如何在 Looker 中按日期计算状态变化？

Question

我在 BigQuery 中有一个任务，其中包含创建日期和上次修改日期。如果可能的话，我希望能够在同一个 table 中按日期报告任务打开和任务关闭事件的数量。

view: tasks {
  derived_table: {
    sql:
        SELECT *
        FROM UNNEST(ARRAY<STRUCT<CREATED_DATE DATE, LAST_MODIFIED DATE, ID INT64, STATE STRING>>[
        ('2020-12-01', '2020-12-01', 1, "OPEN"),
        ('2020-12-01', '2020-12-03', 2, "CLOSED"),
        ('2020-12-02', '2020-12-03', 3, "CLOSED"),
        ('2020-12-03', '2020-12-05', 4, "OPEN"),
        ('2020-12-05', '2020-12-05', 5, "CLOSED")])
      ;;
  }
  dimension_group: created {
    type: time
    datatype: date
    sql: ${TABLE}.created_date ;;
  }
  dimension_group: last_modified {
    type: time
    datatype: date
    sql: ${TABLE}.last_modified ;;
  }
  dimension: id {
    type: number
  }
  dimension: state {
    type: string
  }
  measure: number_of_tasks {
    type: count_distinct
    sql: ${id} ;;
  }
  measure: number_of_open_tasks {
    type: count_distinct
    sql: ${id} ;;
    filters: {
      field: "state"
      value: "OPEN"
    }
  }
  measure: number_of_closed_tasks {
    type: count_distinct
    sql: ${id} ;;
    filters: {
      field: "state"
      value: "CLOSED"
    }
  }
}
explore: tasks {}

我可以使用创建日期获取打开的任务数。

我可以通过对任务进行计数来获取已关闭的任务数，其中最后修改日期在聚合期间并且状态为已关闭，并具有过滤措施。

但是，如果我尝试将这些组合成一个 table，我会为每个日期组合得到一行。

如何按日期统计任务状态变化？

Date	Number of Opened Tasks	Number of Closed Tasks
2020-12-01	2	0
2020-12-02	1	0
2020-12-03	1	2
2020-12-04	0	0
2020-12-05	1	1

Answer 1

一位同事提出了一个解决方案。堆叠任务 table 本身会为每个任务创建（最多）两行。

view: tasks {
  derived_table: {
    sql:
        WITH tab AS (
          SELECT *
          FROM UNNEST(ARRAY<STRUCT<CREATED_DATE DATE, LAST_MODIFIED DATE, ID INT64, STATE STRING>>[
          ('2020-12-01', '2020-12-01', 1, "OPEN"),
          ('2020-12-01', '2020-12-03', 2, "CLOSED"),
          ('2020-12-02', '2020-12-03', 3, "CLOSED"),
          ('2020-12-03', '2020-12-05', 4, "OPEN"),
          ('2020-12-05', '2020-12-05', 5, "CLOSED")])
        )
        SELECT *, 1 open_count, 0 closed_count, created_date AS action_date 
        FROM tab
        UNION DISTINCT
        SELECT *, 0 open_count, 1 closed_count, last_modified AS action_date 
        FROM tab
        WHERE state = "CLOSED"
      ;;
  }
  dimension_group: created {
    type: time
    datatype: date
    sql: ${TABLE}.created_date ;;
  }
  dimension_group: last_modified {
    type: time
    datatype: date
    sql: ${TABLE}.last_modified ;;
  }
  dimension_group: action {
    type: time
    datatype: date
    sql: ${TABLE}.action_date ;;
  }
  dimension: id {
    type: number
  }
  dimension: state {
    type: string
  }
  dimension: open_count {
    type: number
    hidden: yes
  }
  dimension: closed_count {
    type: number
    hidden: yes
  }
  measure: number_opened{
    type: sum
    sql: ${open_count} ;;
  }
  measure: number_closed {
    type: sum
    sql: ${closed_count} ;;
  }
}
explore: tasks {}

然后可以计算打开和关闭的标签。

如何在 Looker 中按日期计算状态变化？

How to count state changes by date in Looker?

sql

business-intelligence

looker