Aggregate functions compute a single result from a set of input values. The built-in general-purpose aggregate functions are listed in Table 9.55 while statistical aggregates are in Table 9.56. The built-in within-group ordered-set aggregate functions are listed in Table 9.57 while the built-in within-group hypothetical-set ones are in Table 9.58. Grouping operations, which are closely related to aggregate functions, are listed in Table 9.59. The special syntax considerations for aggregate functions are explained in Section 4.2.7. Consult Section 2.7 for additional introductory information.
Aggregate functions that support Partial Mode are eligible to participate in various optimizations, such as parallel aggregation.
It should be noted that except for
count, these functions return a null value when no rows are selected. In particular,
sumof no rows returns null, not zero as one might expect, and
array_aggreturns null rather than an empty array when there are no input rows. The
coalescefunction can be used to substitute zero or an empty array for null when necessary.
The aggregate functions
xmlagg, as well as similar user-defined aggregate functions, produce meaningfully different result values depending on the order of the input values. This ordering is unspecified by default, but can be controlled by writing an
ORDER BYclause within the aggregate call, as shown in Section 4.2.7. Alternatively, supplying the input values from a sorted subquery will usually work. For example:
SELECT xmlagg(x) FROM (SELECT x FROM test ORDER BY y DESC) AS tab;
Beware that this approach can fail if the outer query level contains additional processing, such as a join, because that might cause the subquery's output to be reordered before the aggregate is computed.
The boolean aggregates
bool_orcorrespond to the standard SQL aggregates
some. PostgreSQL supports
every, but not
some, because there is an ambiguity built into the standard syntax:
SELECT b1 = ANY((SELECT b2 FROM t2 ...)) FROM t1 ...;
ANYcan be considered either as introducing a subquery, or as being an aggregate function, if the subquery returns one row with a Boolean value. Thus the standard name cannot be given to these aggregates.
Users accustomed to working with other SQL database management systems might be disappointed by the performance of the
countaggregate when it is applied to the entire table. A query like:
SELECT count(*) FROM sometable;
will require effort proportional to the size of the table: PostgreSQL will need to scan either the entire table or the entirety of an index that includes all rows in the table.
Table 9.56 shows aggregate functions typically used in statistical analysis. (These are separated out merely to avoid cluttering the listing of more-commonly-used aggregates.) Functions shown as accepting
numeric_typeare available for all the types
double precision. Where the description mentions
N, it means the number of input rows for which all the input expressions are non-null. In all cases, null is returned if the computation is meaningless, for example when
Table 9.57 shows some aggregate functions that use the ordered-set aggregate syntax. These functions are sometimes referred to as “inverse distribution” functions. Their aggregated input is introduced by
ORDER BY, and they may also take a direct argument that is not aggregated, but is computed only once. All these functions ignore null values in their aggregated input. For those that take a
fractionparameter, the fraction value must be between 0 and 1; an error is thrown if not. However, a null
fractionvalue simply produces a null result.
Each of the “hypothetical-set” aggregates listed in Table 9.58 is associated with a window function of the same name defined in Section 9.22. In each case, the aggregate's result is the value that the associated window function would have returned for the “hypothetical” row constructed from
args, if such a row had been added to the sorted group of rows represented by the
sorted_args. For each of these functions, the list of direct arguments given in
argsmust match the number and types of the aggregated arguments given in
sorted_args. Unlike most built-in aggregates, these aggregates are not strict, that is they do not drop input rows containing nulls. Null values sort according to the rule specified in the
The grouping operations shown in Table 9.59 are used in conjunction with grouping sets (see Section 7.2.4) to distinguish result rows. The arguments to the
GROUPINGfunction are not actually evaluated, but they must exactly match expressions given in the
GROUP BYclause of the associated query level. For example:
=> SELECT * FROM items_sold;
make | model | sales
Foo | GT | 10
Foo | Tour | 20
Bar | City | 15
Bar | Sport | 5
=> SELECT make, model, GROUPING(make,model), sum(sales) FROM items_sold GROUP BY ROLLUP(make,model);
make | model | grouping | sum
Foo | GT | 0 | 10
Foo | Tour | 0 | 20
Bar | City | 0 | 15
Bar | Sport | 0 | 5
Foo | | 1 | 30
Bar | | 1 | 20
| | 3 | 50
0in the first four rows shows that those have been grouped normally, over both the grouping columns. The value
modelwas not grouped by in the next-to-last two rows, and the value
3indicates that neither
modelwas grouped by in the last row (which therefore is an aggregate over all the input rows).