查询以查找hiveQL中不包括前5％的行的平均值

Question

select perecentile(time,0.95) from sometable;
 gives the 95th percentile.

I want average of all the rows whose time values are below this value.

In oracle it would be something like this:-

 select avg(time) from sometable
 where 
 time<(select percentile(time,0.95) from sometable);

But in hive it is not possible to use subqueries in the where clause.When i am using union all I am not able to isolate the tuple that i need to compare the other tuples with.

Answer 1

you can do a Cartesian join with the result of the percentile and then filter all the smaller values.

Something like this :

 select avg(time) from sometable a  
 join (select percentile(time,0.95) perc from sometable) b on (1=1)   
 where a.time < b.perc;

Its not the most efficient way but that's the first that comes in mind..