Peter, yes it's possible to import custom polygons in both 6 & 7. Search these Forums and the Knowledge Base for 'import geocode' and you'll find lot of info on how to do it. Now if I could only read Danish I could fully enjoy your site, but it looks nice. --Shawn
I will see if I can persuade our customer to translate it to english :-).
Purpose of the viz is to show number of athletes for each munichipals in Denmark. And filter it by type of sport (sports association), age and ***
the left row are based on the raw numbers - the right is based on athletes/total population
The top row is current year (År) - the buttom row is the change i percentages between the current year and the base year.
I have looked and not found any new information. The how-to described here: http://kb.tableausoftware.com/articles/knowledgebase/polygon-shaded-maps is not really useful for larger data sets because of the explosion of data due to the join in the data source.
The reduced dataset in the viz is approx 100.000 rows in the membership measuregroup, 10000 in the geography (map polygons (100 polygons)) and 100.000 in the population measuregroup. (the original dataset is 15 mio membership rows)
What I would have liked were to make an custom geocode import of polygons - just like it is possible with ordinary points. And just reference the polygons like you would do with a zipcode in 6.0 or 5.2.
There was lots of discussion of custom geocoding for filled maps during the 7.0 beta, with lots of people suggesting/requesting it. There was also a bit of discussion of the possibility of using spatial datatypes from databases such as SQL server which support it. Unfortunately there's no support for either of these options in Tableau 7.0 (I think Shawn must have just been thinking of normal custom geocoding of locations as opposed to polygons). It was clear from the discussion during the beta that the Tableau folk are well aware of what people would like and are thinking about how to support it one day, though.
The impression I got from the discussions was that the main difficulty lies around availability of data at an appropriate level of detail for visualisation. Most shape files that are available are far too detailed: fine for GIS, not for Tableau visualisations. If you look at the built-in shapes that come with Tableau you'll see that there has been a lot of work that has had to go into simplifying the shapes down to a suitable level of detail so that the performance is OK.
We had the same issue with performance in our first implementation of filled maps in the above linked visualization. We started out with aprox. 40.000 points for the visualization of Denmark but rather quickly reduced it to approx 10.000 points. It was back in version 5.2 and we used a cube to blend and pre aggragate the data the geopoints with the measures. Works fine for an internal application but not good for export to tableau public ;-)
Looking forward to the time when tableau adds custom geocoding of polygons to the products :-)
Peter, not sure if this qualifies as 'custom geocoding on filled maps', but what Richard did here http://www.tableausoftware.com/support/forum/topic/lightning-density-maps might be of interest to you.