needs already exists: preprocessing transformers, the CV engine, and the ensemble core. AutoML orchestrates them — searching preprocessing × model × hyperparameters under a budget, optionally auto-ensembling the top candidates — and returns the best deployable model plus a leaderboard. No new crate; it reuses selection and ensemble.

When the eda engine is enabled, the search is seeded from Profile::suggest_pipeline: the preprocessing is fixed to EDA’s suggestion and only the model varies on top, pruning the space before a single model is fit. Without eda it falls back to searching the scaler as well.

use millwright::prelude::*;
let result = AutoML::classifier()
    .budget(Budget::trials(40))
    .metric(Metric::F1)
    .cv(StratifiedKFold::new(5))
    .fit(&train)?;

println!("{}", result.leaderboard());

Structs

AutoML

An automated model search.

AutoMLResult

The outcome of an AutoML search: the fitted winner and a leaderboard.

Enums

Budget

A search budget.