How to build a query that SELECT rows into columns for SQL?

Question

How can I write a query in SQL Server that returns a column for each variable of each unique ID? Below is a simplified example of what I am trying to accomplish. Would you be able to point me in the right direction?

Table_Exists

ID    |    LOCATION    |    STATUS
__________________________________
001   |     top        |     3
001   |     mid        |     1
001   |     bot        |     4
002   |     top        |     2
002   |     mid        |     2
002   |     bot        |     1
003   |     top        |     1
003   |     mid        |     4
003   |     bot        |     2

Here is a simplified example of what I am trying, obviously without success. The code below returns an error message for incorrect syntax around '=':

SELECT  
  ID,
  LOCATION = top AS Top,
  LOCATION = mid AS Middle,
  LOCATION = bot AS Bottom
FROM 
  Table_Exists

I trying to build:

Table_Desired

ID | Top | Middle | Bottom 
__________________________
001|  3  |   1    |  4
002|  4  |   2    |  2 
003|  1  |   4    |  2

it's called a [pivot query](https://technet.microsoft.com/en-us/library/ms177410%28v=sql.105%29.aspx)... — Marc B, Nov 25 '15 at 19:14
Have a look at the pivot command. http://stackoverflow.com/questions/18116020/sql-server-2008-vertical-data-to-horizontal — domenicr, Nov 25 '15 at 19:18

Lukasz Szozda · Accepted Answer · 2015-11-27T12:22:09.590

3

You can use self join (as long as your LOCATION is NOT NULL and it contains all top/mid/bot values):

 SELECT t1.ID, t1.Status AS Top, t2.Status AS Middle, t3.Status AS Bottom
 FROM Table_Exists t1
 JOIN Table_Exists t2
   ON t1.ID = t2.ID
  AND t1.LOCATION = 'top'
  AND t2.Location = 'mid'
 JOIN Table_Exists t3
   ON t1.ID = t3.ID
  AND t1.LOCATION = 'top'
  AND t3.Location = 'bot';

LiveDemo

For more generic solution use PIVOT:

SELECT
     [ID]
    ,[Top]    = [top] 
    ,[Middle] = [mid]
    ,[Bottom] = [bot]
FROM #Table_Exists t
PIVOT (
   MAX(STATUS) FOR LOCATION IN ([top],[mid],[bot])
) AS Piv

LiveDemo2

edited Nov 27 '15 at 12:22

answered Nov 25 '15 at 19:16

Lukasz Szozda

162,964
23
234
275

Thanks @lad2025 This was very helpful – Bret Shandro Nov 25 '15 at 23:18
If any values for "Status" are 0/null the entire row for that "ID" will be excluded. If Table_Exists is missing data the PIVOT function can be used to return "ID" rows when "Status" is 0/null – Bret Shandro Nov 27 '15 at 00:49

score 2 · Answer 2 · edited Nov 25 '15 at 19:51

2

Use pivot function like this

SELECT
    ID,
    [top] AS [Top],
    [mid] AS Middle,
    [bot] AS Bottom
FROM(SELECT 
    ID, 
    LOCATION,
    STATUS
FROM 
    Table_Exists) AS SourceTable
PIVOT (AVG(STATUS) FOR LOCATION IN ([top],[mid],[bot])) AS PivotTable

edited Nov 25 '15 at 19:51

Mihai

26,325
7
66
81

answered Nov 25 '15 at 19:43

G.Maltby

66
5

I am getting error message "Error converting data type nvarchar to int." Any thoughts? @Mihai – Bret Shandro Nov 27 '15 at 01:07
Is your Status column a nvarchar? If so then CAST(STATUS, int) AS STATUS will let get the average – G.Maltby Nov 27 '15 at 01:17

How to build a query that SELECT rows into columns for SQL?

2 Answers2