常见问题解答

在本页中,您将找到与此特定ML模型相关的常见问题的解答。

此用例背后的架构是什么?

DSW

涉及哪些数据集?

该模型依赖于两个数据集:

  • AEP Demo - Car Insurance Interactions的问题。这是培训和评分输入数据集。 所有客户行为( 如Get Quote和Purchase Insurance - 事件)都存储在此数据集中。

  • AEP Demo - ML Predictions的问题。这是评分输出数据集。 当ML模型计算倾向得分时,它会将该得分存储在此数据集中。

培训数据的外观是什么?

当客户单击“获 取报价 ”按钮时,会收集以下数据:

{
  "header": {
    "datasetId": "5e18200afe31e818a8d616b6",
    "imsOrgId": "907075E95BF479EC0A495C73@AdobeOrg",
    "source": {
      "name": "vangeluw Launch 2"
    },
    "schemaRef": {
      "id": "https://ns.adobe.com/experienceplatform/schemas/cc85bf9611e1df1c6ef9d0a237c3e9ec",
      "contentType": "application/vnd.adobe.xed-full+json;version=1"
    }
  },
  "body": {
    "xdmMeta": {
      "schemaRef": {
        "id": "https://ns.adobe.com/experienceplatform/schemas/cc85bf9611e1df1c6ef9d0a237c3e9ec",
        "contentType": "application/vnd.adobe.xed-full+json;version=1"
      }
    },
    "xdmEntity": {
      "_id": "8194664889569.993",
      "web": {
        "webPageDetails": {
          "URL": "https://platformdemo.net/fsi_carinsurance.html",
          "name": "Car Insurance Simulator"
        }
      },
      "eventType": "carInsuranceGetQuote",
      "timestamp": "2020-03-12T22:55:26Z",
      "--aepTenantId--": {
        "carinsurance": {
          "insuranceKm": "+50000",
          "insuranceAge": "37",
          "insuranceCity": "Brussels",
          "insuranceGender": "female",
          "insuranceCarType": "hatchback",
          "insuranceCountry": "belgium",
          "insuranceLeasing": "yes",
          "insuranceCarBrand": "alfaromeo",
          "insuranceGetQuote": 1,
          "insuranceFullPrice": "1025",
          "insuranceBasicPrice": "574",
          "insuranceMediumPrice": "820",
          "insuranceNationality": "belgium",
          "insurancePrimaryDriver": "yes"
        },
        "identification": {
          "ecid": "39948201868485073490150335797615706955"
        }
      }
    }
  }
}

客户购买汽车保险时,会收集以下数据:

{
  "header": {
    "datasetId": "5e18200afe31e818a8d616b6",
    "imsOrgId": "907075E95BF479EC0A495C73@AdobeOrg",
    "source": {
      "name": "vangeluw Launch 2"
    },
    "schemaRef": {
      "id": "https://ns.adobe.com/experienceplatform/schemas/cc85bf9611e1df1c6ef9d0a237c3e9ec",
      "contentType": "application/vnd.adobe.xed-full+json;version=1"
    }
  },
  "body": {
    "xdmMeta": {
      "schemaRef": {
        "id": "https://ns.adobe.com/experienceplatform/schemas/cc85bf9611e1df1c6ef9d0a237c3e9ec",
        "contentType": "application/vnd.adobe.xed-full+json;version=1"
      }
    },
    "xdmEntity": {
      "_id": "6155761227236.238",
      "web": {
        "webPageDetails": {
          "URL": "https://platformdemo.net/fsi_carinsurance.html",
          "name": "Car Insurance Simulator"
        }
      },
      "eventType": "carInsurancePurchase",
      "timestamp": "2020-03-12T22:56:30Z",
      "--aepTenantId--": {
        "carinsurance": {
          "insurancePurchase": "yes"
        },
        "identification": {
          "ecid": "39948201868485073490150335797615706955"
        }
      }
    }
  }
}

如何创建培训数据集?

根据网站上的客户互动实时填充培训数据集。 数据集配置为以流方式收集数据并激活以进行用户档案。
如以前的问题所示,数据集会将获取报 价事件​作为 购买​事件。
数据通过Launch和在网站上收集 alloy.js,并实时将数据流化到Adobe Experience Platform。

什么是目标预测变量?

目标预测变量是以下字段。 这是表示所需行为(即购买)的值。

"_experienceplatform": {
 "carinsurance": {
  "insurancePurchase": "yes"
},

流量如何从Pipeline转发到RTML端点?

在此演示环境中,启动服务器端转发用于将数据从管道转发到RTML端点。 此功能目前为测试版,将于今年晚些时候推出。

延迟是什么?

从单击“获取报价”按钮到 Adobe Target (根据RTML-endpoint计算的得分)提供体验的端到端延迟为1-1.5秒。

RTML端点的输出如何发送回平台?

鉴于这是自定义实现,将得分输出发回Adobe Experience Platform的方法是将XDM有效负荷发送到HTTP API端点。

我们正在重复使用通过Launch中的平台扩展创建的现有DCS Inlet ID。

RTML写回数据集, EMEA ML Predictions (API) 该数据集是评分输出数据集。 当ML模型计算倾向得分时,它通过向DCS Inlet ID发送XDM有效负荷来将此得分存储在此数据集中。

以下是RTML端点发送到DCS Inlet ID的XDM有效负荷的示例:

{
  "header": {
    "datasetId": "5d918f445dad97163733422e",
    "imsOrgId": "907075E95BF479EC0A495C73@AdobeOrg",
    "source": {
      "name": "vangeluw Launch 2"
    },
    "schemaRef": {
      "id": "https://ns.adobe.com/experienceplatform/schemas/8d8773d1cffdd8ed8de7f6d42778a527",
      "contentType": "application/vnd.adobe.xed-full+json;version=1"
    }
  },
  "body": {
    "xdmMeta": {
      "schemaRef": {
        "id": "https://ns.adobe.com/experienceplatform/schemas/8d8773d1cffdd8ed8de7f6d42778a527",
        "contentType": "application/vnd.adobe.xed-full+json;version=1"
      }
    },
    "xdmEntity": {
      "_id": "6749627040993.491",
      "_experienceplatform": {
        "identification": {
          "ecid": "50684336701781569260739091385360831781"
        },
        "salesPrediction": {
          "carInsuranceSalesPrediction": 82
        }
      }
    }
  }
}

返回模块15

返回到所有模块

在此页面上