mlshell.Metric¶
-
class
mlshell.
Metric
(scorer=None, oid=None, score_func=None, score_func_vector=None, greater_is_better=True, needs_proba=False, needs_threshold=False, needs_custom_kw_args=False)[source]¶ Bases:
object
Unified pipeline interface.
Implements interface to access arbitrary scorer. Interface: pprint and all underlying scorer methods.
-
scorer
¶ Underlying scorer.
- Type
callable, optional (default=None)
-
score_func
¶ Scorer score function, return scalar value.
- Type
callable, optional (default=None)
-
score_func_vector
¶ Scorer vectorized score function, return vector of values for all samples.
- Type
callable, optional (default=None)
-
greater_is_better
¶ Whether score_func is a score function (default), meaning high is good, or a loss function, meaning low is good. In the latter case, the scorer object should sign-flip the outcome of the score_func.
- Type
bool, optional (default=True)
-
needs_proba
¶ Whether score_func requires predict_proba to get probability estimates out of a classifier.
- Type
bool, optional (default=False)
-
needs_threshold
¶ Whether score_func takes a continuous decision certainty. This only works for classification using estimators that have either a decision_function or predict_proba method.
- Type
bool, optional (default=False)
-
needs_custom_kw_args
¶ If True, before score evaluation extract scorer kwargs from pipeline ‘pass_custom’ step (if existed).
- Type
bool, optional (default=False)
Notes
Extended scorer object:
Additional
needs_custom_kw_args
kwarg.
Allows to optimize custom scorer kwargs as hyper-parameters.
Additional
score_func_vector
kwarg.
Allows to evaluate vectorized score for more detailed analyze.
- Attributes
kw_args
dict: Additional kwargs passed to score_func.
Methods
__call__
(estimator, *args, **kwargs)Redirect call to scorer object.
pprint
(score)Pretty print metric result.
-
__init__
(scorer=None, oid=None, score_func=None, score_func_vector=None, greater_is_better=True, needs_proba=False, needs_threshold=False, needs_custom_kw_args=False)[source]¶ Initialize self. See help(type(self)) for accurate signature.
Methods
__init__
([scorer, oid, score_func, …])Initialize self.
pprint
(score)Pretty print metric result.
Attributes
dict: Additional kwargs passed to score_func.
-
property
kw_args
¶ dict: Additional kwargs passed to score_func.
-