Chris, if I remember correctly Tableau's geo database includes cities down to 16,000 15,000 people. The workaround for your issue is to rename your two towns something link Cambridge, MN and Ashland, WI. Then when you bring them assign them lats/longs by hand. And then give them alias to get rid of the state names.
Hope this helps,
Tableau is apparently recognizing them as Cambridge, MN and Ashland, WI on initial view. Tableau recognizes that they are not the MA and OR cities, because they are listed as unfound originally. Why does it then set the lat/long for the MA and OR cities when I set the MN and WI towns.
I could do as you suggest, Shawn, but wouldn't I have to do that at the Extract layer? I am touching a mySQL database for this info and extracting it. Can I edit an extract? Ultimately, this extract will have thousands of rows for each city, so making that edit doesn't seem so functional, unless I am missing something.
So what you have in your data is 2 Ashlands and 2 Cambridges. Tableau has two matching records in it's geo database; and two that don't match. When you tell Tableau that you want to move Ashland to a point in WI, T says OK, and moves both the Ashland marks. The reason is there is nothing distinguishing the two marks from each other, except for state. And when you override T's geo location and put in your own, you by-pass the state field, so both Ashlands move.
There are several ways to get around this (I suggested one above). You can also fix the problem inside Tableau by using this calculation:
IF [city]='Ashland' AND [state]<>'OR'
OR [city]='Cambridge' AND [state]<>'MA'
THEN [city] END
Then use this to create a dual axis map. (See attached.) You could also make the distinction between the two Ashlands using a custom SQL query.
location_query-SW.twbx.zip 53.7 KB
Thanks for the response, Shawn,
Is there anyway to hack the Tableau geo code database? Ultimately, if not, I will probably go the route of SQL query edit