Remove default from metrics (#507)

* remove_default_from_metrics * merge
deepchecks · Jan 5, 2022 · a33f540 · a33f540
1 parent a54f6fe
commit a33f540
Show file tree

Hide file tree

Showing 4 changed files with 23 additions and 23 deletions.
diff --git a/deepchecks/utils/metrics.py b/deepchecks/utils/metrics.py
@@ -56,39 +56,39 @@ class ModelType(enum.Enum):
 
 
 DEFAULT_BINARY_SCORERS = {
-    'Accuracy (Default)': 'accuracy',
-    'Precision (Default)': make_scorer(precision_score, zero_division=0),
-    'Recall (Default)':  make_scorer(recall_score, zero_division=0)
+    'Accuracy': 'accuracy',
+    'Precision': make_scorer(precision_score, zero_division=0),
+    'Recall':  make_scorer(recall_score, zero_division=0)
 }
 
 
 DEFAULT_MULTICLASS_SCORERS = {
-    'Accuracy (Default)': 'accuracy',
-    'Precision - Macro Average (Default)': make_scorer(precision_score, average='macro', zero_division=0),
-    'Recall - Macro Average (Default)': make_scorer(recall_score, average='macro', zero_division=0)
+    'Accuracy': 'accuracy',
+    'Precision - Macro Average': make_scorer(precision_score, average='macro', zero_division=0),
+    'Recall - Macro Average': make_scorer(recall_score, average='macro', zero_division=0)
 }
 
 MULTICLASS_SCORERS_NON_AVERAGE = {
-    'F1 (Default)': make_scorer(f1_score, average=None, zero_division=0),
-    'Precision (Default)': make_scorer(precision_score, average=None, zero_division=0),
-    'Recall (Default)': make_scorer(recall_score, average=None, zero_division=0)
+    'F1': make_scorer(f1_score, average=None, zero_division=0),
+    'Precision': make_scorer(precision_score, average=None, zero_division=0),
+    'Recall': make_scorer(recall_score, average=None, zero_division=0)
 }
 
 
 DEFAULT_REGRESSION_SCORERS = {
-    'Neg RMSE (Default)': 'neg_root_mean_squared_error',
-    'Neg MAE (Default)': 'neg_mean_absolute_error',
-    'R2 (Default)': 'r2'
+    'Neg RMSE': 'neg_root_mean_squared_error',
+    'Neg MAE': 'neg_mean_absolute_error',
+    'R2': 'r2'
 }
 
 
 DEFAULT_SINGLE_SCORER = {
-    ModelType.BINARY: 'Accuracy (Default)',
-    ModelType.MULTICLASS: 'Accuracy (Default)',
-    ModelType.REGRESSION: 'Neg RMSE (Default)'
+    ModelType.BINARY: 'Accuracy',
+    ModelType.MULTICLASS: 'Accuracy',
+    ModelType.REGRESSION: 'Neg RMSE'
 }
 
-DEFAULT_SINGLE_SCORER_MULTICLASS_NON_AVG = 'F1 (Default)'
+DEFAULT_SINGLE_SCORER_MULTICLASS_NON_AVG = 'F1'
 
 
 DEFAULT_SCORERS_DICT = {

diff --git a/tests/checks/performance/model_error_analysis_test.py b/tests/checks/performance/model_error_analysis_test.py
@@ -70,7 +70,7 @@ def test_condition_fail(iris_labeled_dataset, iris_adaboost):
         equal_condition_result(
             is_pass=False,
             name='The performance of the detected segments must not differ by more than 5.00%',
-            details='Change in Accuracy (Default) in features: petal length (cm) exceeds threshold.',
+            details='Change in Accuracy in features: petal length (cm) exceeds threshold.',
             category=ConditionCategory.WARN
         )
     ))

diff --git a/tests/checks/performance/performance_report_test.py b/tests/checks/performance/performance_report_test.py
@@ -170,7 +170,7 @@ def test_condition_degradation_ratio_not_greater_than_not_passed(iris_split_data
     # Assert
     assert_that(result, has_items(
         equal_condition_result(is_pass=False,
-                               details=re.compile(r'F1 \(Default\) for class 1 \(train=0.94 test=0.88\)'),
+                               details=re.compile(r'F1 for class 1 \(train=0.94 test=0.88\)'),
                                name='Train-Test scores relative degradation is not greater than 0')
     ))
 
@@ -198,7 +198,7 @@ def test_condition_class_performance_imbalance_ratio_not_greater_than_not_passed
         equal_condition_result(is_pass=False,
                                details=re.compile('Relative ratio difference between highest and '
                                                   'lowest in Test dataset classes is 14.29%'),
-                               name='Relative ratio difference between labels \'F1 (Default)\' '
+                               name='Relative ratio difference between labels \'F1\' '
                                     'score is not greater than 0%')
     ))
 
@@ -212,6 +212,6 @@ def test_condition_class_performance_imbalance_ratio_not_greater_than_passed(iri
     # Assert
     assert_that(result, has_items(
         equal_condition_result(is_pass=True,
-                               name='Relative ratio difference between labels \'F1 (Default)\' '
+                               name='Relative ratio difference between labels \'F1\' '
                                'score is not greater than 100%')
     ))
diff --git a/tests/checks/performance/simple_model_comparison_test.py b/tests/checks/performance/simple_model_comparison_test.py
@@ -112,7 +112,7 @@ def test_condition_ratio_not_less_than_not_passed(diabetes_split_dataset_and_mod
         equal_condition_result(
             is_pass=False,
             name='Model performance gain over simple model must be at least 40.00%',
-            details='Metrics failed: "Neg RMSE (Default)"')
+            details='Metrics failed: "Neg RMSE"')
     ))
 
 
@@ -127,7 +127,7 @@ def test_condition_failed_for_multiclass(iris_split_dataset_and_model):
         equal_condition_result(
             is_pass=False,
             name='Model performance gain over simple model must be at least 80.00%',
-            details='Metrics failed: "F1 (Default)" - Classes: 1')
+            details='Metrics failed: "F1" - Classes: 1')
     ))
 
 
@@ -157,7 +157,7 @@ def test_condition_pass_for_multiclass_avg_with_classes(iris_split_dataset_and_m
         equal_condition_result(
             is_pass=False,
             name='Model performance gain over simple model must be at least 100%',
-            details='Metrics failed: "F1 (Default)" - Classes: 1, 2'
+            details='Metrics failed: "F1" - Classes: 1, 2'
         ),
         equal_condition_result(
             is_pass=True,