elasticsearch - Elasticsearch/Kibana 中两个聚合列的计算列？

Question

假设我有一个简单的聚合，如下所示:

| Item | Views | Clicks |
+------+-------+--------+
| B    | 123   | 34     |
| A    | 234   | 12     |
| C    | 56    | 2      |

是否可以生成一个额外的列作为 Views 和 Clicks 列的计算结果？例如:

| Item | Views | Clicks | Ratio |
+------+-------+--------+-------+
| B    | 123   | 34     | 0.28  |
| A    | 234   | 12     | 0.05  |
| C    | 56    | 2      | 0.03  |

如果能够在此附加列上进行排序也是理想的选择。

有没有办法通过Elasticsearch(和Kibana)实现这种聚合？

Answer 1

您可以为此使用带有脚本支持的管道聚合。缺点是您必须启用内联脚本或将脚本放在文件中。要启用对内联脚本的支持，请将其添加到您的 yaml 中。

script.inline: on

下面是一个例子:

POST /items/item/_bulk
{ "index": { "_id": "1" }}
{"name": "A","views": 2,"clicks": 3}
{ "index": { "_id": "2" }}
{"name": "A","views": 1,"clicks": 9}
{ "index": { "_id": "3" }}
{"name": "B","views": 5,"clicks": 15}
{ "index": { "_id": "4" }}
{"name": "C","views": 1,"clicks": 1}
{ "index": { "_id": "5" }}
{"name": "A","views": 3,"clicks": 23}

GET /items/item/_search
{
  "size": 0, 
  "aggs": {
    "byName": {
      "terms": {
        "field": "name",
        "size": 10
      },
      "aggs": {
        "countClicks": {
          "sum": {
            "field": "clicks"
          }
        },
        "countViews": {
          "sum": {
            "field": "views"
          }
        },
        "computeRatio": {
          "bucket_script": {
            "buckets_path": {
              "the_clicks": "countClicks",
              "the_views": "countViews"
            },
            "script": "the_clicks / the_views"
          }
        }
      }
    }
  }
}