SQLSERVER 2012计算上一条，下一条数据的函数

用户1177503

发布于 2018-02-26 16:32:13

9860

发布于 2018-02-26 16:32:13

实际需求很普遍，比如求销售数据的每天与头一天的销售增长量。这里用一个汽车行驶数据来做例子：

先初始化数据：

CREATE TABLE [dbo].[CarData](
    [CarID] [int] NULL,
    [Mileage] [int] NULL,
    [M_year] [int] NULL,
    [M_Month] [int] NULL,
    [M_Day] [int] NULL
) ON [PRIMARY]
GO
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (1, 10, 2015, 1, 1)
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (1, 15, 2015, 1, 2)
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (1, 15, 2015, 1, 5)
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (1, 20, 2015, 1, 6)
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (1, 26, 2015, 1, 9)
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (1, 30, 2015, 1, 10)
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (1, 35, 2015, 1, 11)
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (2, 20, 2015, 1, 5)
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (2, 22, 2015, 1, 8)
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (2, 40, 2015, 1, 10)
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (2, 45, 2015, 1, 11)
INSERT [dbo].[CarData] ([CarID], [Mileage], [M_year], [M_Month], [M_Day]) VALUES (3, 50, 2015, 1, 11)

然后，使用下面的SQL来统计：

WITH ONE AS(
    SELECT ROW_NUMBER() OVER(PARTITION BY CarId ORDER BY CarId, M_Year, M_Month, M_Day) AS NodeId
          ,C.CarId
          ,C.Mileage
          ,C.M_Year
          ,C.M_Month
          ,C.M_Day
    FROM  dbo.CarData AS C
)
SELECT *
      ,COALESCE(One.Mileage - LAG(One.Mileage) over(PARTITION BY CarId order by One.NodeId),0) AS '增量'
FROM ONE

这里使用LAG函数来计算。

注意，这个查询只有在SQLSERVER 2012以上才支持，2008不支持，所以采用下面的方法实现：

WITH TWO AS(
SELECT ROW_NUMBER() OVER(PARTITION BY CarId ORDER BY CarId, M_Year, M_Month, M_Day) AS NodeId
          ,C.CarId
          ,C.Mileage
          ,C.M_Year
          ,C.M_Month
          ,C.M_Day
    FROM [dbo].[CarData] AS C
)
SELECT A.*
     , A.Mileage -  COALESCE(B.NextMileage, 0) AS '增量'
FROM TWO AS A
    OUTER APPLY (SELECT Mileage AS NextMileage FROM TWO AS B WHERE B.NodeId = A.NodeId - 1 AND B.CarId = A.CarId ) AS B;

执行查询，将得到下面的结果：

1    1    10    2015    1    1    10
2    1    15    2015    1    2    5
3    1    15    2015    1    5    0
4    1    20    2015    1    6    5
5    1    26    2015    1    9    6
6    1    30    2015    1    10    4
7    1    35    2015    1    11    5
1    2    20    2015    1    5    20
2    2    22    2015    1    8    2
3    2    40    2015    1    10    18
4    2    45    2015    1    11    5
1    3    50    2015    1    11    50

感谢 SOD开发技术群（PWMIS开发框架-SOD会员群 43109929）朋友提供的程序。

本文参与腾讯云自媒体分享计划，分享自作者个人站点/博客。

如有侵权请联系 cloudcommunity@tencent.com 删除

sql server