Sql 2012 - Pivot And Unpivot

I have summarised data in a table which is similar to this: Customer Year Month No_trans spend points 1 2015 1 30 400 10 1 2015 2 20 150 5

Solution 1:

You could use dynamic SQL to transpose table:

DECLARE@cols NVARCHAR(MAX) = 
                STUFF((SELECTDISTINCT','+ QUOTENAME(CONCAT([Year], '_', [Month]))
                      FROM #tab
                      FOR XML PATH(''), TYPE
                     ).value('.', 'NVARCHAR(MAX)') 
                     , 1, 1, '');

DECLARE@query NVARCHAR(MAX) = 
FORMATMESSAGE(
N'SELECT col_name, customer, %s
FROM (SELECT [year_month] = CONCAT([Year], ''_'', [Month]),
             Customer, No_Trans, spend, points
      FROM #tab) AS sub
UNPIVOT
(
    val FOR col_name  IN (No_trans, spend, points)
) AS unpvt
PIVOT
(
    MAX(val) FOR [year_month] IN (%s)
) AS pvt
ORDER BY customer, col_name;', @cols, @cols); 

EXEC [dbo].[sp_executesql] @query;

LiveDemo

Output:

╔══════════╦══════════╦════════╦════════╦════════╗
║   col    ║ customer ║ 2015_1 ║ 2015_2 ║ 2015_3 ║
╠══════════╬══════════╬════════╬════════╬════════╣
║ No_Trans ║        1 ║     30 ║     20 ║     10 ║
║ points   ║        1 ║     10 ║      5 ║     15 ║
║ spend    ║        1 ║    400 ║    150 ║    500 ║
║ No_Trans ║        2 ║      5 ║        ║        ║
║ points   ║        2 ║      7 ║        ║        ║
║ spend    ║        2 ║    100 ║        ║        ║
╚══════════╩══════════╩════════╩════════╩════════╝

If you want zeros at missing position you could use ISNULL/COALESCE. Be aware that not every datatatype could be replaced by 0 (int)

LiveDemo2

EDIT:

FORMATMESSAGE is fancy way to replace %s with string. It could be easily changed by simple REPLACE:

DECLARE@query NVARCHAR(MAX) = 
 N'SELECT col1, col2, <placeholder>
   FROM ...
   ...
   PIVOT( MAX(col) IN col2 IN (<placeholder>)
   ...';

 SET@query= REPLACE(@query, '<placeholder', @cols);

 -- for debug
 PRINT @query;

How it works:

1. Generate [year_month] column in subquery
2. UNPIVOT data (columns to row)
3. PIVOT result (rows to columns)
4. Wrap it with dynamic-SQL to allow generate columns without knowing [year_month] in advance.