What are good ways to (automatedly) plot (or othewise present) spatial data that
is very skewed? E.g.
http://tinyurl.com/dn2VerySkewedSpatialData
produced with fields:image.plot. There are obviously outliers :-) which may or
may not be "for real." At an exploratory stage in the investigation, I
don't want to prejudge, I just want to show both
1 where the outliers are. This is not such a problem in the plot
shown, but is often a problem in others, where the outliers are
just a few pixels/gridcells. I could highlight them by, e.g.,
* importing the output into an editable format
* drawing circles around the outliers
* exporting back to distributable format (e.g., PDF)
but that would be Real Work; automated solution much preferred!
2 how the rest of the data is distributed. Normally I'd do this with a
quantile plot, and in fact that's what I'm doing. The jumble at the
bottom of the legend are the quantiles overwriting each other :-)
Hence my appeal to the vast experience of the group regarding Best Practice, not
least because I can't believe I'm the first person to have this problem.
TIA, Tom Roche <Tom_Roche at pobox.com>
