GROUPING

Reference material for GROUPING

This function is particularly useful when working with GROUPING SETS, ROLLUP or CUBE clauses.

GROUPING SETS, ROLLUP, and CUBE clauses produce aggregates that use NULL values to pad columns that are not part of the current grouping set. This may cause ambiguity if those columns also contain actual NULL values. The GROUPING function helps resolve this ambiguity by distinguishing between columns that are excluded from a grouping set and those that genuinely contain NULL values.

Key points about the GROUPING function:

• It accepts expressions that are also used in the GROUP BY [GROUPING SETS | ROLLUP | CUBE] clause as arguments.
• It returns an integer where each bit represents whether the corresponding column is part of the grouping set or not.
• A bit is set to 1 if the corresponding column is not part of the grouping set; otherwise, it is 0.
• The first argument corresponds to the most significant bit, with subsequent arguments following in order.

Syntax

GROUPING(<expression> [, ...n]) [FILTER ([WHERE] <condition>)]

Parameters

Parameter	Description	Supported input types
<expression>	An expression that also appears as a <grouping_element> in a GROUPING SETS, ROLLUP, or CUBE clause. The maximum number of arguments is 31.	Any
<condition>	An optional boolean expression to filter rows used in aggregation	BOOL

Return Type

GROUPING returns a value of type INT.

Examples

Consider this table addresses:

user_id	country	county	city
3	USA	California	Los Angeles
34	USA	California	Los Angeles
15	Canada	NULL	NULL
1	Canada	Quebec	Montreal
45	Canada	Quebec	Montreal
10	USA	California	Springfield
11	USA	Oregon	Springfield
23	Canada	Ontario	London
30	UK	Greater London	London

Let’s say we are interested in the number of users per region, going from the city level up to country level:

select 
    grouping(country, county, city), 
    country, 
    county, 
    city, 
    count(*)
from 
    locations
group by 
    grouping sets (
        (country, county, city),
        (country, county),
        (country)
    )
order by
    1, 2, 3, 4

The query returns this result:

grouping	country	county	city	count
0	Canada	Ontario	London	1
0	Canada	Quebec	Montreal	2
0	Canada	null	null	1
0	UK	Greater London	London	1
0	USA	California	Los Angeles	2
0	USA	California	Springfield	1
0	USA	Oregon	Springfield	1
1	Canada	Ontario	null	1
1	Canada	Quebec	null	2
1	Canada	null	null	1
1	UK	Greater London	null	1
1	USA	California	null	3
1	USA	Oregon	null	1
3	Canada	null	null	4
3	UK	null	null	1
3	USA	null	null	4

The GROUPING function evaluates to three distinct values: 0, 1, and 3. Each value corresponds to a specific grouping set defined in the GROUP BY GROUPING SETS clause.

• A value of 0 indicates that the row is grouped by all specified columns.
• A value of 1 (0b001 in binary) signifies that the city column is excluded from the grouping.
• A value of 3 (0b011 in binary) indicates that both the county and city columns are excluded from the grouping.

This allows us to differentiate between rows with the same grouping key values, such as Canada, NULL, NULL, by identifying which columns were used for aggregation in each case.