[ https://issues.apache.org/jira/browse/HIVE-18711?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16363357#comment-16363357 ]
Ashutosh Chauhan commented on HIVE-18711: ----------------------------------------- create table employees (name string, department_id tinyint, salary int); SELECT name, salary, department_id, PERCENTILE_CONT(0.5) OVER (PARTITION BY department_id ORDER BY salary DESC) FROM employees; > Add percentile_cont and percentile_disc udaf > -------------------------------------------- > > Key: HIVE-18711 > URL: https://issues.apache.org/jira/browse/HIVE-18711 > Project: Hive > Issue Type: Bug > Components: UDF > Reporter: Ashutosh Chauhan > Priority: Major > > Most common way to implement this is via ordered aggregate which allows users > to specify sort specification with group by clause. Some implementations also > allow to use these with window functions. > Since Hive doesn't have concept of ordered aggregates yet, one possibility is > to support these only for window functions where sort specification is also > taken from window clause. -- This message was sent by Atlassian JIRA (v7.6.3#76005)