describe_by builtin

describe_by(
    values,
    categories,
    *,
    weights=None,
    quantiles=Ellipsis,
    data=None,
)

Weighted statistics per category, as describe, then over all values.

Parameters:

Name	Type	Description	Default
values	array_like or str	Values, or their column in data; NaN is skipped.	required
categories	array_like or str	Category (e.g. domain) of each value, int or str, or its column.	required
weights	array_like or str	Declustering weights; default the volumes of a BlockModel data, else 1.	None
quantiles	sequence of float	Probabilities of the quantile columns.	Ellipsis
data	(PointSet, BlockModel, Table or dict)	Where column names are looked up.	None

Returns:

Type	Description
Table	category (sorted, then "all"), n, mean, variance, std, cv, min, max and one column per quantile named P10, P97.5, ...

used in

• Describing data
• mesh-crossing interval splits
• statistics by domain
• top cuts
• swaths
• capping transform