scikit-learn · helper-uttam · Jun 21, 2021
diff --git a/sklearn/metrics/_scorer.py b/sklearn/metrics/_scorer.py
@@ -46,6 +46,7 @@
     brier_score_loss,
     jaccard_score,
     mean_absolute_percentage_error,
+    root_mean_squared_log_error,
 )
 
 from .cluster import adjusted_rand_score
@@ -585,6 +586,10 @@ def make_scorer(
     needs_threshold=False,
     **kwargs,
 ):
+
+def root_mean_squared_log_error(y_true, y_preds):
+    return np.sqrt(mean_squared_log_error(y_test, y_preds))
+
     """Make a scorer from a performance metric or loss function.
 
     This factory function wraps scoring functions for use in
@@ -690,6 +695,9 @@ def make_scorer(
 neg_mean_absolute_percentage_error_scorer = make_scorer(
     mean_absolute_percentage_error, greater_is_better=False
 )
+neg_root_mean_squared_log_error_scorer = make_scorer(
+    root_mean_squared_log_error, greater_is_better=False
+)
 neg_median_absolute_error_scorer = make_scorer(
     median_absolute_error, greater_is_better=False
 )