On 06/11/2014 11:29 PM, Robert C. Helling wrote:
On 11 Jun 2014, at 20:15, Lutz Vieweg<[email protected]> wrote:
I guess if somebody really wants to visualize an informative
statement on "How deep do I usually dive?" then a histogram
would be the best way to do it:
"Depth" on the X-Axis, "Total time spent in that depth" on the
Y-Axis, and then 20-or-so bars reaching up from the X-Axis.
one could make that even better defined (in the sense of: upon any ascent or
descent I only spend an infinitesimal
amount of time at each depth. So one would need to introduce some binning like
“in depths between x and x+2m” and
the result would depend on the binning) one you use the integral, i.e. plot
“Total time spent at depths deeper than x” which is independent of binning and
finite.
While I totally agree that plotting the integral gets rid of the "binning"
arbitrariness, I'd assume such a plot to be much less easy to understand
for the average user.
Assume for example that you spend a lot of time in depths shallower than
10m and also considerable time in depths deeper than 20m, for whatever
reason (e.g. because your favorite dive spot is a wreck at 20m-40m), but
relatively little time in depths from 10m to 20m. In a classic histogram
plot, such a preference would be easy to spot at first sight. In an
integral plot, you'd have to deduct your seldom stays from 10m to 20m
from the slope of the graph in that interval, which is much less intuitive.
People are very used to the "histogram" type of charts, and if you plot
a reasonably high amount of bars, the relevance of their precise interval
borders is kind of neglectable. And if there are enough depth samples to work
with,
you could even draw one bar per pixel without to much risk of "quantization
noise".
Regards,
Lutz Vieweg
_______________________________________________
subsurface mailing list
[email protected]
http://lists.hohndel.org/cgi-bin/mailman/listinfo/subsurface
