Fangchen Li created SPARK-54901:
-----------------------------------
Summary: Selective column conversion for scalar Pandas UDFs
Key: SPARK-54901
URL: https://issues.apache.org/jira/browse/SPARK-54901
Project: Spark
Issue Type: Improvement
Components: PySpark
Affects Versions: 4.2.0
Reporter: Fangchen Li
When executing a scalar Pandas UDF, PySpark currently converts all Arrow
columns to Pandas Series, even if the UDF only uses a subset of columns. This
is wasteful when working with wide DataFrames, where the UDF needs only a few
columns.
We could convert Arrow columns that are actually used by the UDF(s).
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
