Question

我有两个数据库（MS SQL Server）;一个具有当前的报告表集，另一个具有上周的报告表集。每周我都想比较每一个中的所有测量值（尽管每周的行数可能会有所不同）与前一周相比，并确保我的QTD和YTD数字正在增加。行的顺序可能每周都不同，所以无论我采用什么，都必须考虑到这一点。

另一个考虑因素是我有大约50个这些表要进行比较，因此手动执行每个表都不是一个选项。有一些错误日志也很棒，就像本周YTD值不比上周高的那样。

本周

+-------+------------+--------+
| Item  | TimePeriod | Number |
+-------+------------+--------+
| Pigs  | QTD        |     28 |
| Pigs  | YTD        |     86 |
| Cows  | QTD        |      7 |
| Cows  | YTD        |      9 |
| Bears | QTD        |     67 |
| Bears | YTD        |    132 |
+-------+------------+--------+

上周

+-------+------------+--------+
| Item  | TimePeriod | Number |
+-------+------------+--------+
| Bears | QTD        |     55 |
| Bears | YTD        |    120 |
| Pigs  | QTD        |     22 |
| Pigs  | YTD        |     80 |
| Cows  | QTD        |      5 |
| Cows  | YTD        |     10 |
+-------+------------+--------+

错误

错误：奶牛YTD当前周-9上周第10周

Answer 1

也许这样的事情可以做到：

SELECT 'Error ' + CW.TimePeriod + ' Current week:' + CAST(CW.Number AS VARCHAR) + ' Prev week:' + CAST(PW.Number AS VARCHAR)
FROM [Current Week DB].Table CW 
INNER JOIN [Previous DB].Table PW ON 
  CW.ITEM=PW.ITEM 
  AND CW.TIMEPERIOD=PW.TIMEPERIOD 
  AND CW.Number=PW.Number

加入所有列将使表格之间的行没有变化。需要明确更改表的标识符。

Answer 2

您的问题中没有任何内容可以证明使用多个数据库或多个表格。

相反，请考虑使用这样的表格。

ITEM     TIMEPERIOD  WEEKNUM     NUMBER
Bears    QTD         2013-26     67
Bears    QTD         2013-25     55
Bears    YTD         2013-26     132
Bears    YTD         2013-25     120
Cows     QTD         2013-26     7
Cows     QTD         2013-25     5
Cows     YTD         2013-26     9
Cows     YTD         2013-25     10
Pigs     QTD         2013-26     28
Pigs     QTD         2013-25     22
Pigs     YTD         2013-26     86
Pigs     YTD         2013-25     80

如果您有SQL Server 2012，那么基本查询相对简单明了。

select 
    item, timeperiod, weeknum, number,
    number - lead (number) 
             over (partition by item, timeperiod 
                       order by item, timeperiod, weeknum desc) as change
from table1;

ITEM    TIMEPERIOD   WEEKNUM   NUMBER   CHANGE
--
Bears   QTD          2013-26   67       12
Bears   QTD          2013-25   55       (null)
Bears   YTD          2013-26   132      12
Bears   YTD          2013-25   120      (null)
Cows    QTD          2013-26   7        2
Cows    QTD          2013-25   5        (null)
Cows    YTD          2013-26   9       -1
Cows    YTD          2013-25   10       (null)
Pigs    QTD          2013-26   28       6
Pigs    QTD          2013-25   22       (null)
Pigs    YTD          2013-26   86       6
Pigs    YTD          2013-25   80       (null)

如果您只想要那些具有负面更改的行，请使用上一个查询作为派生表，并添加WHERE子句

select *
from 
  (select 
      item, timeperiod, weeknum, number,
      number - lead (number) 
               over (partition by item, timeperiod 
                         order by item, timeperiod, weeknum desc) as change
  from table1) t
where change < 0
  and timeperiod = 'YTD';

ITEM    TIMEPERIOD   WEEKNUM   NUMBER   CHANGE
--
Cows    YTD          2013-26   9       -1

针对历史数据的QA SQL表

2 个答案: