Re: Help: Flattened aggregations (with limiting and sorting)

[Matt Traynham]

Hey Adrien,

Say I have two fields in my index with values: 

genre = {Action, Adventure}
actor = {Tom Cruise, Jason Statham}

I'm looking for a way to get the distinct combinations of values with doc 
counts, so I use a sub-aggregation:
"aggs":{
"genreAgg": {
 "terms": {
 "field": "genre"
 },
 "aggs": {
 "actorAgg": {
 "terms": {
 "field": "actor"
 },
 "aggs": {
 "measureAgg": { "sum": { "field" : "docCount"} }
 }
 } 
 }
}
}

When I get the data back, I flatten it into a CSV format:
Action, Tom Cruise, 50
Adventure, Tom Cruise, 40
Action, Jason Statham, 20
Adventure, Jason Statham, 40

My question is, is there a better way to do this?  I'm not entirely worried 
about recursively flattening the data.  My point of interest is:
1. Performance - My top aggregation may not be the one with the lowest 
cardinality, can ES handle that for me?
2. Sorting & Limiting - I have to fetch all the data for these fields.  Say 
I want to "sort by actor, limit 1".  Where do you apply the sort?  It can't 
be on the genre field.  Actor's field seems logical, but I still can't 
limit the genre field at all.  Fetching all the data and then flattening 
works because I can sort correctly then limit.

I have seen that you can use script fields to return back single rows.  But 
can you sort and limit by a script field?

-- 
You received this message because you are subscribed to the Google Groups 
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to elasticsearch+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/elasticsearch/28b210de-9822-4967-a95a-c5f74f426ff9%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: Help: Flattened aggregations (with limiting and sorting)

[Adrien Grand]

Hi Matt,

I don't understand what the problem is, can you maybe try to elaborate a
bit?

Thanks.

On Fri, Oct 24, 2014 at 4:00 PM, Matt Traynham  wrote:

> Thoughts, anybody?  I saw that you can somewhat do this with "scripts" and
> letting the top aggregation encompass all term fields, but is that any more
> performant?
>
> --
> You received this message because you are subscribed to the Google Groups
> "elasticsearch" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to elasticsearch+unsubscr...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/elasticsearch/c79e42a7-bd82-4aed-9e1d-4e882566f630%40googlegroups.com
> 
> .
>
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Adrien Grand

-- 
You received this message because you are subscribed to the Google Groups 
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to elasticsearch+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/elasticsearch/CAL6Z4j65OMxJQkCxRBORW9V2%3Dy2wUNxDmbWiWKdJh_27qjsX1g%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

Re: Help: Flattened aggregations (with limiting and sorting)

[Matt Traynham]

Thoughts, anybody?  I saw that you can somewhat do this with "scripts" and 
letting the top aggregation encompass all term fields, but is that any more 
performant?

-- 
You received this message because you are subscribed to the Google Groups 
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to elasticsearch+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/elasticsearch/c79e42a7-bd82-4aed-9e1d-4e882566f630%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Help: Flattened aggregations (with limiting and sorting)

[Matt Traynham]

Here's a bit of background info:

I'm interested in using aggregations to produce distinct keys for multiple 
"term" fields and then getting a "measure" value for those keys.  This can 
be accomplished by "tree"-ing term aggregations together and whatever 
"measure" terms are applied to the lowest sub-aggregation.

For instance:
"aggs":{
"field1Agg": {
 "terms": {
 "field": "field1"
 },
 "aggs": {
 "field2Agg": {
 "terms": {
 "field": "field2"
 },
 "aggs": {
 "measure1Agg": { "sum": { "field" : "measure1"} }
 }
 } 
 }
}
}

Now, when I get this data back, I just recursively flatten the results into 
a single List.  I then apply whatever sorting and limiting after the fact.

The bad:
This actually requires me to request for every record from ElasticSearch, 
which is not ideal.

So is there a particular way to accomplish the sorting/limiting on 
ElasticSearch rather than after I flatten the data?  I saw the "top_hits" 
aggregation, but I'm not sure how it applies...

-- 
You received this message because you are subscribed to the Google Groups 
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to elasticsearch+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/elasticsearch/e8eafbce-3636-4532-847b-409730241a8f%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.