Skip to content

[Doc]: Clarity on behaviour when setting partition_by=None (default) with nw.Expr.over #3542

@danielarifmurphy

Description

@danielarifmurphy

What type of report is this?

Improvement

Please describe the issue.

For lazy operations where we can't rely on order dependence, something like .over(order_by="date") is useful!

From the docs for over, it may not be immediately obvious to the reader what happens when partition_by=None, as there are only examples where this is specified to be a different column (similarly in the polars docs!), e.g., nw.col("a").cum_sum().over("b")

This is a suggestion then to include an explicit note on default behaviour here, i.e., that upon collection the expression is performed on a sorted frame and results returned with original row order (with original dimensions).

Cheers!

If you have a suggestion on how it should be, add it below.

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    documentationImprovements or additions to documentation

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions