models.cross_validate()

Fits model on each split’s training rows and scores it on the

Usage

Source

models.cross_validate(
    model,
    design,
    y,
    splits,
    score,
    n_jobs=None,
)

test rows.

Folds run on threads; the Rust fits release the GIL, so they run in parallel.

Parameters

model: object

Anything with fit(design, y) returning an object with predict(design): Glm, ElasticNet, Gam.

design: Design

y: list of float

splits: list of (list of int, list of int)

(train, test) rows, as k_fold returns.

score: callable

score(y_test, predicted, test_design) -> float, a loss (lower is better), such as deviance_score("poisson").

n_jobs: int = None

Threads; one per split by default.

Returns

list of float

One score per split.

Examples

>>> from prospicio.models import Design, Glm, cross_validate, deviance_score, k_fold
>>> d = Design([[1.0] * 6, [0.0, 1.0, 2.0, 3.0, 4.0, 5.0]], ["(Intercept)", "x"])
>>> y = [1.0, 2.9, 5.1, 7.0, 8.9, 11.2]
>>> scores = cross_validate(Glm("gaussian"), d, y, k_fold(6, 3, 1), deviance_score("gaussian"))
>>> len(scores)

3