lale.sklearn_compat module

class lale.sklearn_compat.DefaultsVisitor

Bases: lale.util.Visitor.Visitor

classmethod run(op: lale.operators.Operator) → Dict[str, Any]

visitIndividualOp(op: lale.operators.IndividualOp) → Dict[str, Any]

visitOperatorChoice(op: lale.operators.OperatorChoice) → Dict[str, Any]

visitPipeline(op: lale.operators.PlannedPipeline) → Dict[str, Any]

visitPlannedIndividualOp(op: lale.operators.IndividualOp) → Dict[str, Any]

visitPlannedPipeline(op: lale.operators.PlannedPipeline) → Dict[str, Any]

visitTrainableIndividualOp(op: lale.operators.IndividualOp) → Dict[str, Any]

visitTrainablePipeline(op: lale.operators.PlannedPipeline) → Dict[str, Any]

visitTrainedIndividualOp(op: lale.operators.IndividualOp) → Dict[str, Any]

visitTrainedPipeline(op: lale.operators.PlannedPipeline) → Dict[str, Any]

class lale.sklearn_compat.SKlearnCompatWrapper(**kwargs)

Bases: object

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

fixup_params_internal(**params)

get_params(deep: bool = True) → Dict[str, Any]

get_params_internal(out: Dict[str, Any])

hyperparam_defaults() → Dict[str, Any]

init_params_internal(**kwargs)

classmethod make_wrapper(base: lale.operators.Operator)

set_params(**impl_params)

set_params_internal(**impl_params)

to_lale() → lale.operators.Operator

class lale.sklearn_compat.WithoutGetParams(base)

Bases: object

This wrapper forwards everything except “get_attr” to what it is wrapping

classmethod clone_wgp(obj: lale.sklearn_compat.WithoutGetParams) → lale.sklearn_compat.WithoutGetParams

lale.sklearn_compat.clone_lale(op: lale.operators.Operator) → lale.operators.Operator

lale.sklearn_compat.make_sklearn_compat(op: lale.operators.Operator) → lale.sklearn_compat.SKlearnCompatWrapper

Top level function for providing compatibiltiy with sklearn operations This returns a wrapper around the provided sklearn operator graph which can be passed to sklearn methods such as clone and GridSearchCV The wrapper may modify the wrapped lale operator/pipeline as part of providing compatibility with these methods. After the sklearn operation is complete, SKlearnCompatWrapper.to_lale() can be called to recover the wrapped lale operator for future use

lale.sklearn_compat.nest_HPparam(name: str, key: str)

lale.sklearn_compat.nest_HPparams(name: str, grid: Dict[str, V]) → Dict[str, V]

lale.sklearn_compat.nest_all_HPparams(name: str, grids: List[Dict[str, V]]) → List[Dict[str, V]]

Given the name of an operator in a pipeline, this transforms every key(parameter name) in the grids to use the operator name as a prefix (separated by __). This is the convention in scikit-learn pipelines.

lale.sklearn_compat.partition_sklearn_params(d: Dict[str, Any]) → Dict[str, Dict[str, Any]]

lale.sklearn_compat.set_operator_params(op: lale.operators.Operator, **impl_params) → lale.operators.TrainableOperator

May return a new operator, in which case the old one should be overwritten

lale.sklearn_compat.sklearn_compat_clone(impl: Any) → Any

lale.sklearn_compat.unnest_HPparams(k: str) → List[str]