7.11 Determining Whether Values are Unique
7.11.1 Problem
You want to know whether table values are unique.
7.11.2 Solution
Use HAVING in
conjunction with COUNT( ).
7.11.3 Discussion
You can use HAVING to find unique values in
situations to which DISTINCT does not apply.
DISTINCT eliminates duplicates, but
doesn't show which values actually were duplicated
in the original data. HAVING can tell you which
values were unique or non-unique.
The following queries show the days on which only one driver was
active, and the days on which more than one driver was active.
They're based on using HAVING and
COUNT( ) to determine which
trav_date values are unique or non-unique:
mysql> SELECT trav_date, COUNT(trav_date)
-> FROM driver_log
-> GROUP BY trav_date
-> HAVING COUNT(trav_date) = 1;
+------------+------------------+
| trav_date | COUNT(trav_date) |
+------------+------------------+
| 2001-11-26 | 1 |
| 2001-11-27 | 1 |
| 2001-12-01 | 1 |
+------------+------------------+
mysql> SELECT trav_date, COUNT(trav_date)
-> FROM driver_log
-> GROUP BY trav_date
-> HAVING COUNT(trav_date) > 1;
+------------+------------------+
| trav_date | COUNT(trav_date) |
+------------+------------------+
| 2001-11-29 | 3 |
| 2001-11-30 | 2 |
| 2001-12-02 | 2 |
+------------+------------------+
This technique works for combinations of values, too. For example, to
find message sender/recipient pairs between whom only one message was
sent, look for combinations that occur only once in the
mail table:
mysql> SELECT srcuser, dstuser
-> FROM mail
-> GROUP BY srcuser, dstuser
-> HAVING COUNT(*) = 1;
+---------+---------+
| srcuser | dstuser |
+---------+---------+
| barb | barb |
| gene | tricia |
| phil | barb |
| tricia | gene |
| tricia | phil |
+---------+---------+
Note that this query doesn't print the count. The
first two examples did so, to show that the counts were being used
properly, but you can use a count in a HAVING
clause without including it in the output column list.
|
