using doc_count from one aggregation into another aggregation to calculate additional percentage

[rajeshetty]

Lets say I have doc like this

      {
       user_id : "12312321321"
       location_shared_with { [ ]},
       location {location_name:.... 
                      location_verification[{} ]
                } 
       }

In here location_verification is an array of all the friends who have verified the location and it will always be equal or less than location_shared_with array.

mapping

    { 
        "user" : 
            {"properties" : 
                 { 
                    "locations":
                    {
                        "type" : "nested",
                        "properties" : 
                        {
                            "location_name": { "type": "string", "index": "not_analyzed"},
                            "location_verification":
                            {
                                "type" : "nested",
                                "properties" : 
                                {
                                    "rating": { "type": "short", "store":"true" }
                                }
                            }
                        }
                    },
                    "location_shared_with":
                    {
                        "type" : "nested",
                        "properties" : 
                        {
                            "status": { "type": "string", "store":"true", "index": "analyzed"},
                            "friend_id": { "type": "string", "store" : "true"}
                        }
                    }
                 }
            }
    }'

So I need to calculate the % verified per location array . I have created right "nested" mapping between main doc (user) and location , location and location_verification. I have terms aggregation that tells me

when I execute

<pre>
GET _search?search_type=count
{
  "query": {
    "bool": {
      "must": [
        {
          "term": {
            "user_id": "1"
          }
        }
      ]
    }
  },  
  "aggs": {
    "location_shared_with_count":{
      "nested": {
        "path": "location_shared_with"
      }
     },
    "locations_count": {
      "nested": {
        "path": "locations"
      },
      "aggs": {
        "locations_aggs": {
          "terms": {
            "field": "locations.location_name"
          },
          "aggs": {
              "location_verification_count": {
                "nested": {
                  "path": "locations.location_verification"
                }

              }
            }          
        }
      }
    }
  },   
    "size": 10
} 

i do get following

    {
       "took": 2,
       "timed_out": false,
       "_shards": {
          "total": 5,
          "successful": 5,
          "failed": 0
       },
       "hits": {
          "total": 1,
          "max_score": 0,
          "hits": []
       },
       "aggregations": {
          "location_shared_with_count": {
             "doc_count": 6
          },
          "locations_count": {
             "doc_count": 2,
             "locations_aggs": {
                "buckets": [
                   {
                      "key": "Boston",
                      "doc_count": 1,
                      "location_verification_count": {
                         "doc_count": 2
                      }
                   },
                   {
                      "key": "Miami",
                      "doc_count": 1,
                      "location_verification_count": {
                         "doc_count": 0
                      }
                   }
                ]
             }
          }
       }
    }

• Location verification count (doc_count) per location - perfect . So I do get e.g. "Boston" location was verified by 2 friends (doc_count) .

What I want to do is calculate % verified per location .
Formula = [ (location_verification_count -> doc_count / location_shared_with_count->doc_count) x 100 ].

Question is

• How can I get or use existing doc_count in another sub-aggregation to calculate above %
• What kind of aggregation type I would use to perform above ?

[clintongormley]

Hi @rajeshetty

Please ask questions like these in the mailing list, rather than on the github issues. That said, you would have to do this calculation application side today. However when #8110 is merged, you should be able to do things like this.

[rajeshetty]

when is 8110 going into main release ?. what release would that be ?

[jpountz]

@rajeshetty It is scheduled for elaticsearch 2.0 (see tags of the issue), which I don't see released before at least several months.

[rajeshetty]

Appreciate your quick response on this. Which mailing list you recommend asking these kind of questions.? - Stackoverflow or somewhere else?.

You see your answer about new reducers was very helpful even though i can't use it anywhere in coming months till 2.0 is out, but still knowing that ES is working on it is useful , So which forum would give me such a quick turn around on my queries around questions I might have ?

[clintongormley]

@rajeshetty http://www.elasticsearch.org/community