This is a follow-up to a thread started by my esteemed colleague, Joe Cole, pertaining to changing user perspective without needing huge data joins (most recently: community.tableausoftware.com/message/215326).
We want to allow the user to select any one of a number of locations as "My Location" and then see the world from that perspective. Our initial approach was using data joins to create unique data sets for each location. Unfortunate, this approach is unsustainable due to a relatively large data set combined with hundreds of potential locations.
Thanks to the help from Jonathan Drummey and Robin Kennedy in the thread referenced above, we were able to use a filter from a secondary data source and table calcs to measure distance from "my location" to all other location without increasing the core data size. This approach also worked well for me on a different criteria: "only view job titles I employee".
However, when I combined these two attributes (i.e., provided the ability to filter on both distance from my location and titles I employee) the dashboards became unusable slow. Instead of taking a ~5 seconds, like it did with just one filter, it takes 30-45 seconds with both.
I have attached a .twbx with fake data where the user can change perspective of their location and then filter on both distances from and titles they employ at the selected location. It works really well in the attached demo, presumably because of the relatively small data set and only a handful of locations. But this same approach comes to a grinding halt with my real data set and calculation-heavy viz.
I am trying to determine if I am running into limitations of Tableau and table calcs or if there is a more elegant and efficient way to combine these two filters. Again, it works great with just one "Compute Using" variable but something about having the table calcs address two variables brings my real viz to it's knees.
Any thoughts appreciated on what might be wrong or even ways to problem shoot what is causing my slow down.