Question

我正在研究一些调查数据，并且想知道我是否可以重新排列数据以使其更加可用。结果被分类为1-5，我希望首选表按值计算结果并逐个计算。

原始表：

year | month | customer_id | survey | q1 | q2 | q3 | q4 | q5 | q6 ----> q29
-----|-------|-------------|--------|----|----|----|----|----|---
2016 | Oct   | ABC12345678 | 1      | 1  | 2  | 3  | 1  | 2  | 3 
2016 | Oct   | DEF12345678 | 1      | 2  | 1  | 4  | 2  | 1  | 1
2016 | Oct   | GHI12345678 | 1      | 4  | 2  | 1  | 1  | 3  | 2 
2016 | Oct   | JKL12345678 | 1      | 2  | 3  | 2  | 4  | 1  | 3 
2016 | Oct   | MNO12345678 | 1      | 5  | 2  | 3  | 1  | 2  | 3 
2016 | Oct   | PQR12345678 | 1      | 3  | 4  | 4  | 2  | 4  | 4 
2016 | Oct   | STU12345678 | 1      | 1  | 5  | 3  | 1  | 2  | 5 
2016 | Oct   | VWX12345678 | 1      | 2  | 2  | 4  | 2  | 1  | 1

首选表：

Year | Month | Survey | Question | 1  | 2  | 3  | 4  | 5  |
-----|-------|--------|----------|----|----|----|----|----|
2016 | Oct   |   1    |   q1     | 80 | 45 | 25 | 63 | 89 |  
2016 | Oct   |   1    |   q2     | 65 | 75 | 35 | 53 | 69 |

我可以使用基本的选择查询来执行此操作，但是为每个问题执行此操作最终会有29个联合，并且必须有更快的方式。

此致

尼尔

Answer 1

在有人发布更好的解决方案之前，我会使用这个：

    <!-- language: lang-sql -->

use tempdb;
create table #tempsurvey (year int, month varchar(32), customer_id varchar(32), survey int, [q1] int, [q2] int, [q3] int, [q4] int, [q5] int, [q6] int, [q7] int, [q8] int, [q9] int, [q10] int, [q11] int, [q12] int, [q13] int, [q14] int, [q15] int, [q16] int, [q17] int, [q18] int, [q19] int, [q20] int, [q21] int, [q22] int, [q23] int, [q24] int, [q25] int, [q26] int, [q27] int, [q28] int, [q29] int);
insert into #tempsurvey values (2016,'Oct', 'ABC12345678', 1, 1,2,3,1,2,3,1,2,3,1,2,3,1,2,3,1,2,3,1,2,3,1,2,3,1,2,3,1,2);
insert into #tempsurvey values (2016,'Oct', 'DEF12345678', 1, 4,5,1,4,5,1,4,5,1,4,5,1,4,5,1,4,5,1,4,5,1,4,5,1,4,5,1,4,5);

with cte as (
  select t.[year], t.[month], t.customer_id, t.survey, x.question, x.answer
    from #tempsurvey t
      cross apply (values ('q1',q1) ,('q2',q2) ,('q3',q3) ,('q4',q4) ,('q5',q5) ,('q6',q6) ,('q7',q7) ,('q8',q8) ,('q9',q9) ,('q10',q10) ,('q11',q11) ,('q12',q12) ,('q13',q13) ,('q14',q14) ,('q15',q15) ,('q16',q16) ,('q17',q17) ,('q18',q18) ,('q19',q19) ,('q20',q20) ,('q21',q21) ,('q22',q22) ,('q23',q23) ,('q24',q24) ,('q25',q25) ,('q26',q26) ,('q27',q27) ,('q28',q28) ,('q29',q29)) 
      as x (Question,Answer)
 )
  select [year], [month], [survey], question, [1]=sum(case when answer=1 then 1 else 0 end), [2]=sum(case when answer=2 then 1 else 0 end), [3]=sum(case when answer=3 then 1 else 0 end), [4]=sum(case when answer=4 then 1 else 0 end), [5]=sum(case when answer=5 then 1 else 0 end)
    from cte
      group by [year], [month], [survey], question;

  drop table #tempsurvey;

Brad Schulz对交叉申请：http://bradsruminations.blogspot.com/search/label/CROSS%20APPLY

Answer 2

肖恩是对的。它会是这样的：

with subquery as (
  select year, month, survey, question, tempVal from #table
  unpivot
  (tempVal for question in (q1, q2, q3, q4, q5, q6, q7, ..., q29)) as up
)
select year, month, survey, question,
  sum(case when tempVal = 1 then 1 else 0 end) as a1,
  sum(case when tempVal = 2 then 1 else 0 end) as a2,
  sum(case when tempVal = 3 then 1 else 0 end) as a3,
  sum(case when tempVal = 4 then 1 else 0 end) as a4,
  sum(case when tempVal = 5 then 1 else 0 end) as a5
from subquery
group by year, month, survey, question

重新排列数据集

2 个答案: