Bigquery - Select 一列未将它们分组在按子句中

Question

我有基于 device_category(desktop/mobile/tablet) 和 user_type(new user/returning user) 拆分的谷歌分析数据的日表。

我的要求是，查询当月表现最好的产品，只知道设备和用户的类型。 我不想根据 device_category、user_type 对它们进行分组。

当从我的查询中排除它们时会出现错误提示 - “查询错误：SELECT 列表表达式引用列 device_category 既不在 [3:21] 分组也不聚合”

不工作的查询（这是我的要求）

 SELECT
  month, 
  year, 
  device_category, 
  user_type, 
  product_name, 
  round(sum(item_revenue),2) as item_revenue 
FROM 
  `ProjectName.DatasetName.GA_REPORT_3_*` 
where 
  _table_suffix between '20201101' and '20210131' 
  and channel_grouping = 'Organic Search' 
group by 
  month, 
  year, 
  channel_grouping, 
  product_name 
order by 
  item_revenue desc;

有效的查询

SELECT 
  month, 
  year, 
  device_category, 
  user_type, 
  product_name, 
  round(sum(item_revenue),2) as item_revenue 
FROM 
  `ProjectName.DatasetName.GA_REPORT_3_*` 
where 
  _table_suffix between '20201101' and '20210131' 
  and channel_grouping = 'Organic Search' 
group by 
  month, 
  year, 
  channel_grouping, 
  product_name, 
  device_category, 
  user_type 
order by 
  item_revenue desc;

样本数据

我知道在常规的 SQL 工作台中，我们可以 select 列在 SQL 中，而不是在 Group By 子句中。

你能帮我解决这个问题吗？

Answer 1

从技术上讲，您可以使用ANY_VALUE或MAX或MIN封装device_category和user_type ：

 SELECT
  month, 
  year, 
  ANY_VALUE(device_category), 
  ANY_VALUE(user_type), 
  product_name, 
  round(sum(item_revenue),2) as item_revenue 
FROM 
  `ProjectName.DatasetName.GA_REPORT_3_*` 
where 
  _table_suffix between '20201101' and '20210131' 
  and channel_grouping = 'Organic Search' 
group by 
  month, 
  year, 
  channel_grouping, 
  product_name 
order by 
  item_revenue desc;

Answer 2

您可以使用子查询来实现此目的：

SELECT 
  x.month, 
  x.year, 
  x.device_category, 
  x.user_type, 
  x.product_name, 
  ROUND(SUM(x.item_revenue),2) as item_revenue 
FROM 
 (SELECT
    month,
    year,
    device_category,
    user_type,
    product_name,
    item_revenue
  FROM `ProjectName.DatasetName.GA_REPORT_3_*` 
  WHERE _table_suffix BETWEEN '20201101' and '20210131' 
  AND channel_grouping = 'Organic Search'
 ) x 
GROUP BY 
  x.month, 
  x.year,  
  x.product_name, 
  x.device_category, 
  x.user_type 
ORDER BY ROUND(SUM(x.item_revenue),2) DESC;