Question

My dataset looks like this:

COLA  | COLB
Name1 | 218
Name2 | 157
Name3 | 134
Name4 | 121

I need this output:

COLA  | COLB| COLC
Name1 | 218 | 0.34
Name2 | 157 | 0.60
Name3 | 134 | 0.71
Name4 | 121 | 1

My SQL looks like this so far:

SELECT COLA, COLB, COLB/SUM(COLB) FROM #MyTempTable

Two problems with this SQL. One, COLC is 0 everytime and I don’t understand that. Two, even if it did result in the % it’s not a cumulative %.

I’ve seen some similar threads on StackOverflow, but I wasn’t able to make the answers from those threads work in my exact scenario.

Thanks in advance for any suggestions!

Answer 1

I think you’re looking for something like this, though your example calculations may be off a little:

SELECT
    COLA,
    COLB,
    ROUND(
        -- Divide the running total...
        (SELECT CAST(SUM(COLB) AS FLOAT) FROM #MyTempTable WHERE COLA <= a.COLA) /
        -- ...by the full total
        (SELECT CAST(SUM(COLB) AS FLOAT) FROM #MyTempTable),
        2
    ) AS COLC
FROM #MyTempTable AS a
ORDER BY COLA

EDIT: I’ve added rounding.

This gives us the following output:

COLA    COLB    COLC
Name1   218     0.35
Name2   157     0.6
Name3   134     0.81
Name4   121     1

The reason that your results are 0 (or 1) is because you are dividing ints by ints, thus giving you an int (see Datatype precedence).

UPDATE:

I should add that this uses a “triangular join” to get the running total (WHERE COLA <= a.COLA). Depending upon your SQL Server version, you may compare this to other options if performance becomes a concern.