Remove buckets from being returned in Aggregation

Elastic Stack Elasticsearch
1

Is there a way to prevent the buckets array from being returned as a part of the aggregation result? I am using a sum_bucket to total all of the buckets and I don't need all of the buckets to be returned. I'm just using key and displayCount. I worry that with my dataset, there could be hundreds of buckets returned for hundreds of keys.

Meaning, if my default result is below:

{
  "aggregations" : {
    "Sports" : {
      "doc_count_error_upper_bound" : 0,
      "sum_other_doc_count" : 0,
      "buckets" : [
        {
          "key" : "Football",
          "doc_count" : 5,
          "States" : {
            "doc_count_error_upper_bound" : 0,
            "sum_other_doc_count" : 0,
            "buckets" : [
              {
                "key" : "Maryland",
                "doc_count" : 1,
                "countStates" : {
                  "value" : 1
                }
              },
              {
                "key" : "New York",
                "doc_count" : 1,
                "countStates" : {
                  "value" : 1
                }
              },
              {
                "key" : "Ohio",
                "doc_count" : 1,
                "countStates" : {
                  "value" : 1
                }
              },
              {
                "key" : "Pennsylvania",
                "doc_count" : 1,
                "countStates" : {
                  "value" : 1
                }
              },
              {
                "key" : "Virginia",
                "doc_count" : 1,
                "countStates" : {
                  "value" : 1
                }
              }
            ]
          },
          "displayCount" : {
            "value" : 5.0
          }
        },
        {
          "key" : "Baseball",
          "doc_count" : 3,
          "States" : {
            "doc_count_error_upper_bound" : 0,
            "sum_other_doc_count" : 0,
            "buckets" : [
              {
                "key" : "Maryland",
                "doc_count" : 1,
                "countStates" : {
                  "value" : 1
                }
              },
              {
                "key" : "Ohio",
                "doc_count" : 1,
                "countStates" : {
                  "value" : 1
                }
              },
              {
                "key" : "Pennsylvania",
                "doc_count" : 1,
                "countStates" : {
                  "value" : 1
                }
              }
            ]
          },
          "displayCount" : {
            "value" : 3.0
          }
        }
	}
}

Can I remove the buckets array within the fields to only return the following:

{
  "aggregations" : {
    "Sports" : {
      "doc_count_error_upper_bound" : 0,
      "sum_other_doc_count" : 0,
      "buckets" : [
        {
          "key" : "Football",
          "doc_count" : 5,
          "States" : {
            "doc_count_error_upper_bound" : 0,
            "sum_other_doc_count" : 0
          }
          "displayCount" : {
            "value" : 5.0
          }
        },
        {
          "key" : "Baseball",
          "doc_count" : 3,
          "States" : {
            "doc_count_error_upper_bound" : 0,
            "sum_other_doc_count" : 0,
          },
          "displayCount" : {
            "value" : 3.0
          }
        }
  }
}
2

I'll add that I am aware I could use filter_paths list to get what I want, but that feels like overkill for removing just one array from the entire result set. Also I am using the Java API, which filter_paths is not available.

3

Hi Eric,
I'm not sure what produces the displayCount in your response?

I suspect you may be interested in the cardinality aggregation

4

I'm not familiar enough with cardinality to know if it work with my use case, but I am more than happen to try if it does. Essentially I am creating a faceted search and I want my categories to display a count of potential results if that facet is selected. I can accomplish this with an aggs query in my aggregation like shown below.

Can cardinality do something similar? I am working with around 2 million documents so I worry if the solution is to run a script that performance would suffer.

{
  "query": {
    "match_all": {}
  },
  "post_filter": {
    "bool": {
      "must": [
        {
          "terms": {
            "sport.name.keyword": [
              "Football"
            ]
          }
        }
      ]
    }
  },
  "aggregations": {
    "States": {
      "terms": {
        "field": "state.name.keyword"
      },
      "aggs": {
        "Sports": {
          "terms": {
            "field": "sport.name.keyword",
            "include": [
              "Football"
            ],
            "size": 10000
          },
          "aggs": {
            "countSports": {
              "value_count": {
                "field": "_id"
              }
            }
          }
        },
        "displayCount": {
          "sum_bucket": {
            "buckets_path": "Sports>countSports"
          }
        }
      }
    }
  }
}
5

When you swap a terms agg for a cardinality agg you’re effectively asking for a count of how many terms buckets would match without seeing the string values of any of them.