Druid聚合函数

Question

我正在使用德鲁伊创建一个用于生成报告的UI。对于脚本，我使用以下代码：

        {
      "type" : "doubleSum",
      "name" : "impressions",
      "fieldName" : "impressions"
    },
    {
      "type" : "doubleSum",
      "name" : "clicks",
      "fieldName" : "clicks"
    },
    {
      "type" : "doubleSum",
      "name" : "pvconversions",
      "fieldName" : "pvconversions"
    },
    {
      "type" : "doubleSum",
      "name" : "pcconversions",
      "fieldName" : "pcconversions"
    }

我还需要两个字段：

Total Conversions = pvconversions+pcconversions
CTR = Clicks / Impressions

关于这件事，我找不到任何关于如何写的东西。有人能帮上忙吗。

谢谢

Answer 1

您的问题可以使用聚合和postAggregations来解决，如下所示：

{
  "queryType":"timeseries",
  "dataSource":"data",
  "granularity":"hour",
  "descending":"false",
  "aggregations":[
    {"type":"doubleSum", "name":"sum-pvconversions", "fieldName":"pvconversions"},
    {"type":"doubleSum", "name":"sum-pcconversions", "fieldName":"pcconversions"},
    {"type":"doubleSum", "name":"sum-clicks", "fieldName":"clicks"},
    {"type":"doubleSum", "name":"sum-impressions", "fieldName":"impressions"}
    ],
  "postAggregations":[
    {
      "type":"arithmetic",
      "name":"Conversions",
      "fn":"+",
      "fields":[
        {"type":"fieldAccess", "name":"postAgg-proceed", "fieldName":"sum-pvconversions"},
        {"type":"fieldAccess", "name":"postAgg-numbers", "fieldName":"sum-pcconversions"}
        ]
    },
    {
      "type":"arithmetic",
      "name":"CTR",
      "fn":"/",
      "fields":[
        {"type":"fieldAccess", "name":"postAgg-click", "fieldName":"sum-clicks"},
        {"type":"fieldAccess", "name":"postAgg-impression", "fieldName":"sum-impressions"}
        ]
    }
    ],
  "intervals":["2016-08-22T01/2016-08-29T03"],
  "context":{
    "skipEmptyBuckets":"true"
  }
}

Answer 2

您可以通过在timeseries查询中使用聚合来完成此操作。这不是你要找的吗？

Answer 3

您必须在查询中使用post聚合。来自德鲁伊的文献

后聚合是处理的规范，当它们从Druid中出来时，应该发生在聚合值上。如果将post聚合作为查询的一部分，请确保包含后聚合程序所需的所有聚合器。

例如，要计算CTR，以下是post聚合：

"postAggregations" : [{
     "type"   : "arithmetic",
     "name"   : "average",
     "fn"     : "*",
    "fields" : [
       { "type"   : "arithmetic",
         "name"   : "CTR",
         "fn"     : "/",
         "fields" : [
           { "type" : "fieldAccess", "name" : "clicks", "fieldName" : "clicks" },
           { "type" : "fieldAccess", "name" : "impressions", "fieldName" : "impressions" }
         ]
       }