Oracle® OLAP DML Reference 10g Release 2 (10.2) Part Number B14346-03 |
|
|
PDF · Mobi · ePub |
The RANK function computes the rank of values in a numeric expression.
DECIMAL
RANK(expression method [attributes] [BASEDON dimension-list])
where attributes is one or more of the following:
The numeric expression for which rankings are to be computed.
The method to use in computing the rank of the values in expression. The method argument can be one of the following keywords. See also "Results of Method Values".
Table 22-1 Methods for Computing RANK
Method | Description |
---|---|
MIN |
Identical values get the same minimum rank. |
MAX |
Identical values get the same maximum rank. |
AVERAGE |
Identical values get the same average rank. |
PACKED |
Identical values get the same rank but the results are packed into consecutive |
UNIQUE |
All values get a unique rank; for identical values the rank is arbitrary. |
PERCENTILE |
Values are ranked from 1 to 100, based on the relative frequency of their occurrence in the expression. |
DECILE |
Values are ranked from 1 to 10, based on the relative frequency of their occurrence in the expression. |
QUARTILE |
Values are ranked from 1 to 4, based on the relative frequency of their occurrence in the expression. |
Changes how Oracle OLAP computes RANK within a looping statement (for example, an assignment statement):
When you do not specify RESET, Oracle OLAP ranks the members of each group only once and, then caches those ranked values. As the looping statement continues to execute and RANK executes against same set of values, Oracle OLAP uses those cached values to return values for RANK.
When you include RESET, Oracle OLAP recomputes RANK each and every time it executes within the looping statement. This behavior significantly increases the time it takes for Oracle OLAP to execute the looping statement that contains RANK.
Within a looping statement, the only time you use RANK with RESET is when you know that within any group the rankings of members within that group will change during the execution of the looping statement.
Specifies that Oracle OLAP converts all NA values to the largest positive decimal number or (10**308)
before ranking the values.
Note:
An NA expression value produces an NA rank unless you specify either NAFIRST or NALIST.Specifies that Oracle OLAP converts all NA values to the largest negative decimal number or -(10**308)
before ranking the values.
Note:
An NA expression value produces an NA rank unless you specify either NAFIRST or NALIST.Specifies the status of the dimensions that Oracle OLAP uses when calculating RANK. By specifying LIMITSAVE within the RANK function, rather than specify CHGDIMS with LIMITSAVE, you insure that Oracle OLAP evaluates the status only once when RANK needs to calculate a new result.
The dimension values that Oracle OLAP uses to determine dimension status while executing RANK. For the limit-expression argument, you can specify any expression including a valueset, a LIMIT function, or a SORT function.
Specifies how Oracle OLAP sequences values of equal rank.
Any expression including a valueset, a LIMIT function, or a SORT function. Oracle OLAP executes the tiebreaker-expressions in the order in which they are specified. The status of the dimensions of each tiebreaker-expression is the current status of the dimensions or the status specified in the LIMITSAVE clause, if any.
Note:
When you specify a valueset for tiebreaker-expression, Oracle OLAP returns the ranked items in -(STATRANK) order.An optional list of one or more of the dimensions of expression to include in the ranking. When you do not specify the dimensions, then RANK bases the ranking on all of the dimensions of expression.
Note:
When the current value of a BASEDON dimension is not in ranking status, Oracle OLAP returns a rank ofNA
.Monitoring the Behavior of RANK
The OLAPO DML provides the RANK_CALLS, RANK_CELLS, and RANK_SORTS options that you can use to monitor the behavior of the RANK function.
RANK_CALLS The RANK_CALLS option is an INTEGER
, read-only option that holds the number of calls that Oracle OLAP has made to the RANK function.
RANK_CELLS The RANK_CELLS option is an INTEGER
, read-only option that holds the number of values that Oracle OLAP has computed when executing the RANK function.
RANK_SORTS The RANK_SORTS option holds the number of sorts that have been triggered by the execution of the RANK function
This note describes the results of the different methods of ranking values. The results are based on the sales2
variable, which is described in "Ranking Values", with the geography
dimension limited to G2
as the following statements demonstrate.
