lale.lib.sklearn.logistic_regression module

class lale.lib.sklearn.logistic_regression.LogisticRegression(*, solver='lbfgs', penalty='l2', dual=False, C=1.0, tol=0.0001, fit_intercept=True, intercept_scaling=1.0, class_weight=None, random_state=None, max_iter=100, multi_class='auto', verbose=0, warm_start=False, n_jobs=None, l1_ratio=None)

Bases: PlannedIndividualOp

Logistic regression linear model from scikit-learn for classification.

This documentation is auto-generated from JSON schemas.

Parameters

  • solver (‘lbfgs’, ‘liblinear’, ‘newton-cg’, ‘newton-cholesky’, ‘sag’, or ‘saga’, default ‘lbfgs’) –

    Algorithm to use in the optimization problem. Default is ‘lbfgs’. To choose a solver, you might want to consider the following aspects:

    For small datasets, ‘liblinear’ is a good choice, whereas ‘sag’ and ‘saga’ are faster for large ones; For multiclass problems, only ‘newton-cg’, ‘sag’, ‘saga’ and ‘lbfgs’ handle multinomial loss; ‘liblinear’ and is limited to one-versus-rest schemes. ‘newton-cholesky’ is a good choice for n_samples >> n_features, especially with one-hot encoded categorical features with rare categories. Note that it is limited to binary classification and the one-versus-rest reduction for multiclass classification. Be aware that the memory usage of this solver has a quadratic dependency on n_features because it explicitly computes the Hessian matrix.

    See also constraint-1, constraint-2, constraint-3, constraint-4, constraint-6.

  • penalty (‘l1’, ‘l2’, ‘elasticnet’, or None, not for optimizer, default ‘l2’) –

    Norm used in the penalization.

    See also constraint-1, constraint-2, constraint-4, constraint-5, constraint-6.

  • dual (boolean, default False) –

    Dual or primal formulation. Prefer dual=False when n_samples > n_features.

    See also constraint-2.

  • C (float, >0.0, >=0.03125 for optimizer, <=32768 for optimizer, loguniform distribution, not for optimizer, default 1.0) – Inverse regularization strength. Smaller values specify stronger regularization.

  • tol (float, >0.0, >=1e-08 for optimizer, <=0.01 for optimizer, default 0.0001) – Tolerance for stopping criteria.

  • fit_intercept (boolean, default True) – Specifies whether a constant (bias or intercept) should be added to the decision function.

  • intercept_scaling (float, >=0.0, <=1.0, uniform distribution, default 1.0) – Useful only when the solver ‘liblinear’ is used and self.fit_intercept is set to True. In this case, X becomes [X, self.intercept_scaling], i.e. a “synthetic” feature with constant value equal to intercept_scaling is appended to the instance vector. The intercept becomes “intercept_scaling * synthetic_feature_weight”. Note! the synthetic feature weight is subject to l1/l2 regularization as all other features. To lessen the effect of regularization on synthetic feature weight (and therefore on the intercept) intercept_scaling has to be increased.

  • class_weight (union type, not for optimizer, default None) –

    • None

      By default, all classes have weight 1.

    • or ‘balanced’

      Uses the values of y to automatically adjust weights inversely proportional to class frequencies in the input data as “n_samples / (n_classes * np.bincount(y))”.

    • or dict, not for optimizer

      Weights associated with classes in the form “{class_label: weight}”.

  • random_state (union type, not for optimizer, default None) –

    Seed of pseudo-random number generator for shuffling data when solver == ‘sag’, ‘saga’ or ‘liblinear’.

    • None

      RandomState used by np.random

    • or numpy.random.RandomState

      Use the provided random state, only affecting other users of that same random state instance.

    • or integer

      Explicit seed.

  • max_iter (integer, >=1, >=10 for optimizer, <=1000 for optimizer, uniform distribution, default 100) – Maximum number of iterations for solvers to converge.

