Choose Max Value Size for Proportional Symbol

11-16-2010 03:25 PM
Status: Open
New Contributor III

Many times I would like to map point data based on a value proportional to the size, but the largest circle sizes take over the map.  I would like to be able to set both the Min Value AND the Max Value, and then have the software scale the symbols based on the data. 

I should probably call this a 'scaled symbol' since I realize it is not technically proportional when you add this option.  This is also an issue when you have data that is similar.  I have a file with point values ranging from 140-215, and the points all look almost the same size.  This would be much better than having to set groups in a graduated symbol since it would still show variation in size within the groups.   
I understand the problem, but I'm not sure about how to solve it.  If you provide a min and a max, then you are limiting proportionality.  That is, the reason something takes over the screen is because it is that much larger relative to the other data points.  In other words, it is being accurately displayed, proportionately.

Adding both min and max makes proportional more of a hybrid between graduated symbols and proportional.

As soon as you clip, some of the data is not accurately represented.  To maintain proportionality, you have to specify either min or max, but not both.

However, maybe there would be some creative ways to use secondary attributes (e.g., saturation?) to indicate scale beyond the clipping range.  Saturation could be automatically increased or decreased between the clipping point and the max data value.  Obviously, one could label just the large values above the max clipping point, but in many datasets this would produce very crowded and unattractive results.  I'm not sure what other options would work.

Another alternative would be to use a non-linear proportionality scheme (log?) that fits things between the min and max along some kind of curve.  However, this might not be intuitive because the primary attribute of size would not correspond directly.  That is, a point symbol twice the size (area) of another point symbol would not represent data that was twice as great.  The difference would be dependent upon the fitting curve, which would probably not be intuitive..

So, I'm promoting the concept, but with some questions about possible implementations and side-effects.