deepchecks - npm Package Compare versions

Comparing version 0.19.0 to 0.18.0.dev1

deepchecks.egg-info/PKG-INFO

 Metadata-Version: 2.1
 Name: deepchecks
-Version: 0.19.0
+Version: 0.18.0.dev1
 Summary: Package for validating your machine learning model and data
@@ -9,3 +9,3 @@ Home-page: https://github.com/deepchecks/deepchecks
 License: UNKNOWN
-Download-URL: https://github.com/deepchecks/deepchecks/releases/download/0.19.0/deepchecks-0.19.0.tar.gz
+Download-URL: https://github.com/deepchecks/deepchecks/releases/download/0.18.0.dev1/deepchecks-0.18.0.dev1.tar.gz
 Project-URL: Documentation, https://docs.deepchecks.com
@@ -12,0 +12,0 @@ Project-URL: Bug Reports, https://github.com/deepchecks/deepchecks

deepchecks.egg-info/requires.txt

@@ -1,3 +0,3 @@
-pandas<2.2.0,>=1.1.5
-scikit-learn<1.4.0,>=0.23.2
+pandas>=1.1.5
+scikit-learn>=0.23.2
 jsonpickle>=2
@@ -8,3 +8,3 @@ PyNomaly>=0.3.3
 category-encoders>=2.3.0
-scipy<=1.10.1,>=1.4.1
+scipy>=1.4.1
 plotly>=5.13.1
@@ -11,0 +11,0 @@ matplotlib>=3.3.4

deepchecks/tabular/metric_utils/scorers.py

@@ -24,6 +24,12 @@ # ----------------------------------------------------------------------------
 from sklearn.metrics import get_scorer, log_loss, make_scorer, mean_absolute_error, mean_squared_error
-from sklearn.metrics._scorer import _BaseScorer, _ProbaScorer
+from sklearn.metrics._scorer import _BaseScorer
 from sklearn.preprocessing import OneHotEncoder
 
 try:
+    from sklearn.metrics._scorer import _ProbaScorer, _ThresholdScorer
+    has_proba_scorer_import = True
+except ImportError:
+    has_proba_scorer_import = False
+
+try:
     from deepchecks_metrics import f1_score, jaccard_score, precision_score, recall_score  # noqa: F401
@@ -120,14 +126,14 @@ except ImportError:
     'accuracy': get_scorer('accuracy'),
-    'precision_macro': make_scorer(precision_score, average='macro', zero_division=0),
-    'precision_micro': make_scorer(precision_score, average='micro', zero_division=0),
-    'precision_weighted': make_scorer(precision_score, average='weighted', zero_division=0),
-    'recall_macro': make_scorer(recall_score, average='macro', zero_division=0),
-    'recall_micro': make_scorer(recall_score, average='micro', zero_division=0),
-    'recall_weighted': make_scorer(recall_score, average='weighted', zero_division=0),
-    'f1_macro': make_scorer(f1_score, average='macro', zero_division=0),
-    'f1_micro': make_scorer(f1_score, average='micro', zero_division=0),
-    'f1_weighted': make_scorer(f1_score, average='weighted', zero_division=0),
-    'jaccard_macro': make_scorer(jaccard_score, average='macro', zero_division=0),
-    'jaccard_micro': make_scorer(jaccard_score, average='micro', zero_division=0),
-    'jaccard_weighted': make_scorer(jaccard_score, average='weighted', zero_division=0),
+    'precision_macro': make_scorer(precision_score, average='macro', zero_division=0, pos_label=None),
+    'precision_micro': make_scorer(precision_score, average='micro', zero_division=0, pos_label=None),
+    'precision_weighted': make_scorer(precision_score, average='weighted', zero_division=0, pos_label=None),
+    'recall_macro': make_scorer(recall_score, average='macro', zero_division=0, pos_label=None),
+    'recall_micro': make_scorer(recall_score, average='micro', zero_division=0, pos_label=None),
+    'recall_weighted': make_scorer(recall_score, average='weighted', zero_division=0, pos_label=None),
+    'f1_macro': make_scorer(f1_score, average='macro', zero_division=0, pos_label=None),
+    'f1_micro': make_scorer(f1_score, average='micro', zero_division=0, pos_label=None),
+    'f1_weighted': make_scorer(f1_score, average='weighted', zero_division=0, pos_label=None),
+    'jaccard_macro': make_scorer(jaccard_score, average='macro', zero_division=0, pos_label=None),
+    'jaccard_micro': make_scorer(jaccard_score, average='micro', zero_division=0, pos_label=None),
+    'jaccard_weighted': make_scorer(jaccard_score, average='weighted', zero_division=0, pos_label=None),
 }
@@ -230,2 +236,6 @@
         self.observed_classes = observed_classes
+        self._is_proba_scorer = (
+            (has_proba_scorer_import and isinstance(self.scorer, (_ProbaScorer, _ThresholdScorer))) or
+            (not has_proba_scorer_import and 'predict_proba' in getattr(self.scorer, '_response_method', ''))
+        )
 
