Use signature instead of poping kwargs

imblearn/metrics/classification.py

@@ -14,6 +14,8 @@
 import logging
 import functools
 
+from inspect import getcallargs
+
 import numpy as np
 
 from sklearn.metrics.classification import (_check_targets, _prf_divide,
@@ -22,6 +24,12 @@
 from sklearn.utils.fixes import bincount
 from sklearn.utils.multiclass import unique_labels
 
+try:
+ from inspect import signature
+except ImportError:
+ from sklearn.externals.funcsigs import signature
+
+
 LOGGER = logging.getLogger(__name__)
 
 
@@ -603,21 +611,16 @@ def compute_score(*args, **kwargs):
  # Square if desired
  if squared:
  _score = np.power(_score, 2)
- # args will contain the y_pred and y_true
- # kwargs will contain the other parameters
- labels = kwargs.get('labels', None)
- pos_label = kwargs.get('pos_label', 1)
- average = kwargs.get('average', 'binary')
- sample_weight = kwargs.get('sample_weight', None)
- # Compute the sensitivity and specificity
- dict_sen_spe = {
- 'labels': labels,
- 'pos_label': pos_label,
- 'average': average,
- 'sample_weight': sample_weight
- }
- sen, spe, _ = sensitivity_specificity_support(*args,
- **dict_sen_spe)
+ # Create the list of tags
+ tags_scoring_func = getcallargs(scoring_func, *args, **kwargs)
+ # Get the signature of the sens/spec function
+ sens_spec_sig = signature(sensitivity_specificity_support)
+ # Filter the inputs required by the sens/spec function
+ tags_sens_spec = sens_spec_sig.bind(**tags_scoring_func)
+ # Call the sens/spec function
+ sen, spe, _ = sensitivity_specificity_support(
+ *tags_sens_spec.args,
+ **tags_sens_spec.kwargs)
  # Compute the dominance
  dom = sen - spe
  return (1. + alpha * dom) * _score