Question

我有一组看起来像这样的数据（非常简化）：

productId    Qty   dateOrdered
---------    ---   -----------
       1       2    10/10/2008
       1       1    11/10/2008
       1       2    10/10/2009
       2       3    10/12/2009
       1       1    10/15/2009
       2       2    11/15/2009

除此之外，我们正在尝试创建一个查询来获取类似的内容：

productId  Year  Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
---------  ----  --- --- --- --- --- --- --- --- --- --- --- ---
        1  2008    0   0   0   0   0   0   0   0   0   2   1   0
        1  2009    0   0   0   0   0   0   0   0   0   3   0   0
        2  2009    0   0   0   0   0   0   0   0   0   3   2   0

我现在这样做的方式，我正在做12次选择，每个月一次，并把它们放在临时表中。然后我做了一个巨大的加入。一切正常，但这家伙很慢。

我知道这并不多，但我知道我几乎没有资格成为db世界中的tyro，我想知道是否有一个更好的高级方法，我可能会尝试。（我猜是有的。）

（我正在使用MS Sql Server，因此特定于该数据库的答案很好。）

（我刚刚开始将“PIVOT”视为可能的帮助，但我对此一无所知，所以如果有人想对此发表评论，那也可能会有所帮助。）

Answer 1

select productId, Year(dateOrdered) Year
  ,isnull(sum(case when month(dateOrdered) = 1 then Qty end), 0) Jan
  ,isnull(sum(case when month(dateOrdered) = 2 then Qty end), 0) Feb 
  ,isnull(sum(case when month(dateOrdered) = 3 then Qty end), 0) Mar
  ,isnull(sum(case when month(dateOrdered) = 4 then Qty end), 0) Apr
  ,isnull(sum(case when month(dateOrdered) = 5 then Qty end), 0) May
  ,isnull(sum(case when month(dateOrdered) = 6 then Qty end), 0) Jun
  ,isnull(sum(case when month(dateOrdered) = 7 then Qty end), 0) Jul
  ,isnull(sum(case when month(dateOrdered) = 8 then Qty end), 0) Aug
  ,isnull(sum(case when month(dateOrdered) = 9 then Qty end), 0) Sep
  ,isnull(sum(case when month(dateOrdered) = 10 then Qty end), 0) Oct
  ,isnull(sum(case when month(dateOrdered) = 11 then Qty end), 0) Nov
  ,isnull(sum(case when month(dateOrdered) = 12 then Qty end), 0) Dec
from Table1
group by productId, Year(dateOrdered)

SQL Fiddle

Answer 2

SELECT productId, YEAR,
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=1),0) as 'JAN',
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=2),0) as 'FEB',
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=3),0) as 'MAR',
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=4),0) as 'APR',
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=5),0) as 'MAY',
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=6),0) as 'JUN',
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=7),0) as 'JUL',
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=8),0) as 'AUG',
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=9),0) as 'SEP',
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=10),0) as 'OCT',
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=11),0) as 'NOV',
ISNULL((SELECT SUM(Qty) FROM Product WHERE productId=X.productId AND YEAR=YEAR(dateOrdered) AND MONTH(dateOrdered)=12),0) as 'DEC'
FROM (
SELECT productId, YEAR(dateOrdered) AS YEAR FROM Product
GROUP BY YEAR(dateOrdered),ProductId) X

Answer 3

您可以使用查询联合而不是临时表，也可以使用数据透视表选项。

以下是关于它的论坛讨论：

Sql Server Forums - Show the row-wise data as column-wise

Answer 4

这有资格作为演示问题演示文稿和SQL并不总是很好地混合。

在应用程序层中隔离您的表示逻辑：

节省您的维护时间 - 更改您的应用程序代码，但保持您的SQL完好无损;
使您能够更快地适应短暂的客户需求;
比摆弄一个交叉表或数据透视表让你更满意，这些表几乎完全符合你的想法。

下面是一个如何在Python中执行此操作的示例（您可以使用优秀的pyodbc模块连接到SQL Server）：

from collections import defaultdict
from datetime import date

dd = defaultdict(int)

# input 
rows = [(1,2,date(2008,10,10)), (1,1,date(2008,11,10)), 
        (1,2,date(2009,10,10)), (2,3,date(2009,10,12)), 
        (1,1,date(2009,10,15)), (2,2,date(2009,11,15))]

for row in rows:
    # row[0] == productId
    # row[1] == Qty
    # row[2] == dateOrdered
    # pyodbc enables referring to column names by name
    dd[(row[2].year, row[2].month, row[0])] += row[1]

presentation_rows = sorted(set((i[0], i[2]) for i in dd.keys()))

for i in presentation_rows:
  print i[1], i[0], 
  for j in range(0,13):
    try:
      print dd[i[0], j, i[1]], 
    except IndexError:
      print 0,
  print

# output
# 1 2008 0 0 0 0 0 0 0 0 0 0 2 1 0
# 1 2009 0 0 0 0 0 0 0 0 0 0 3 0 0
# 2 2009 0 0 0 0 0 0 0 0 0 0 3 2 0

Answer 5

对于使用Big Query的用户，您可以使用以下内容：

select *
from UNNEST(GENERATE_DATE_ARRAY('2015-10-01', '2019-10-01', INTERVAL 1 MONTH))

请参见https://cloud.google.com/bigquery/docs/reference/standard-sql/functions-and-operators?hl=fr#generate_date_array

Answer 6

试试这个。因此，此代码将在特定时间范围内选择数据，然后将其转换为新列。例如，在我的sql代码中：它从列'L_dt'中选择'2014-10-01'和'2014-10-31'之间的时间范围，然后创建一个名为“October”的新列。通过这种方式，我们可以在源自一列的不同列中布置数据。

select 
sum(case when L_dt between '2014-10-01' and '2014-10-31' then 1 else 0 end) October,
sum(case when L_dt between '2014-11-01' and '2014-11-30' then 1 else 0 end) November,
sum(case when L_dt between '2014-12-01' and '2014-12-31' then 1 else 0 end) December
from Table;

如果输入如下： L_dt
2014年10月13日情节中字情节中字 2014年10月

然后输出

+---------+----------+----------+
| October | November | December |
+---------+----------+----------+
|    2    |    1     |    1     |
+---------+----------+----------+

如何在Sql中创建“月”列？

6 个答案: