[SPARK-36367] Fix the behavior to follow pandas >= 1.3

• Type: Umbrella
• Status: Resolved
• Priority: Major
• Resolution: Done
• Affects Version/s: 3.2.0
• Fix Version/s: 3.2.0
• Component/s: PySpark
• Assignee: Haejoon Lee
• Reporter: Takuya Ueshin
• Created: 30/Jul/21 23:25
• Updated: 12/Dec/22 18:10
• Resolved: 26/Aug/21 23:56

Description

Pandas 1.3 has been released. We should follow the new pandas behavior.

Issue Links

• links to: [Github] Pull Request #33598 (ueshin)
• links to: [Github] Pull Request #33598 (ueshin)
• links to: [Github] Pull Request #33614 (ueshin)
• links to: [Github] Pull Request #33687 (Cedric-Magnan)
• links to: [Github] Pull Request #33687 (Cedric-Magnan)

Sub-Tasks

• Fix CategoricalOps.astype to follow pandas 1.3: Resolved
• Fix Index.union to follow pandas 1.3: Resolved
• Avoid using SelectionMixin._builtin_table which is removed in pandas 1.3: Resolved
• Fix DataFrame groupby-expanding to follow pandas 1.3: Resolved
• Fix DataFrame groupby-rolling to follow pandas 1.3: Resolved
• Fix Series.astype from datetime to nullable string: Resolved
• Revisit disabled tests for CategoricalDtype: Resolved