Skip to content

[SPARK-52394][PS] Fix autocorr divide-by-zero error under ANSI mode #51192

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

Closed
wants to merge 4 commits into from
Closed

[SPARK-52394][PS] Fix autocorr divide-by-zero error under ANSI mode #51192

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
7b45422
chk F.covar_samp
xinrong-meng Aug 4, 2025
6d3b339
test
xinrong-meng Aug 4, 2025
75ef14a
use head
xinrong-meng Aug 5, 2025
e3876f1
comment
xinrong-meng Aug 5, 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
13 changes: 8 additions & 5 deletions python/pyspark/pandas/series.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3396,11 +3396,14 @@ def autocorr(self, lag: int = 1) -> float:
else:
lag_scol = F.lag(scol, lag).over(Window.orderBy(NATURAL_ORDER_COLUMN_NAME))
lag_col_name = verify_temp_column_name(sdf, "__autocorr_lag_tmp_col__")
corr = (
sdf.withColumn(lag_col_name, lag_scol)
.select(F.corr(scol, F.col(lag_col_name)))
Copy link
Contributor

Choose a reason for hiding this comment

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

how does corr affected by ansi?

Copy link
Member Author

@xinrong-meng xinrong-meng Aug 4, 2025 •
edited
Loading

Choose a reason for hiding this comment

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

The example below shows how F.corr(col1, col2) fails with a DIVIDE_BY_ZERO error when ANSI on

>>> df = spark.createDataFrame(
...     [(1, None), (0, 1), (0, 0), (0, 0)],
...     ["val", "lag"]
... )
>>> df.show()
+---+----+
|val| lag|
+---+----+
|  1|NULL|
|  0|   1|
|  0|   0|
|  0|   0|
+---+----+
>>> df.select(F.corr("val", "lag")).show()
...
org.apache.spark.SparkArithmeticException: [DIVIDE_BY_ZERO] Division by zero. Use `try_divide` to tolerate divisor being 0 and return NULL instead. If necessary set "spark.sql.ansi.enabled" to "false" to bypass this error. SQLSTATE: 22012
== DataFrame ==
"corr" was called from
...

>>> spark.conf.set("spark.sql.ansi.enabled", False)
>>> df.select(F.corr("val", "lag")).show()
+--------------+
|corr(val, lag)|
+--------------+
|          NULL|
+--------------+

.head()[0]
)

sdf_lag = sdf.withColumn(lag_col_name, lag_scol)
if is_ansi_mode_enabled(sdf.sparkSession):
cov_value = sdf_lag.select(F.covar_samp(scol, F.col(lag_col_name))).first()[0]
Copy link
Member

Choose a reason for hiding this comment

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

nit: Why is this using first() instead of head()?

Copy link
Member Author

Choose a reason for hiding this comment

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

Used head for consistency.

Copy link
Member

Choose a reason for hiding this comment

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

I'm not sure the relationship between these functions and corr. Could you add a comment to explain why it works?

Copy link
Member Author

Choose a reason for hiding this comment

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

Good point, added comments

Copy link
Member

Choose a reason for hiding this comment

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

cc @zhengruifeng

if cov_value is None or cov_value == 0.0:
return np.nan
corr = sdf_lag.select(F.corr(scol, F.col(lag_col_name))).head()[0]

return np.nan if corr is None else corr

def corr(
Expand Down
3 changes: 3 additions & 0 deletions python/pyspark/pandas/tests/series/test_stat.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -607,6 +607,9 @@ def test_autocorr(self):
with self.assertRaisesRegex(TypeError, r"lag should be an int; however, got"):
psser.autocorr(1.0)

psser = ps.Series([1, 0, 0, 0])
self.assertTrue(bool(np.isnan(psser.autocorr())))

def _test_autocorr(self, pdf):
psdf = ps.from_pandas(pdf)
for lag in range(-10, 10):
Expand Down