0

输入： 有表'A'

Store   Category   Sales   Day
11      aaa        1.5     Sunday
11      aaa        0.5     Monday
11      aaa        2.5     Tuesday
11      aaa        2.0     Wednesday
11      aaa        3.0     Thursday
11      aaa        3.5     Friday
11      aaa        0.5     Saturday
22      bbb        0.5     Sunday
22      bbb        1.5     Monday
22      bbb        2.3     Tuesday
22      bbb        0.3     Wednesday
22      bbb        1.4     Thursday
22      bbb        4.1     Friday
22      bbb        0.2     Saturday

场景： 必须取按商店、类别分组的销售额平均值并保存在单独的列中，并选择前 3 个销售日并保存在单独的列中。结果，一个商店一行，示例输出中的类别如下所示。

预期输出：

Store   Category   AvgSales   PeakDay1   PeakDay2   PeakDay3   
11      aaa        1.92       Friday     Thursday   Tuesday
22      bbb        1.47       Friday     Tuesday    Monday

尝试查询：

SELECT
  Store,
  Category,
  avg(Sales) as AvgSales,
  ARRAY_AGG(Sales ORDER BY Sales DESC LIMIT 3)  #but this line will not produce results in 3 separate columns
FROM A
GROUP BY Site, Category

提前致谢！

4

2 回答 2

1

您可以使用数组 agg，但row_number()看起来很简单：

select store, category, avg(sales),
       max(case when seqnum = 1 then day end) as peakday1,
       max(case when seqnum = 2 then day end) as peakday2,
       max(case when seqnum = 3 then day end) as peakday3
from (select store, category, day, 
             row_number() over (partition by store, category order by sales desc) as seqnum
      from a
     ) a
group by store, category;

如果你想把它放到一个数组列中，你可以使用：

SELECT Store, Category, avg(Sales) as AvgSales,
       ARRAY_AGG(day ORDER BY Sales DESC LIMIT 3) 
FROM A
GROUP BY Site, Category
于 2020-07-16T16:13:32.357 回答
1

以下是 BigQuery 标准 SQL

#standardSQL
SELECT Store, Category, AvgSales,
  Days[OFFSET(0)] PeakDay1,
  Days[SAFE_OFFSET(1)] PeakDay2,
  Days[SAFE_OFFSET(2)] PeakDay3
FROM (
  SELECT Store, Category, 
    ROUND(AVG(Sales), 2) AvgSales,
    ARRAY_AGG(Day ORDER BY Sales DESC LIMIT 3) Days
  FROM `project.dataset.table` t
  GROUP BY Store, Category
)

如果适用于您的问题的样本数据 - 输出是

Row Store   Category    AvgSales    PeakDay1    PeakDay2    PeakDay3     
1   11      aaa         1.93        Friday      Thursday    Tuesday  
2   22      bbb         1.47        Friday      Tuesday     Monday
于 2020-07-16T16:23:03.770 回答