Question

哦，所以我有这个东西。我想知道每个实体每周的订单数量。我从这个问题开始：

select 
DATEPART(YEAR, OrderDate) AS 'Year',
DATEPART(WEEK, OrderDate) AS 'Week',
COUNT(*) as 'Amount of Orders',
EntityID
FROM tblOrder
WHERE DATEPART(YEAR, tblOrder.OrderDate) = 2015
GROUP BY EntityID,OrderDate
ORDER BY Week asc

结果很好，但到处都是双排：

Year | Week | Amount of Orders | EntityID
2015 |  1   |        1         |     1
2015 |  1   |        1         |     1
2015 |  1   |        1         |     1
2015 |  1   |        1         |     1
2015 |  1   |        1         |     1
2015 |  1   |        1         |    15  
2015 |  1   |        1         |    15

我想要达到的目标是：

Year | Week | Amount of Orders | EntityID
2015 |  1   |        5         |     1
2015 |  1   |        2         |     2
2015 |  1   |        4         |     3
2015 |  1   |        9         |     4

等等。我的查询出了什么问题？

Answer 1

GROUP BY年和周而不是orderdate：

select [Year], [week], EntityID, count(*)
from
(
select DATEPART(YEAR, OrderDate) AS 'Year',
       DATEPART(WEEK, OrderDate) AS 'Week',
       EntityID
FROM tblOrder
WHERE DATEPART(YEAR, tblOrder.OrderDate) = 2015
) dt
GROUP BY [Year], [Week], EntityID
ORDER BY Week asc

因为同一年/周可以有来自不同日期的订单。如果您是GROUP BY日期，那么您将获得该年/周的多行。

Answer 2

只需在group by子句中使用相同的表达式。

pyspark --master spark://myurl:7077

在SQL中添加行

2 个答案: