SELECT <aggregate_function(column_name)> KEEP
(DENSE_RANK FIRST ORDER BY <column_name> [<ASC|DESC> NULLS <FIRST|LAST>)
OVER (PARTITION BY <column_name>)
FROM <table_name>
GROUP BY <column_name>;
 
conn oe/oe
 
SELECT last_name, department_id, salary,
MIN(salary) KEEP (DENSE_RANK FIRST ORDER BY commission_pct)
OVER (PARTITION BY department_id) "Worst",
MAX(salary) KEEP (DENSE_RANK LAST ORDER BY commission_pct)
OVER (PARTITION BY department_id) "Best"
FROM employees
WHERE department_id IN (30, 60)
ORDER BY department_id, salary;