R help - May 2011 - Using error in histograms

Hello!

I am trying to produce a histogram of measurement data (orientation of
archaeological structures) that are a subject to measurement error.
The normal histogram just computes frequencies, but does not take into
account that a particular value is spread over a range of values (in
my case the spread is different for reach measurement and is larger
than the bin size).

The closest approach is kernel density estimation (in the image is a
comparison of histogram and KDE):
http://en.wikipedia.org/wiki/Kernel_density_estimation
http://en.wikipedia.org/wiki/File:Comparison_of_1D_histogram_and_KDE.png
However in my case the kernel size is different for each value. I
wrote a program in IDL that performs the plotting, but am just
wondering if such a function is available in R. Basically it is a
problem of summing (and later plotting) several data distributions.

I would appreciate also any hint to a book that might be dealing with
the problem. I am not an expert in statistics and I might not be using
use the correct terminology in web searches.

Regards,

Kristof

The logspline package does density estimation in a different way than KDE, but
it does allow for interval censored data (I know this value is between a and b,
but not where in that range) using the oldlogspline function.  This may be what
you are looking for.

-- 
Gregory (Greg) L. Snow Ph.D.
Statistical Data Center
Intermountain Healthcare
greg.snow at imail.org
801.408.8111

> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-
> project.org] On Behalf Of Kristof Ostir
> Sent: Thursday, May 05, 2011 4:16 AM
> To: r-help at r-project.org
> Subject: [R] Using error in histograms
> 
> Hello!
> 
> I am trying to produce a histogram of measurement data (orientation of
> archaeological structures) that are a subject to measurement error.
> The normal histogram just computes frequencies, but does not take into
> account that a particular value is spread over a range of values (in
> my case the spread is different for reach measurement and is larger
> than the bin size).
> 
> The closest approach is kernel density estimation (in the image is a
> comparison of histogram and KDE):
> http://en.wikipedia.org/wiki/Kernel_density_estimation
> http://en.wikipedia.org/wiki/File:Comparison_of_1D_histogram_and_KDE.pn
> g
> However in my case the kernel size is different for each value. I
> wrote a program in IDL that performs the plotting, but am just
> wondering if such a function is available in R. Basically it is a
> problem of summing (and later plotting) several data distributions.
> 
> I would appreciate also any hint to a book that might be dealing with
> the problem. I am not an expert in statistics and I might not be using
> use the correct terminology in web searches.
> 
> Regards,
> 
> Kristof
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-
> guide.html
> and provide commented, minimal, self-contained, reproducible code.