Hey! That's a map of Piscataway! I recognize it!
As for your question, uploading a sample workbook would really help here. The granularization of the pie charts is probably a function of the dimensions you have on the sheet. It might take table calcs or LODs to roll them up properly. And (at least as far as I am concerned) I would have to play with a working example to figure that out.
Yes I agree with needing the workbook to try out a few things. Having said that I'm wondering if something as simple as moving the green pill from the size mark to the pie angle would solve the issue.
No workbook necessary - instead of giving us the workbook, give us a subset of the data, a screen shot of a subset of data is all we need. Likely this is a primary key or dimension being subtly off, generating a duplicate of your data. Lat long can be fun to play with, good job so far!
MY MONEY is on one of those extra dimensions has the building or building number spelled wrong, or with extra letters, spaces, words, etc.
Here's what you're aiming for:
Filter down to a small segment of data that you know and understand.
In your screenshot there are several dimensions that are currently impacting your granularity.
The 2 shapes in one is two events, that do not share a common characteristic, from a level of detail perspective.
*The not so nerdy explanation*
I would focus on how you're generating your lat/long VS what your most granular detail is.
1. Filter down to a small segment.
2. Highlight everything - open the data view by hovering over a mark.
3. Look at the row level values and see where the duplicate is coming from.
Two circles stacked means two events in the same spot. It's not generating a PIE because the two values do not equal - likely because it's a duplicate value and there is a dimension causing this problem, or maybe it's your LAT/LONG having a subtle difference.
Start here - this is 3 dimensions, which generates 3 slices of the data.
Also if your lat long is subtly off, even .00001 off, it can stack on itself and then not generate your pie chart. Often times corporations see this due to data integrity concerns. Meaning, someone typed something wrong, or something along these lines
PS Don't worry about sending workbooks, just send over a screenshot of the data, explained here.
Example: highlight a few marks, and then click the far left top button. Looks like stacked lines broken by columns
Thanks for all this help! Truly appreciate it.. and Piscataway was guessed correctly .
It does seem to be a latitude/longitude issue.. When I switch the calculation to median over average, it is turning out okay.
The weird part is that the longitude and latitude are from the building level - so the #'s are identical (unless tableau does weird things when calculating the average?) Any knowledge on how that is done? Or maybe how to format the lat/long before bringing the data in?
**Update: there are a LOT of options, here are two cool options that are user friendly, and you could be 'aggregating' up to a rounded lat/long.
For example; Latitude has many decimals, what if you could round up, X.XXXXXX a few at a time will offer a new 'higher level view' of your current data.
Which offers a 'drill down' from a visual perspective.
Then you can drill to the 'areas' causing the most 'pain' or 'success' depending on your data and analysis needs.
http://powertoolsfortableau.com/tools/drawing-tool I mapped out an non-profit HOA division once, using this tool.
It can be very powerful if the data coming in is CLEAN.
Here's the brainstorm session in powerpoint to help me understand how to get there:
And here's the scary granular view.
There's a lot that can be done with mapping, I would recommend 'drawing it out' and then go from there.