Description
The import for CategoricalDtype changed in Pandas from 0.23 to 1.0 and currently pyspark checks 2 places to import. It would be better check the type as a string and avoid any imports.
Attachments
Issue Links
- is related to
-
SPARK-25351 Handle Pandas category type when converting from Python with Arrow
- Resolved
-
SPARK-31963 Support both pandas 0.23 and 1.0
- Resolved
- links to