I have the following table in DB2:
COLUMN_NAME | DATA_TYPE | TYPE_NAME | COLUMN_SIZE | COLUMN_TEXT |
---|---|---|---|---|
DMYEAR | 3 | DECIMAL | 4 | Fiscal Year |
DMYTYP | -2 | CHAR () FOR BIT DATA | 1 | Year Type |
DMPTYP | -2 | CHAR () FOR BIT DATA | 1 | Period Type |
DMTYPE | -2 | CHAR () FOR BIT DATA | 6 | Type of Data |
DMPROD | -2 | CHAR () FOR BIT DATA | 35 | Product Code |
DMVL01 | 3 | DECIMAL | 17 | Value Period 1 |
DMVL02 | 3 | DECIMAL | 17 | Value Period 2 |
DMVL03 | 3 | DECIMAL | 17 | Value Period 3 |
DMVL04 | 3 | DECIMAL | 17 | Value Period 4 |
DMVL05 | 3 | DECIMAL | 17 | Value Period 5 |
DMVL06 | 3 | DECIMAL | 17 | Value Period 6 |
DMVL07 | 3 | DECIMAL | 17 | Value Period 7 |
DMVL08 | 3 | DECIMAL | 17 | Value Period 8 |
DMVL09 | 3 | DECIMAL | 17 | Value Period 9 |
DMVL10 | 3 | DECIMAL | 17 | Value Period 10 |
DMVL11 | 3 | DECIMAL | 17 | Value Period 11 |
DMVL12 | 3 | DECIMAL | 17 | Value Period 12 |
The value periods are Jan-Feb and can sum up various data types for each month like below:
SELECT
DMYEAR,
DMPTYP,
DMPROD,
Sum(CASE WHEN DMTYPE = 'COST' THEN DMVL01 ELSE 0 END) AS COST,
Sum(CASE WHEN DMTYPE = 'COST' THEN DMVL02 ELSE 0 END) AS COST,
Sum(CASE WHEN DMTYPE = 'COST' THEN DMVL03 ELSE 0 END) AS COST,
Sum(CASE WHEN DMTYPE = 'RTNCST' THEN DMVL01 ELSE 0 END) AS RTNCST,
Sum(CASE WHEN DMTYPE = 'RTNCST' THEN DMVL02 ELSE 0 END) AS RTNCST,
Sum(CASE WHEN DMTYPE = 'RTNCST' THEN DMVL03 ELSE 0 END) AS RTNCST
FROM DWM
WHERE DMPTYP = 'M'
GROUP BY DMYEAR, DMPTYP, DMPROD
ORDER BY 1, 3
Which produces output like this:
DMYEAR | DMPTYP | DMPROD | COST | COST | COST | RTNCST | RTNCST | RTNCST |
---|---|---|---|---|---|---|---|---|
2010 | M | 11105 | 435798.00000 | 441252.00000 | 479774.00000 | -2921.00000 | 0.00000 | 0.00000 |
2010 | M | 11350 | 538938.00000 | 593764.00000 | 511088.00000 | -4074.00000 | -2503.00000 | 0.00000 |
2010 | M | 11364 | 2905557.00000 | 3167448.00000 | 3534924.00000 | -5691.00000 | 0.00000 | 0.00000 |
2010 | M | 11374 | 854274.00000 | 829392.00000 |
But I would like to create a new column for month denoting 1-12 and adjusting the sum to produce the corresponding values so it looks like this:
DMYEAR | DMMONTH | DMPTYP | DMPROD | COST | RTNCST |
---|---|---|---|---|---|
2011 | 1 | M | 11105 | 278021.00000 | -7522.00000 |
2011 | 2 | M | 11105 | 615232.00000 | -938.00000 |
2011 | 3 | M | 11105 | 2153272.00000 | -2805.00000 |
I think it is possible with subqueries, but keeping in mind I have 8-9 different DMTYPEs, so I was hoping there is a more elegant solution than to make 9 case statements per DMTYPE.
Try this adding the corresponding expressions to the inner CASE
to get all 12 (x, DMVLx) pairs:
WITH
T (DMMONTH) AS (VALUES 1 UNION ALL SELECT DMMONTH + 1 FROM T WHERE DMMONTH < 12)
SELECT
DMYEAR, DMMONTH, DMPTYP, DMPROD
, Sum (CASE WHEN DMTYPE = 'COST' THEN CASE DMMONTH WHEN 1 THEN DMVL01 WHEN 2 THEN DMVL02 ... WHEN 12 THEN DMVL12 END ELSE 0 END) AS COST
, Sum (CASE WHEN DMTYPE = 'RTNCST' THEN CASE DMMONTH WHEN 1 THEN DMVL01 WHEN 2 THEN DMVL02 ... WHEN 12 THEN DMVL12 END ELSE 0 END) AS RTNCST
FROM DWM, T
WHERE DMPTYP = 'M'
GROUP BY DMYEAR, DMMONTH, DMPTYP, DMPROD
ORDER BY 1, 3