| Re: non-integer binsize [message #43857] |
Thu, 05 May 2005 13:23 |
patrick.gatlin
Messages: 7
Registered: June 2004
|
Junior Member |
|
|
Thank you David and JD for the help and some insight into the
histogram. I wanted to make sure I was not misinterpreting the hist_2d
output.
At least you didn't have to break out the red pen on this one David.
;-)
Thanks again!
|
|
|
|
| Re: non-integer binsize [message #43862 is a reply to message #43857] |
Thu, 05 May 2005 09:44  |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
On Wed, 04 May 2005 17:28:04 -0700, patrick.gatlin wrote:
> This may be a very elementary question, but I was wondering if the
> histogram function can accept a binsize < 1?
>
> I am using the hist_2d function to determine the density field of some
> lat/lon locations in a dataset. These values range 4 deg lat and 4 deg
> lon. In order to determine the # of points per square km, I have setup
> the hist_2d function as follows:
>
> IDL> result=hist_2d(lon,lat,max1=lonmax,min1=lonmin,max2=latmax,$
> min2=latmin,bin1=1/111.,bin2=1/111.)
>
> where the size of bin1 and bin2 are the approximate conversions from km
> to degrees (1km=.09 deg).
> Since hist_2d needs two bins, then is the result equal the number of
> points within a box that is 1/111. long on all sides? Thus am assuming
> result will give me the # of points per square kilometer. Is this a
> correct assumption based upon the way hist_2d works?
If you peek inside HIST_2D (or HIST_ND for that matter), you'll see that
the way non-integer bin-sizes work is by first converting the data to some
suitable integer representation over the range of interest, typically
where the binsize is normalized to 1, and then taking the histogram. This
is also quite likely how HISTOGRAM itself works internally. You might
also like to read up on the rounding issues you can have when working
with floating point histograms:
http://www.dfanning.com/math_tips/razoredge.html
JD
|
|
|
|
| Re: non-integer binsize [message #43875 is a reply to message #43862] |
Wed, 04 May 2005 17:54 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
patrick.gatlin@msfc.nasa.gov writes:
> This may be a very elementary question, but I was wondering if the
> histogram function can accept a binsize < 1?
Since the example in the on-line help for HIST_2D uses a
bin size of 0.02, I think you might be in good shape here. ;-)
> I am using the hist_2d function to determine the density field of some
> lat/lon locations in a dataset. These values range 4 deg lat and 4 deg
> lon. In order to determine the # of points per square km, I have setup
> the hist_2d function as follows:
>
> IDL> result=hist_2d(lon,lat,max1=lonmax,min1=lonmin,max2=latmax,$
> min2=latmin,bin1=1/111.,bin2=1/111.)
>
> where the size of bin1 and bin2 are the approximate conversions from km
> to degrees (1km=.09 deg).
> Since hist_2d needs two bins, then is the result equal the number of
> points within a box that is 1/111. long on all sides? Thus am assuming
> result will give me the # of points per square kilometer. Is this a
> correct assumption based upon the way hist_2d works?
This explanation reminds me of editing my middle son's
big essay last night (sigh...), but I think you've probably
got the gist of it, anyway. :-)
I think a better way of looking at it is that the box is
1/111. times the number of bins between the min and max values,
but I think you are correct that the numbers you get back
represent the number of points in a square kilometer.
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
