PERCENTILE_CONT
Purpose
PERCENTILE_CONT
is an inverse distribution function and expects as an input parameter a percentile value and a sorting specification which defines the rank of each element within a group. The functions return the percentile of the sort order as per the following example - in the case of percentile 0.7 and 100 values, the 70th value is returned.
If the percentile cannot be assigned exactly to an element, then the linear interpolation between the two nearest values is returned. For example, in the case of percentile 0.71 and ten values, the interpolation between the 7th and 8th value.
Syntax
percentile_cont::=
Usage Notes
NULL
values are ignored for the computation.- The
expr
must be anumeric
value between 0 and 1. Theexpr
must be constant within a partition. This implies thatexpr
may depend on the partitioning attribute, so that it may vary in different partitions but must remain constant for all rows within a partition. Exasol returns an error when this condition is not satisfied. - The
order_clause
is necessary forPERCENTILE_CONT
function. - The
over_clause
must not contain anorder_clause
and must not contain awindow_frame_clause
. - For information about the
over_clause
and analytic functions in general, refer to the Analytic Functions section. - For additional information, also refer to the inverse distribution functions PERCENTILE_DISC and MEDIAN.
Example
Analytic Function
SELECT
id, department, current_salary,
PERCENTILE_CONT(0.7) WITHIN GROUP (ORDER BY current_salary)
OVER (PARTITION BY department) PERCENTILE_CONT
FROM employee_table ORDER BY department, current_salary;
Result
ID | DEPARTMENT | CURRENT_SALARY | PERCENTILE_CONT |
2001 | ACCOUNTS | 55000 | 78000 |
2002 | ACCOUNTS | 65000 | 78000 |
2004 | ACCOUNTS | 70000 | 78000 |
2003 | ACCOUNTS | 80000 | 78000 |
2005 | ACCOUNTS | 80000 | 78000 |
1001 | HR | 55000 | 72000 |
1002 | HR | 70000 | 72000 |
1004 | HR | 70000 | 72000 |
1003 | HR | 90000 | 72000 |