mojolearn.ARDRegression#

class mojolearn.ARDRegression(*, max_iter=300, tol=0.001, alpha_1=1e-06, alpha_2=1e-06, lambda_1=1e-06, lambda_2=1e-06, compute_score=False, threshold_lambda=10000.0, fit_intercept=True, copy_X=True, verbose=False)[source]#

Bases: _LinearRegressorMixin, NumericModeMixin

Bayesian ARD (automatic relevance determination) regression.

Follows scikit-learn’s ARDRegression: one precision per weight, estimated by evidence maximization; features whose precision exceeds threshold_lambda are pruned (their weight is 0). Needs at least 2 samples. compute_score=True and predict(return_std=True) raise ValueError; fit takes no sample_weight.

Parameters:

  • max_iter (int, default 300) – Maximum number of iterations.

  • tol (float, default 1e-3) – Stop when the weights change by less than this.

  • alpha_1, alpha_2 (float, default 1e-6) – Shape and rate of the Gamma prior over alpha.

  • lambda_1, lambda_2 (float, default 1e-6) – Shape and rate of the Gamma prior over each lambda.

  • compute_score (bool, default False) – Not supported: True raises ValueError.

  • threshold_lambda (float, default 1e4) – Precision above which a weight is pruned.

  • fit_intercept (bool, default True) – Whether to fit an intercept.

  • copy_X (bool, default True) – Accepted for compatibility; X is never modified.

  • verbose (bool, default False) – Accepted for compatibility; ignored.

  • numeric_mode (str or None, default None) – Numeric mode for this model ('identical', 'deterministic' or 'fast', which runs on Apple GPUs only); None uses the process default (see mojolearn.set_numeric_mode()).

Variables:

  • coef_ (float32 array of shape (n_features,)) – Posterior mean of the weights.

  • intercept_ (float) – Intercept.

  • alpha_ (float) – Estimated noise precision.

  • lambda_ (float32 array of shape (n_features,)) – Estimated precision of each weight.

  • n_iter_ (int) – Number of iterations run.

  • n_features_in_ (int) – Number of features seen in fit.

Examples

>>> import numpy as np, mojolearn
>>> rng = np.random.default_rng(0)
>>> X = rng.random((1000, 8), dtype=np.float32)
>>> y = 3 * X[:, 0] - 2 * X[:, 1]
>>> model = mojolearn.ARDRegression().fit(X, y)
__init__(*, max_iter=300, tol=0.001, alpha_1=1e-06, alpha_2=1e-06, lambda_1=1e-06, lambda_2=1e-06, compute_score=False, threshold_lambda=10000.0, fit_intercept=True, copy_X=True, verbose=False)[source]#

Methods

fit(X, y)

Fit the model.

get_params([deep])

numeric_mode_used()

The numeric mode this estimator runs in.

predict(X[, return_std])

Predict targets with the posterior mean of the weights.

score(X, y)

Coefficient of determination R^2 of predict(X) against y.

set_params(**params)

Set constructor parameters.

vendor_used()

The GPU backend this estimator runs on.
fit(X, y)[source]#

Fit the model.

Parameters:

  • X (array-like of shape (n_samples, n_features)) – Training data, converted to float32; values must be finite.

  • y (array-like of shape (n_samples,)) – Targets.

Returns:

self (ARDRegression) – The fitted model.

Raises:

ValueError – If compute_score is True or there are fewer than 2 samples.

predict(X, return_std=False)[source]#

Predict targets with the posterior mean of the weights.

Parameters:

  • X (array-like of shape (n_samples, n_features)) – Samples, converted to float32.

  • return_std (bool, default False) – Not supported: True raises ValueError.

Returns:

y_pred (float32 array of shape (n_samples,)) – Predicted values.