2 9.22. Window Functions #
4 Window functions provide the ability to perform calculations across
5 sets of rows that are related to the current query row. See Section 3.5
6 for an introduction to this feature, and Section 4.2.8 for syntax
9 The built-in window functions are listed in Table 9.67. Note that these
10 functions must be invoked using window function syntax, i.e., an OVER
13 In addition to these functions, any built-in or user-defined ordinary
14 aggregate (i.e., not ordered-set or hypothetical-set aggregates) can be
15 used as a window function; see Section 9.21 for a list of the built-in
16 aggregates. Aggregate functions act as window functions only when an
17 OVER clause follows the call; otherwise they act as plain aggregates
18 and return a single row for the entire set.
20 Table 9.67. General-Purpose Window Functions
26 row_number () → bigint
28 Returns the number of the current row within its partition, counting
33 Returns the rank of the current row, with gaps; that is, the row_number
34 of the first row in its peer group.
36 dense_rank () → bigint
38 Returns the rank of the current row, without gaps; this function
39 effectively counts peer groups.
41 percent_rank () → double precision
43 Returns the relative rank of the current row, that is (rank - 1) /
44 (total partition rows - 1). The value thus ranges from 0 to 1
47 cume_dist () → double precision
49 Returns the cumulative distribution, that is (number of partition rows
50 preceding or peers with current row) / (total partition rows). The
51 value thus ranges from 1/N to 1.
53 ntile ( num_buckets integer ) → integer
55 Returns an integer ranging from 1 to the argument value, dividing the
56 partition as equally as possible.
58 lag ( value anycompatible [, offset integer [, default anycompatible ]]
61 Returns value evaluated at the row that is offset rows before the
62 current row within the partition; if there is no such row, instead
63 returns default (which must be of a type compatible with value). Both
64 offset and default are evaluated with respect to the current row. If
65 omitted, offset defaults to 1 and default to NULL.
67 lead ( value anycompatible [, offset integer [, default anycompatible
70 Returns value evaluated at the row that is offset rows after the
71 current row within the partition; if there is no such row, instead
72 returns default (which must be of a type compatible with value). Both
73 offset and default are evaluated with respect to the current row. If
74 omitted, offset defaults to 1 and default to NULL.
76 first_value ( value anyelement ) → anyelement
78 Returns value evaluated at the row that is the first row of the window
81 last_value ( value anyelement ) → anyelement
83 Returns value evaluated at the row that is the last row of the window
86 nth_value ( value anyelement, n integer ) → anyelement
88 Returns value evaluated at the row that is the n'th row of the window
89 frame (counting from 1); returns NULL if there is no such row.
91 All of the functions listed in Table 9.67 depend on the sort ordering
92 specified by the ORDER BY clause of the associated window definition.
93 Rows that are not distinct when considering only the ORDER BY columns
94 are said to be peers. The four ranking functions (including cume_dist)
95 are defined so that they give the same answer for all rows of a peer
98 Note that first_value, last_value, and nth_value consider only the rows
99 within the “window frame”, which by default contains the rows from the
100 start of the partition through the last peer of the current row. This
101 is likely to give unhelpful results for last_value and sometimes also
102 nth_value. You can redefine the frame by adding a suitable frame
103 specification (RANGE, ROWS or GROUPS) to the OVER clause. See
104 Section 4.2.8 for more information about frame specifications.
106 When an aggregate function is used as a window function, it aggregates
107 over the rows within the current row's window frame. An aggregate used
108 with ORDER BY and the default window frame definition produces a
109 “running sum” type of behavior, which may or may not be what's wanted.
110 To obtain aggregation over the whole partition, omit ORDER BY or use
111 ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING. Other frame
112 specifications can be used to obtain other effects.
116 The SQL standard defines a RESPECT NULLS or IGNORE NULLS option for
117 lead, lag, first_value, last_value, and nth_value. This is not
118 implemented in PostgreSQL: the behavior is always the same as the
119 standard's default, namely RESPECT NULLS. Likewise, the standard's FROM
120 FIRST or FROM LAST option for nth_value is not implemented: only the
121 default FROM FIRST behavior is supported. (You can achieve the result
122 of FROM LAST by reversing the ORDER BY ordering.)