4 The catalog pg_statistic stores statistical data about the contents of
5 the database. Entries are created by ANALYZE and subsequently used by
6 the query planner. Note that all the statistical data is inherently
7 approximate, even assuming that it is up-to-date.
9 Normally there is one entry, with stainherit = false, for each table
10 column that has been analyzed. If the table has inheritance children or
11 partitions, a second entry with stainherit = true is also created. This
12 row represents the column's statistics over the inheritance tree, i.e.,
13 statistics for the data you'd see with SELECT column FROM table*,
14 whereas the stainherit = false row represents the results of SELECT
15 column FROM ONLY table.
17 pg_statistic also stores statistical data about the values of index
18 expressions. These are described as if they were actual data columns;
19 in particular, starelid references the index. No entry is made for an
20 ordinary non-expression index column, however, since it would be
21 redundant with the entry for the underlying table column. Currently,
22 entries for index expressions always have stainherit = false.
24 Since different kinds of statistics might be appropriate for different
25 kinds of data, pg_statistic is designed not to assume very much about
26 what sort of statistics it stores. Only extremely general statistics
27 (such as nullness) are given dedicated columns in pg_statistic.
28 Everything else is stored in “slots”, which are groups of associated
29 columns whose content is identified by a code number in one of the
30 slot's columns. For more information see
31 src/include/catalog/pg_statistic.h.
33 pg_statistic should not be readable by the public, since even
34 statistical information about a table's contents might be considered
35 sensitive. (Example: minimum and maximum values of a salary column
36 might be quite interesting.) pg_stats is a publicly readable view on
37 pg_statistic that only exposes information about those tables that are
38 readable by the current user.
40 Table 52.51. pg_statistic Columns
46 starelid oid (references pg_class.oid)
48 The table or index that the described column belongs to
50 staattnum int2 (references pg_attribute.attnum)
52 The number of the described column
56 If true, the stats include values from child tables, not just the
57 values in the specified relation
61 The fraction of the column's entries that are null
65 The average stored width, in bytes, of nonnull entries
69 The number of distinct nonnull data values in the column. A value
70 greater than zero is the actual number of distinct values. A value less
71 than zero is the negative of a multiplier for the number of rows in the
72 table; for example, a column in which about 80% of the values are
73 nonnull and each nonnull value appears about twice on average could be
74 represented by stadistinct = -0.4. A zero value means the number of
75 distinct values is unknown.
79 A code number indicating the kind of statistics stored in the Nth
80 “slot” of the pg_statistic row.
82 staopN oid (references pg_operator.oid)
84 An operator used to derive the statistics stored in the Nth “slot”. For
85 example, a histogram slot would show the < operator that defines the
86 sort order of the data. Zero if the statistics kind does not require an
89 stacollN oid (references pg_collation.oid)
91 The collation used to derive the statistics stored in the Nth “slot”.
92 For example, a histogram slot for a collatable column would show the
93 collation that defines the sort order of the data. Zero for
98 Numerical statistics of the appropriate kind for the Nth “slot”, or
99 null if the slot kind does not involve numerical values
103 Column data values of the appropriate kind for the Nth “slot”, or null
104 if the slot kind does not store any data values. Each array's element
105 values are actually of the specific column's data type, or a related
106 type such as an array's element type, so there is no way to define
107 these columns' type more specifically than anyarray.