sklearn.linear_model (original) (raw)

A variety of linear models.

User guide. See the Linear Models section for further details.

The following subsections are only rough guidelines: the same estimator can fall into multiple categories, depending on its parameters.

Linear classifiers#

Classical linear regressors#

Regressors with variable selection#

The following estimators have built-in variable selection fitting procedures, but any estimator using a L1 or elastic-net penalty also performs variable selection: typically SGDRegressor or SGDClassifier with an appropriate penalty.

Bayesian regressors#

Multi-task linear regressors with variable selection#

These estimators fit multiple regression problems (or tasks) jointly, while inducing sparse coefficients. While the inferred coefficients may differ between the tasks, they are constrained to agree on the features that are selected (non-zero coefficients).

Outlier-robust regressors#

Any estimator using the Huber loss would also be robust to outliers, e.g., SGDRegressor with loss='huber'.

Generalized linear models (GLM) for regression#

These models allow for response variables to have error distributions other than a normal distribution.

Miscellaneous#