Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[SPARK-50884][ML][PYTHON][CONNECT] Support isLargerBetter in Evaluator #49620

Closed
wants to merge 3 commits into from
+83 −0
Closed

[SPARK-50884][ML][PYTHON][CONNECT] Support isLargerBetter in Evaluator #49620

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
a948963
Support isLargerBetter in Evaluator
wbo4958 Jan 23, 2025
a28d7f1
Merge remote-tracking branch 'upstream/master' into isLargerBetter
wbo4958 Jan 24, 2025
2945920
comments
wbo4958 Jan 24, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
29 changes: 29 additions & 0 deletions python/pyspark/ml/evaluation.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -311,6 +311,9 @@ def setParams(
kwargs = self._input_kwargs
return self._set(**kwargs)

def isLargerBetter(self) -> bool:
return True


@inherit_doc
class RegressionEvaluator(
Expand Down Expand Up @@ -467,6 +470,11 @@ def setParams(
kwargs = self._input_kwargs
return self._set(**kwargs)

def isLargerBetter(self) -> bool:
if self.getMetricName() in ["r2", "var"]:
return True
return False
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
if self.getMetricName() in ["r2", "var"]:
return True
return False
return self.getMetricName() in ["r2", "var"]

we can simplify it a bit



@inherit_doc
class MulticlassClassificationEvaluator(
Expand Down Expand Up @@ -700,6 +708,16 @@ def setParams(
kwargs = self._input_kwargs
return self._set(**kwargs)

def isLargerBetter(self) -> bool:
if self.getMetricName() in [
"weightedFalsePositiveRate",
"falsePositiveRateByLabel",
"logLoss",
"hammingLoss",
]:
return False
return True


@inherit_doc
class MultilabelClassificationEvaluator(
Expand Down Expand Up @@ -843,6 +861,11 @@ def setParams(
kwargs = self._input_kwargs
return self._set(**kwargs)

def isLargerBetter(self) -> bool:
if self.getMetricName() == "hammingLoss":
return False
return True


@inherit_doc
class ClusteringEvaluator(
Expand Down Expand Up @@ -1002,6 +1025,9 @@ def setWeightCol(self, value: str) -> "ClusteringEvaluator":
"""
return self._set(weightCol=value)

def isLargerBetter(self) -> bool:
return True


@inherit_doc
class RankingEvaluator(
Expand Down Expand Up @@ -1138,6 +1164,9 @@ def setParams(
kwargs = self._input_kwargs
return self._set(**kwargs)

def isLargerBetter(self) -> bool:
return True


if __name__ == "__main__":
import doctest
Expand Down
54 changes: 54 additions & 0 deletions python/pyspark/ml/tests/test_evaluation.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,6 +42,7 @@ def test_ranking_evaluator(self):

# Initialize RankingEvaluator
evaluator = RankingEvaluator().setPredictionCol("prediction")
self.assertTrue(evaluator.isLargerBetter())

# Evaluate the dataset using the default metric (mean average precision)
mean_average_precision = evaluator.evaluate(dataset)
Expand Down Expand Up @@ -94,6 +95,25 @@ def test_multilabel_classification_evaluator(self):
self.assertEqual(evaluator2.getPredictionCol(), "prediction")
self.assertEqual(str(evaluator), str(evaluator2))

for metric in [
"subsetAccuracy",
"accuracy",
"precision",
"recall",
"f1Measure",
"precisionByLabel",
"recallByLabel",
"f1MeasureByLabel",
"microPrecision",
"microRecall",
"microF1Measure",
]:
evaluator.setMetricName(metric)
self.assertTrue(evaluator.isLargerBetter())

evaluator.setMetricName("hammingLoss")
self.assertTrue(not evaluator.isLargerBetter())

def test_multiclass_classification_evaluator(self):
dataset = self.spark.createDataFrame(
[
Expand Down Expand Up @@ -163,6 +183,29 @@ def test_multiclass_classification_evaluator(self):
log_loss = evaluator.evaluate(dataset)
self.assertTrue(np.allclose(log_loss, 1.0093, atol=1e-4))

for metric in [
"f1",
"accuracy",
"weightedPrecision",
"weightedRecall",
"weightedTruePositiveRate",
"weightedFMeasure",
"truePositiveRateByLabel",
"precisionByLabel",
"recallByLabel",
"fMeasureByLabel",
]:
evaluator.setMetricName(metric)
self.assertTrue(evaluator.isLargerBetter())
for metric in [
"weightedFalsePositiveRate",
"falsePositiveRateByLabel",
"logLoss",
"hammingLoss",
]:
evaluator.setMetricName(metric)
self.assertTrue(not evaluator.isLargerBetter())

def test_binary_classification_evaluator(self):
# Define score and labels data
data = map(
Expand All @@ -180,6 +223,8 @@ def test_binary_classification_evaluator(self):
dataset = self.spark.createDataFrame(data, ["raw", "label", "weight"])

evaluator = BinaryClassificationEvaluator().setRawPredictionCol("raw")
self.assertTrue(evaluator.isLargerBetter())

auc_roc = evaluator.evaluate(dataset)
self.assertTrue(np.allclose(auc_roc, 0.7083, atol=1e-4))

Expand Down Expand Up @@ -226,6 +271,8 @@ def test_clustering_evaluator(self):
dataset = self.spark.createDataFrame(data, ["features", "prediction", "weight"])

evaluator = ClusteringEvaluator().setPredictionCol("prediction")
self.assertTrue(evaluator.isLargerBetter())

score = evaluator.evaluate(dataset)
self.assertTrue(np.allclose(score, 0.9079, atol=1e-4))

Expand Down Expand Up @@ -300,6 +347,13 @@ def test_regression_evaluator(self):
through_origin = evaluator_with_weights.getThroughOrigin()
self.assertEqual(through_origin, False)

for metric in ["mse", "rmse", "mae"]:
evaluator.setMetricName(metric)
self.assertTrue(not evaluator.isLargerBetter())
for metric in ["r2", "var"]:
evaluator.setMetricName(metric)
self.assertTrue(evaluator.isLargerBetter())


class EvaluatorTests(EvaluatorTestsMixin, unittest.TestCase):
def setUp(self) -> None:
Expand Down
Loading