@@ -246,3 +256,3 @@ @classmethod
         if isinstance(self.scorer, _BaseScorer):
-            if y_proba is not None and isinstance(self.scorer, _ProbaScorer):
+            if y_proba is not None and self._is_proba_scorer:
                 pred_to_use = y_proba
@@ -307,4 +317,3 @@ else:
         # If scorer 'needs_threshold' or 'needs_proba' than the model has to have a predict_proba method.
-        if ('needs' in self.scorer._factory_args()) and not hasattr(model,  # pylint: disable=protected-access
-                                                                    'predict_proba'):
+        if self._is_proba_scorer and not hasattr(model, 'predict_proba'):  # pylint: disable=protected-access
             raise errors.DeepchecksValueError(
@@ -347,3 +356,3 @@ f'Can\'t compute scorer {self.scorer} when predicted probabilities are '
             if model.is_binary and len(original_label_col.unique()) == 1 and len(model.predictions.unique()) == 1 and \
-                        original_label_col[0] == model.predictions[0]:
+                    original_label_col[0] == model.predictions[0]:
                 seen_class = original_label_col[0]
@@ -350,0 +359,0 @@ unseen_class = self.model_classes[0] if seen_class == self.model_classes[1] \

PKG-INFO

 Metadata-Version: 2.1
 Name: deepchecks
-Version: 0.19.0
+Version: 0.18.0.dev1
 Summary: Package for validating your machine learning model and data
@@ -9,3 +9,3 @@ Home-page: https://github.com/deepchecks/deepchecks
 License: UNKNOWN
-Download-URL: https://github.com/deepchecks/deepchecks/releases/download/0.19.0/deepchecks-0.19.0.tar.gz
+Download-URL: https://github.com/deepchecks/deepchecks/releases/download/0.18.0.dev1/deepchecks-0.18.0.dev1.tar.gz
 Project-URL: Documentation, https://docs.deepchecks.com
@@ -12,0 +12,0 @@ Project-URL: Bug Reports, https://github.com/deepchecks/deepchecks

requirements/requirements.txt

@@ -1,5 +0,5 @@
-pandas>=1.1.5,<2.2.0
+pandas>=1.1.5
 numpy>=1.19; python_version < '3.8'
 numpy>=1.22.2; python_version >= '3.8'
-scikit-learn>=0.23.2, <1.4.0
+scikit-learn>=0.23.2
 jsonpickle>=2
@@ -29,5 +29,3 @@ PyNomaly>=0.3.3
 statsmodels>=0.13.5; python_version >= '3.7'
-# Remove the <=1.10.1 dependency below once sklearn's issue is fixed. The higher version causes
-# issues with sklearn's _most_frequent() function using scipy's mode() function
-scipy>=1.4.1, <=1.10.1
+scipy>=1.4.1
 dataclasses>=0.6; python_version < '3.7'
@@ -34,0 +32,0 @@ plotly>=5.13.1

VERSION

@@ -1,1 +0,1 @@
-0.19.0
+0.18.0.dev1

Alert delta unavailable

Currently unable to show alert delta for PyPI packages.

Improved metrics

Total package byte prevSize

19922447

0

Lines of code

70255

0.01%