Update metrics

Author: lixfz

hypernets/tabular/metrics.py

@@ -2,6 +2,7 @@
 """
 
 """
+import inspect
 import math
 import os
 import pickle
@@ -18,6 +19,10 @@
 
 _MIN_BATCH_SIZE = 100000
 
+_DEFAULT_RECALL_OPTIONS = {}
+if 'zero_division' in inspect.signature(sk_metrics.recall_score).parameters.keys():
+    _DEFAULT_RECALL_OPTIONS['zero_division'] = 0.0
+
 
 def _task_to_average(task):
     if task == const.TASK_MULTICLASS:
@@ -37,10 +42,15 @@ def calc_score(y_true, y_preds, y_proba=None, metrics=('accuracy',), task=const.
     if len(y_preds.shape) == 2 and y_preds.shape[-1] == 1:
         y_preds = y_preds.reshape(-1)
 
+    recall_options = _DEFAULT_RECALL_OPTIONS.copy()
+
     if average is None:
         average = _task_to_average(task)
+    recall_options['average'] = average
+
+    if classes is not None:
+        recall_options['labels'] = classes
 
-    recall_options = dict(average=average, labels=classes)
     if pos_label is not None:
         recall_options['pos_label'] = pos_label
 
@@ -112,7 +122,8 @@ def metric_to_scoring(metric, task=const.TASK_BINARY, pos_label=None):
         raise ValueError(f'Not found matching scoring for {metric}')
 
     if metric_lower in metric2fn.keys():
-        options = dict(average=_task_to_average(task))
+        options = _DEFAULT_RECALL_OPTIONS.copy()
+        options['average'] = _task_to_average(task)
         if pos_label is not None:
             options['pos_label'] = pos_label
         scoring = sk_metrics.make_scorer(metric2fn[metric_lower], **options)