SQL transpose full table

Selrac picture Selrac · Mar 8, 2013 · Viewed 61.1k times · Source

I need to do the following transpose in MS SQL

from:

Day  A  B 
---------
Mon  1  2
Tue  3  4
Wed  5  6
Thu  7  8
Fri  9  0

To the following:

Value Mon Tue Wed Thu Fri 
--------------------------
A      1   3   5   7   9
B      2   4   6   8   0

I understand how to do it with PIVOT when there is only one column (A) but I can not figure out how to do it when there are multiple columns to transpose (A,B,...)

Example code to be transposed:

select LEFT(datename(dw,datetime),3) as DateWeek, 
  sum(ACalls) as A, 
  Sum(BCalls) as B 
from DataTable
group by LEFT(datename(dw,datetime),3)

Table Structure:

Column DataType
DateTime Datetime
ACalls int
BCalls int

Any help will be much appreciated.

Answer

Taryn picture Taryn · Mar 8, 2013

In order to transpose the data into the result that you want, you will need to use both the UNPIVOT and the PIVOT functions.

The UNPIVOT function takes the A and B columns and converts the results into rows. Then you will use the PIVOT function to transform the day values into columns:

select *
from
(
  select day, col, value
  from yourtable
  unpivot
  (
    value
    for col in (A, B)
  ) unpiv
) src
pivot
(
  max(value)
  for day in (Mon, Tue, Wed, Thu, Fri)
) piv

See SQL Fiddle with Demo.

If you are using SQL Server 2008+, then you can use CROSS APPLY with VALUES to unpivot the data. You code would be changed to the following:

select *
from
(
  select day, col, value
  from yourtable
  cross apply
  (
    values ('A', A),('B', B)
  ) c (col, value)
) src
pivot
(
  max(value)
  for day in (Mon, Tue, Wed, Thu, Fri)
) piv

See SQL Fiddle with Demo.

Edit #1, applying your current query into the above solution you will use something similar to this:

select *
from
(
  select LEFT(datename(dw,datetime),3) as DateWeek,
    col, 
    value
  from DataTable 
  cross apply 
  (
    values ('A', ACalls), ('B', BCalls)
  ) c (col, value)
) src
pivot
(
  sum(value)
  for dateweek in (Mon, Tue, Wed, Thu, Fri)
) piv