QUANTILE_CONT() GROUP BY ...

All functions > GROUP BY > QUANTILE_CONT() GROUP BY ...

Returns the continuous quantile of the group at the given probability in (0, 1); result type is DOUBLE for numeric inputs and matches the expression type for DATE or TIMESTAMP.

Syntax

QUANTILE_CONT(expr, quantile)
 [ FILTER (WHERE condition) ]
  [ GROUP BY feature [, feature ...] ]

QUANTILE_CONT(expr, quantile)
 [ FILTER (WHERE condition) ]
  [ GROUP BY feature [, feature ...] ]

Notes

• NULL values are ignored where the underlying aggregate skips NULLs
• Can be used with FILTER (WHERE condition) and GROUP BY like other aggregates
• quantile uses DOUBLE (for example 0.5e0 for the median along the continuous definition)

Examples

SELECT
    -- Continuous 0.5 quantile over one group
    f1 := ZIP(ARRAY(1, 1, 1) AS g, ARRAY(1e0, 2e0, 3e0) AS x).TRANSFORM(
        SELECT QUANTILE_CONT(x, 0.5e0) GROUP BY g
    ).UNWRAP_ONE()
;