  • multi_class (‘auto’, ‘ovr’, or ‘multinomial’, default ‘auto’) –

    If the option chosen is ovr, then a binary problem is fit for each label. For multinomial the loss minimised is the multinomial loss fit across the entire probability distribution, even when the data is binary. multinomial is unavailable when solver=`liblinear`. auto selects ovr if the data is binary, or if solver=`liblinear`, and otherwise selects multinomial.

    See also constraint-3.

  • verbose (integer, not for optimizer, default 0) – For the liblinear and lbfgs solvers set verbose to any positive number for verbosity.

  • warm_start (boolean, not for optimizer, default False) – When set to True, reuse the solution of the previous call to fit as initialization, otherwise, just erase the previous solution. Useless for liblinear solver.

  • n_jobs (union type, not for optimizer, default None) –

    Number of CPU cores when parallelizing over classes if multi_class is ovr. This parameter is ignored when the “solver” is set to ‘liblinear’ regardless of whether ‘multi_class’ is specified or not.

    • None

      1 unless in joblib.parallel_backend context.

    • or -1

      Use all processors.

    • or integer, >=1

      Number of CPU cores.

  • l1_ratio (union type, optional, not for optimizer, default None) –

    The Elastic-Net mixing parameter.

    • float, >=0.0, <=1.0

    • or None

    See also constraint-5.

Notes

constraint-1 : union type

The newton-cg, sag, and lbfgs solvers support only l2 or no penalties.

  • solver : negated type of ‘newton-cg’, ‘newton-cholesky’, ‘sag’, or ‘lbfgs’

  • or penalty : ‘l2’, ‘none’, or None

constraint-2 : union type

The dual formulation is only implemented for l2 penalty with the liblinear solver.

  • dual : False

  • or dict

    • penalty : ‘l2’

    • solver : ‘liblinear’

constraint-3 : union type

The multi_class multinomial option is unavailable when the solver is liblinear or newton-cholesky.

  • multi_class : negated type of ‘multinomial’

  • or solver : negated type of ‘liblinear’ or ‘newton-cholesky’

constraint-4 : union type, not for optimizer

penalty=’none’ is not supported for the liblinear solver

  • solver : negated type of ‘liblinear’

  • or penalty : negated type of ‘none’ or None

constraint-5 : union type, not for optimizer

When penalty is elasticnet, l1_ratio must be between 0 and 1.

  • penalty : negated type of ‘elasticnet’

  • or l1_ratio : float, >0.0, <=1.0

constraint-6 : union type, not for optimizer

Only ‘saga’ solver supports elasticnet penalty

  • penalty : negated type of ‘elasticnet’

  • or solver : ‘saga’

decision_function(X)

Confidence scores for all classes.

Note: The decision_function method is not available until this operator is trained.

Once this method is available, it will have the following signature:

Parameters

X (array of items : array of items : float) – Features; the outer array is over samples.

Returns

result – Confidence scores for samples for each class in the model.

  • array of items : array of items : float

    In the multi-way case, score per (sample, class) combination.

  • or array of items : float

    In the binary case, score for self._classes[1].

Return type

union type

fit(X, y=None, **fit_params)

Train the operator.

Note: The fit method is not available until this operator is trainable.

Once this method is available, it will have the following signature:

Parameters

  • X (array of items : array of items : float) – Features; the outer array is over samples.

  • y (union type) –

    Target class labels; the array is over samples.

    • array of items : float

    • or array of items : string

    • or array of items : boolean

predict(X, **predict_params)

Make predictions.

Note: The predict method is not available until this operator is trained.

Once this method is available, it will have the following signature:

Parameters

X (array of items : array of items : float) – Features; the outer array is over samples.

Returns

result – Predicted class label per sample.

  • array of items : float

  • or array of items : string

  • or array of items : boolean

Return type

union type

predict_proba(X)

Probability estimates for all classes.

Note: The predict_proba method is not available until this operator is trained.

Once this method is available, it will have the following signature:

Parameters

X (array of items : array of items : float) – Features; the outer array is over samples.

Returns

result – Probability of the sample for each class in the model.

Return type

array of items : array of items : float