Question

I’m looking to compute a column based on two other columns, but I have some special cases where the computation has to take multiple rows into account.

Here’s the query as it stands now:

SELECT
    UserName
    , EntryDate
    , Project
    , HoursWorked
    , HoursAvailable
    , UtilPct as HoursWorked / HoursAvailable
FROM
    MyDatabase
ORDER BY
    EntryDate

Which results in:


UserName EntryDate Project HoursWorked HoursAvailable UtilPct
Justin 12/17/2012 ABC 8 8 100
Justin 12/18/2012 ABC 4 8 50
Justin 12/18/2012 DEF 4 8 50


But 50% utilization for the last two entries is wrong, because both happened on the same day. Both of those rows should show 100%.

I need the query to…

1. Sum up all of the HoursWorked for a certain day
2. Divide that by the max HoursAvailable for that day
3. Put that value into the UtilPct column for each row

How do I make it do that?

Answer 1

I need the query to…

Sum up all of the HoursWorked for a certain day
Divide that by the max HoursAvailable for that day
Put that value into the UtilPct column for each row

How do I make it do that?

Sum up all of the HoursWorked:

SUM(HoursWorked) ... GROUP BY EntryDate

Divide that by the max HoursAvailable for that day

SUM(HoursWorked)/MAX(HoursAvailable) ... GROUP BY EntryDate

Put that value into the UtilPct column for each row:

SELECT ... SUM(HoursWorked)/MAX(HoursAvailable)*100.0 AS UtilPct
    FROM MyDatabase
    GROUP BY EntryDate
    ORDER BY EntryDate

But now you won’t be able to divide the rows by Project, you will only see the time utilization any single day. If you try to GROUP BY Project, you will get 50% instead of 100% again.

That is because you are asking for 4/8 to be 100%, and that would get you semantically nonsensical answers, such as “On Tuesday, I worked 100% on project ABC, and also 100% on project DEF”, to which one could answer “So on Tuesdays there’s 200% of you?”

You can solve the quandary using a JOIN between the table and itself to get the TotalHoursWorked in a single day, which is semantically different from the HoursWorked on a single project:

SELECT
    UserName
    , md1.EntryDate
    , Project
    , HoursWorked
    , HoursAvailable
    , (TotalHoursWorked / HoursAvailable)*100.0 AS UtilPct
FROM
    MyDatabase AS md1
    JOIN ( 
        SELECT EntryDate, SUM(HoursWorked) AS TotalHoursWorked
        FROM MyDatabase GROUP BY EntryDate
    ) AS md2 ON (md1.EntryDate = md2.EntryDate)
ORDER BY
    md1.EntryDate

Now the UtilPct will refer to the time usage of that day, which means you could get a 100% value on a project even if you worked at it for five minutes only, provided those five minutes were all that you had available that day. And you will still be able to add another column to indicate the percentage of HoursWorked over HoursAvailable for that project that day (in most systems, it ought to be roughly proportional to that project’s priority).