Elasticsearch带权重的指标聚合方式 weighted_avg

聚合权重求平均：weighted_avg

关于聚合有权重时，求平均的方式，总结了以下几点：
1、首先确定平均值字段和权重字段，字段都为数值类型
2、权重的字段
​ ​ ​ ​ ​ ​如果权重都相等，则和常规求平均结果一致
​ ​ ​ ​ ​ 如果权重都为0，则平均值为null
​ ​ ​ ​ ​ ​如果权重值不同，可以根据下面的公式进行计划​ ​

废话少说，直接上代码，验证下：

1、先创建Index

//创建Index
PUT /demo_avg_test
{
  "mappings": {
    "properties": {
      "lesson":{
        "type": "text"
      },
      "scores": {
        "type": "double"
      },
      "weighttest": {
        "type": "integer"
      }
    }
  }
}

2、导入文档有两种形式：单个或批量

1. 单个导入

//单个文档导入
PUT /demo_avg_test/_doc/1
{
  "id":1,
  "lesson":"语言",
  "scores":10,
  "weighttest":20
}
PUT /demo_avg_test/_doc/2
{
  "id":2,
  "lesson":"历史",
  "scores":40,
  "weighttest":100
}

PUT /demo_avg_test/_doc/3
{
  "id":3,
  "lesson":"数学",
  "scores":50,
  "weighttest":100
}

2.批量导入文档

//批量导入文档
PUT /_bulk
{"index":{"_index":"demo_avg_test","_id":1}}
{"lesson":"语言","scores":10,"weighttest":20}
{"index":{"_index":"demo_avg_test","_id":2}}
{"lesson":"历史","scores":20,"weighttest":40}
{"index":{"_index":"demo_avg_test","_id":3}}
{"lesson":"数学","scores":50,"weighttest":100}

3、查询全部：验证下数据

//查询全部
GET /demo_avg_test/_search
{
  "query": {"match_all": {}}
}

4、聚合权重求平均值

聚合权重平均查询

#聚合权重平均查询
GET /demo_avg_test/_search
{
  "size": 3,
  "aggs": {
    "weighted_type": {
      "weighted_avg": {
        "value": {
          "field": "scores"
        },
        "weight": {
          "field": "weighttest"
        }
      }
    }
  }
}

说明：
－求平均值字段：scores
－权重字段：weighttest

平均分计算规则：
加权平均值：∑(值*权重) / ∑(权重)

公式：
$$加权平均值为\sum (值\ast 权重)/\sum (权重)$$

1）权重值不等时

#结果1
{
  "took" : 281,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 3,
      "relation" : "eq"
    },
    "max_score" : 1.0,
    "hits" : [
      {
        "_index" : "demo_avg_test",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 1.0,
        "_source" : {
          "lesson" : "语言",
          "scores" : 10,
          "weighttest" : 20
        }
      },
      {
        "_index" : "demo_avg_test",
        "_type" : "_doc",
        "_id" : "2",
        "_score" : 1.0,
        "_source" : {
          "lesson" : "历史",
          "scores" : 20,
          "weighttest" : 40
        }
      },
      {
        "_index" : "demo_avg_test",
        "_type" : "_doc",
        "_id" : "3",
        "_score" : 1.0,
        "_source" : {
          "lesson" : "数学",
          "scores" : 50,
          "weighttest" : 100
        }
      }
    ]
  },
  "aggregations" : {
    "weighted_type" : {
      "value" : 37.5
    }
  }
}

平均分数：37.5
1020+2040+50*100/(20+40+100)=37.5

2）权重值为0，则平均值为null

#结果2

{
  "took" : 719,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 3,
      "relation" : "eq"
    },
    "max_score" : 1.0,
    "hits" : [
      {
        "_index" : "demo_avg_test",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 1.0,
        "_source" : {
          "lesson" : "语言",
          "scores" : 10,
          "weighttest" : 0
        }
      },
      {
        "_index" : "demo_avg_test",
        "_type" : "_doc",
        "_id" : "2",
        "_score" : 1.0,
        "_source" : {
          "lesson" : "历史",
          "scores" : 20,
          "weighttest" : 0
        }
      },
      {
        "_index" : "demo_avg_test",
        "_type" : "_doc",
        "_id" : "3",
        "_score" : 1.0,
        "_source" : {
          "lesson" : "数学",
          "scores" : 50,
          "weighttest" : 0
        }
      }
    ]
  },
  "aggregations" : {
    "weighted_type" : {
      "value" : null
    }
  }
}

平均分数：null

3）权重值都相等时

#结果3
#如果将权重值都修改为相同值，则权重如常规变量1一样
下面将权重修改为100，其平均值为：(10+20+50)/3=26.6666666
{
  "took" : 296,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 3,
      "relation" : "eq"
    },
    "max_score" : 1.0,
    "hits" : [
      {
        "_index" : "demo_avg_test",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 1.0,
        "_source" : {
          "lesson" : "语言",
          "scores" : 10,
          "weighttest" : 100
        }
      },
      {
        "_index" : "demo_avg_test",
        "_type" : "_doc",
        "_id" : "2",
        "_score" : 1.0,
        "_source" : {
          "lesson" : "历史",
          "scores" : 20,
          "weighttest" : 100
        }
      },
      {
        "_index" : "demo_avg_test",
        "_type" : "_doc",
        "_id" : "3",
        "_score" : 1.0,
        "_source" : {
          "lesson" : "数学",
          "scores" : 50,
          "weighttest" : 100
        }
      }
    ]
  },
  "aggregations" : {
    "weighted_type" : {
      "value" : 26.666666666666668
    }
  }
}

平均值：26.6
下面将权重修改为100，其平均值为：(10+20+50)/3=26.6

到此结束了，关于聚合的权重求平均，还有脚本的方式，以及参数缺省的情况还有多种情况，后面再介绍了。

关于聚合其实官网提供了多种指标聚合方式，求最大、最小、平均、汇总、折叠等
今天在看文档时，看到这个挺意思，自测了下，作个记录。

官网链接：7.6.2.

加油！日拱一卒无有尽，功不唐捐终入海！