Either you have to differentiate those rows with an additional dimension (such as date-time entered, for example), and then grab the first one using the unique value (MIN([Date-time field]), for example), or instead of doing SUM([Hours]), do AVG([Hours]). (Assuming the hours value for each row is the same.)
You also have a [Number of Records] measure that Tableau creates for you. You can use that to determine the number of rows in a given course, and then do your own average calculation.
SUM([Hours]) / SUM([Number of Records]) would be evaluated at the [Course Number] level if you have [Course Number] on the sheet.
I think the issue you are facing has to do with the database connection you are using. It looks like you are using a "legacy" connection, and that has limited functionality in Tableau. For example, Tableau certainly has the COUNTD function, but not for legacy connections, apparently. If there is some way you can change to a regular connection, rather than legacy, I think you will get all of the functionality of Tableau back again, and you should be able to do much more.
Here is a Level of Detail (LOD) Expression that I think gives you what you're looking for, even if [Class Title] is not part of the view's level of detail:
I tested it in isolation and it seems to behave as you would want it to. I also dropped it into your Southeast view, but I can't tell for certain that it's yielding the results you were anticipating. Let me know.
NOTE: Your data sources are legacy Excel files, which do not support LOD Expressions. I had to create an extract of your data to enable LOD Expressions.
Workbook attached (version 10.1).
Let me know if this is what you're looking for!
TAB PCC Workbook01_JC-10.1.twbx 25.5 MB
That didn't quite eliminate the issue of duplicate records, but it has
been very helpful! As my boss and I tried to solve the issue, LOD
calculations have been essential (so thank you!). I'm not sure where we
are right now (haven't chatted with my boss yet today), but I might check
in later with additional questions.
Thank you again for the help!
On Wed, Mar 29, 2017 at 5:44 PM, Jamieson Christian <