LIMIT geography TO 'G2' SORT items D sales2 REPORT DOWN geography sales2
The preceding statements produce the following output.
------------------------SALES2------------------------ ------------------------ITEMS------------------------- GEOGRAPHY ITEM4 ITEM2 ITEM3 ITEM1 ITEM5 -------------- ---------- ---------- ---------- ---------- ---------- G2 25.00 20.00 20.00 15.00 7.00
Table 22-2, "Results of Different Methods of Ranking" shows the results of the different methods of ranking that are produced by a statement of the form
REPORT DOWN geography RANK(sales2 MIN BASEDON items)
with the different method keywords substituted for MIN.
Table 22-2 Results of Different Methods of Ranking
Methods | (ITEM4, G2) = 25 | (ITEM2, G2) = 20 | (ITEM3, G2) = 20 | (ITEM1,G2) = 15 | (ITEM5,G2) = 7 |
---|---|---|---|---|---|
MIN |
1 |
2 |
2 |
4 |
5 |
MAX |
1 |
3 |
3 |
4 |
5 |
AVERAGE |
1 |
2.5 |
2.5 |
4 |
5 |
PACKED |
1 |
2 |
2 |
3 |
4 |
UNIQUE |
1 |
2 |
3 |
4 |
5 |
PERCENTILE |
100 |
62 |
62 |
25 |
1 |
DECILE |
10 |
7 |
7 |
3 |
1 |
QUARTILE |
4 |
3 |
3 |
1 |
1 |
Note that the value that is returned by the UNIQUE method for Item2
and Item3
can be either 2
or 3
, since the RANK function randomly assigns a unique rank for identical values in the expression.
Assume that your analytic workspace contains geography
and items
dimensions and sales2
variable.
DEFINE geography DIMENSION TEXT MAINTAIN geography ADD 'g1' 'g2' 'g3' DEFINE items DIMENSION TEXT MAINTAIN items ADD 'Item1' 'Item2' 'Item3' 'Item4' 'Item5' DEFINE sales2 DECIMAL <geography items>
Assume the sales2
variable has the following data values.
-------------SALES2------------- -----------GEOGRAPHY------------ ITEMS g1 g2 g3 -------------- ---------- ---------- ---------- Item1 30.00 15.00 12.00 Item2 10.00 20.00 18.00 Item3 15.00 20.00 24.00 Item4 30.00 25.00 25.00 Item5 NA 7.00 21.00
This statement reports the results of using the MIN method to rank the sales2
values based on the items
dimension.
report rank(sales2 min basedon items)
The preceding statement produces the following output.
-RANK(SALES2 MIN BASEDON ITEMS)- -----------GEOGRAPHY------------ ITEMS g1 g2 g3 -------------- ---------- ---------- ---------- Item1 1.00 4.00 5.00 Item2 4.00 2.00 4.00 Item3 3.00 2.00 2.00 Item4 1.00 1.00 1.00 Item5 NA 5.00 3.00
This statement reports the results of using the MIN method to rank the sales2
values based on the geography dimension.
REPORT RANK(sales2 MIN BASEDON geography)
The preceding statement produces the following output.
----RANK(SALES2 MIN BASEDON----- -----------GEOGRAPHY)----------- -----------GEOGRAPHY------------ ITEMS g1 g2 g3 -------------- ---------- ---------- ---------- Item1 1.00 2.00 3.00 Item2 3.00 1.00 2.00 Item3 3.00 2.00 1.00 Item4 1.00 2.00 2.00 Item5 NA 2.00 1.00
This statement reports the results of using the MIN method to rank the sales2
values based on all of its dimensions.
REPORT RANK(sales2, MIN)
The preceding statement produces the following output.
-------RANK(SALES2, MIN)-------- -----------GEOGRAPHY------------ ITEMS g1 g2 g3 -------------- ---------- ---------- ---------- Item1 1.00 10.00 12.00 Item2 13.00 7.00 9.00 Item3 10.00 7.00 5.00 Item4 1.00 3.00 3.00 Item5 NA 14.00 6.00