how to use distinctCount and group by together

[sooch...@gmail.com]

 I need use distinctCount and group by together in senseidb.

Such as,  select distinctCount(a) group by b ;

So,  I want to ask,  how to use?

[Volodymyr Zhabiuk]

As far as I know this has not been implemented yet. Each aggregation function in Sensei is implemented as the Sensei Map/Reduce function http://senseidb.github.io/sensei/map-reduce.html. The problem is that distinctCount with group by would consume too much memory to store intermediate results on heap if cardinalities of b and a are big. 

In future we might leverage HyperLogLog or BloomFilter to do this more effectively, but results might not be accurate

Thanks,

Volodymyr

2013/11/13 <sooch...@gmail.com>

 I need use distinctCount and group by together in senseidb.

Such as,  select distinctCount(a) group by b ;

So,  I want to ask,  how to use?

--
You received this message because you are subscribed to the Google Groups "Sensei" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sensei-searc...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[sooch...@gmail.com]

Thanks,

But now, have a other method to replace this.

在 2013年11月14日星期四UTC+8上午1时39分28秒，Volodymyr Zhabiuk写道：