SQL - 前 7 天销售（每个客户，每个日期）答案

【问题标题】：SQL - Previous Seven Days Sale (Per Customer, Per Date)SQL - 前 7 天销售（每个客户，每个日期）
【发布时间】：2016-10-19 05:15:30
【问题描述】：

基于包含 CustomerID、Date 和 Daily Sales 字段的表，我需要计算每个 CustomerID、Date 组合的前 7 天的销售额。样本数据如下：

CustomerID  Date    Daily Sales
    1   10/18/16    $100
    1   10/17/16    $50
    1   9/1/16  $20
    1   8/5/16  $20
    1   7/1/16  $20
    1   6/15/16 $20
    1   1/1/16  $20
    2   10/18/16    $50
    2   10/17/16    $50
    2   10/16/16    $50
    2   10/15/16    $50
    2   10/14/16    $50

想要的输出如下：

CustomerID  Date    Daily Sales Last 7 Days Sales
1   10/18/16    $100    $150
1   10/17/16    $50 $50
1   9/1/16  $20 $20
1   8/5/16  $20 $20
1   7/1/16  $20 $20
1   6/15/16 $20 $20
1   1/1/16  $20 $20
2   10/18/16    $50 $250
2   10/17/16    $50 $200
2   10/16/16    $50 $150
2   10/15/16    $50 $100
2   10/14/16    $50 $50

这是我正在处理的查询：

SELECT
t1.[CustomerID]
,CAST(t1.[Date] AS DATE) AS Date
,SUM(t1.[DailySales]) AS [DailySales]
,SUM(t2.[DailySales]) AS [Last7DaysSales]
FROM TABLE_NAME t1 LEFT OUTER JOIN TABLE_NAME t2 ON t1.[CustomerID] = t2.[CustomerID]
AND CAST(t2.[Date] AS DATE) BETWEEN DATEADD(DAY, -7, t1.[Date]) and t1.[Date]

GROUP BY t1.[CustomerID], CAST(t1.[Date] AS DATE)
ORDER BY t1.[CustomerID], CAST(t1.[Date] AS DATE) DESC;

按预期返回 Last7DaysSales 的值，但对于当天的销售额不正确。我确定我在加入导致问题的过程中做了一些愚蠢的事情......

【问题讨论】：

欢迎来到 StackOverflow。您应该与我们分享您迄今为止所做的工作以及您面临的问题。请查看此页面以了解如何提出好的问题：stackoverflow.com/help/how-to-ask
安德鲁，我的加入可能是个白痴（SQL 新手） - 通过这个查询，我能够正确获取 Last7DaysSales（每个客户，每个日期），但是它只是针对那个单独的日期返回了不正确的销售额！同样，我确定我是个白痴。我编辑了我的帖子以提供查询。

标签： sql-server sql-server-2012

【解决方案1】：

SELECT SUM(Column) FROM TABLE_NAME WHERE date < (GETDATE()-7) AND CustomerId=1 - 为您提供客户 id = 1 的每日销售额。

【讨论】：

嗨，弗拉德 - 我需要返回数据集中每一天的每位客户前 7 天的销售额（而不是从 GETDATE() 开始的前 -7 天）

【解决方案2】：

您可以构建基本查询以获取除 Last7DaysSales 值之外的所有内容，然后使用子查询进行计算：

SELECT
t1.[CustomerID]
,CAST(t1.[Date] AS DATE) AS Date
,SUM(t1.[DailySales]) AS [DailySales]
,(select sum(t2.[DailySales]) 
  from TABLE_NAME t2 
  where t1.CustomerID = t2.CustomerID
  and cast(t2.[Date] as date) between dateadd(day, -7, cast(t1.[Date] as date)) and cast(t1.[Date] as date)
 ) as [Last7DaysSales]
FROM TABLE_NAME t1 LEFT OUTER JOIN TABLE_NAME t2 ON t1.[CustomerID] = t2.[CustomerID]
AND CAST(t2.[Date] AS DATE) BETWEEN DATEADD(DAY, -7, t1.[Date]) and t1.[Date]
GROUP BY t1.[CustomerID], CAST(t1.[Date] AS DATE)
ORDER BY t1.[CustomerID], CAST(t1.[Date] AS DATE) DESC;

请注意，窗口函数是一种更有效的方法，但这种方法可能更直观一些。

【讨论】：