Sum aggregation on Unique Data in ElasticSearch

First, You can defind index mapping which have field amt as float type. like below:

{
  "mappings": {
    "properties": {
      "amt":{
        "type": "float"
      },
      "recName":{
        "type": "keyword"
      }
    }
  }
}

After, You can use aggregation for your expected result as below:

{
  "size": 0,
  "aggs": {
    "RecName": {
      "terms": {
        "field": "recName",
        "size": 10,
        "order": {
          "_key": "asc"
        }
      },
      "aggs": {
        "amount": {
          "avg": {
            "field": "amt"
          }
        }
      }
    },
    "unique_count": {
      "cardinality": {
        "field": "recName"
      }
    },
    "sum_total":{
      "sum_bucket": {
        "buckets_path": "RecName>amount"
      }
    }
  }
}

• First use Term Aggregation on recName field which will give unique name.
• Second use Avg Aggregation on amt field (inner aggregation insideRecName to just get amout per item).
• Third, use Cardinality Aggregation on recName field for getting distinct values.
• Forth, use Sum Bucket Aggregation to sum on amt field.

Response

"aggregations" : {
    "RecName" : {
      "doc_count_error_upper_bound" : 0,
      "sum_other_doc_count" : 0,
      "buckets" : [
        {
          "key" : "Item1",
          "doc_count" : 1,
          "amount" : {
            "value" : 1.7999999523162842
          }
        },
        {
          "key" : "Item2",
          "doc_count" : 2,
          "amount" : {
            "value" : 2.9000000953674316
          }
        }
      ]
    },
    "unique_count" : {
      "value" : 2
    },
    "sum_total" : {
      "value" : 4.700000047683716
    }
  }