Got Sample Data?
User Tool Day Number of Uses A T1 9/1/2014 1 A T2 9/1/2014 0 A T3 9/1/2014 0 A T4 9/1/2014 2 B T1 9/1/2014 0 B T2 9/1/2014 1 B T3 9/1/2014 0 B T4 9/1/2014 1 C T1 9/1/2014 3 C T2 9/1/2014 2 C T3 9/1/2014 0 C T4 9/1/2014 0 A T1 9/2/2014 0 A T2 9/2/2014 0 A T3 9/2/2014 0 A T4 9/2/2014 0 B T1 9/2/2014 5 B T2 9/2/2014 2 B T3 9/2/2014 0 B T4 9/2/2014 1 C T1 9/2/2014 0 C T2 9/2/2014 2 C T3 9/2/2014 0 C T4 9/2/2014 1
If this is the data with no filters applied, then by summing the number of Uses by User, we see that no User has no Tool usage. Even though A has an entire day without usage, for the entire unfiltered time frame they used something on one of the days. So a count of users would be 0, and the list of non-users would be empty.
But, if we filtered to 9/2, the sum of Uses by User would show that A has no usage. So the count would be 1, and the list of non-users would include A.
Or, if we filtered by Tool, we'd see that for T2, User A is the only one who has no usage (despite User C having no usage for a single day). Filtering on T3, we'd see that no one has usage during the two days, so the non-user list would have A, B, C, and D, and the count distinct would be 4.
This is why summing is important because, otherwise, looking at one row at a time for instances of 0 Uses, all four Users would show up in the list just because there is a 0 associated to them somewhere in the dataset. Expand this out to a whole year, and it would be like punishing a user because they missed a single day out of the entire 365.
Is this the RAW data structure, or what you are seeing inside Tableau? If its your raw data structure, you can take this table of data, and copy/paste it directly into Tableau as a new data source, in order to post a sample packaged workbook with your view setup with the work you've completed here in the thread. To attach, click "Use advanced editor" in the top-right corner, then the "Attach" option appears in the bottom right, next to @ Mention.
For me, its very difficult to answer these questions without interacting with Tableau. Knowing what you are expecting to see as an output (mockup) can be helpful as well. But there is always the chance someone else can answer with the information you've provided here.
Thanks for the tip. I've whipped together a quick workbook.
Output we'd expect to see:
With no filters applied (data as is), countd of users would be 0, list of non-users would be empty.
Filter on 9/2: countd of users would be 1, non-user list would be A.
Filter on T2: countd of users would be 1, non-user list would be A.
Filter on T3: countd of users would be 4, non-user list would be A, B, C, D
Filter on T3 and T4: countd of users would be 0, list of non-user would be empty.
Hope this makes more sense now.
Samp Count by Sum.twbx 19.9 KB
I'm sorry, but I really don't follow the results and "filter settings" you are describing. But perhaps this will help--I was able to write a formula similar to the one you had mentioned previously, something like:
IF SUM([Number of Uses]) == 0 or ISNULL(SUM([Number of Uses]))
then "Non User"
Does not throw an error; does that help you at all? That's probably as far as I can take this without understanding more about the situations you are describing (there is no action filter from a heat map, for instance, and I'm just not following your logic in your filter scenarios)