Add option to display errored metrics

sdv/evaluation.py

@@ -94,7 +94,7 @@ def _select_metrics(synthetic_data, metrics):
 
 
 def evaluate(synthetic_data, real_data=None, metadata=None, root_path=None,
-             table_name=None, metrics=None, aggregate=True):
+             table_name=None, metrics=None, aggregate=True, report_errors=False):
     """Apply multiple metrics at once.
 
     Args:
@@ -117,6 +117,9 @@ def evaluate(synthetic_data, real_data=None, metadata=None, root_path=None,
             If ``get_report`` is ``False``, whether to compute the mean of all the normalized
             scores to return a single float value or return a ``dict`` containing the score
             that each metric obtained. Defaults to ``True``.
+        report_errors (bool):
+            If ``True``, report the metrics that errored out and their corresponding errors.
+            If ``False``, omit the metrics that errored out.
 
     Return:
         float or sdmetrics.MetricsReport
@@ -133,7 +136,14 @@ def evaluate(synthetic_data, real_data=None, metadata=None, root_path=None,
         synthetic_data = synthetic_data[table]
 
     scores = sdmetrics.compute_metrics(metrics, real_data, synthetic_data, metadata=metadata)
+
+    if report_errors:
+        errored_metrics = scores[~scores['error'].isnull()]
+        errored_metrics = errored_metrics[['metric', 'name', 'error']]
+        print(f'The following metrics errored out: \n %s', errored_metrics.to_string())
+
     scores.dropna(inplace=True)
+    scores.drop(columns=['error'], errors='ignore')
 
     if aggregate:
         return scores.normalized_score.mean()