R help - Apr 2012 - Looking for the name of a certain kind of quantile plot

Hi,
While playing with quantile-quantile plots, I wrote up some code which 
plots something strangely different.  Here's the pseudocode:

testhist <- hist(sample_data)
refhist <- hist(rnorm(n, mean=0,sd=1))  # for some large-ish n
cumtest <- cumsum(testhist)
cumref <- cumsum(refhist)

plot(cumref,cumtest)

This produces a straight line of slope 1 for a sample with the same 
parameters as the reference sample, and produces S-curves for samples 
with different sigmas.  A sample with nonzero mean looks almost 
exponential (or logarithmic, depending on the sign of the mean).

So my question is: is there a name for this sort of plot,  and is it of 
any real use in statistical analysis?

thanks.
Carl


-- 

Sent from my Cray XK6
"Quidvis recte factum, quamvis humile, praeclarum."

Carl Witthoft <carl at witthoft.com> writes:
> Hi,
> While playing with quantile-quantile plots, I wrote up some code which
> plots something strangely different.  Here's the pseudocode:
>
> testhist <- hist(sample_data)
> refhist <- hist(rnorm(n, mean=0,sd=1))  # for some large-ish n
> cumtest <- cumsum(testhist)
> cumref <- cumsum(refhist)
>
> plot(cumref,cumtest)

Sounds like a 'pp-plot'. See

      http://en.wikipedia.org/wiki/P-P_plot

Why not provide reproducible code?

Chuck
>
> This produces a straight line of slope 1 for a sample with the same
> parameters as the reference sample, and produces S-curves for samples
> with different sigmas.  A sample with nonzero mean looks almost
> exponential (or logarithmic, depending on the sign of the mean).
>
> So my question is: is there a name for this sort of plot,  and is it
> of any real use in statistical analysis?
>
> thanks.
> Carl
-- 
Charles C. Berry                            Dept of Family/Preventive Medicine
cberry at ucsd edu			    UC San Diego
http://famprevmed.ucsd.edu/faculty/cberry/  La Jolla, San Diego 92093-0901

Thanks, Chuck.  That looks about right.

And I would have provided repro code, but I was forced to write it in 
Matlab on account of corporate rules, and haven't translated to R yet.


Carl Witthoft <carl at witthoft.com> writes:

 > Hi,
 > While playing with quantile-quantile plots, I wrote up some code which
 > plots something strangely different.  Here's the pseudocode:
 >
 > testhist <- hist(sample_data)
 > refhist <- hist(rnorm(n, mean=0,sd=1))  # for some large-ish n
 > cumtest <- cumsum(testhist)
 > cumref <- cumsum(refhist)
 >
 > plot(cumref,cumtest)


Sounds like a 'pp-plot'. See

       http://en.wikipedia.org/wiki/P-P_plot

Why not provide reproducible code?

Chuck

 >
 > This produces a straight line of slope 1 for a sample with the same
 > parameters as the reference sample, and produces S-curves for samples
 > with different sigmas.  A sample with nonzero mean looks almost
 > exponential (or logarithmic, depending on the sign of the mean).
 >
 > So my question is: is there a name for this sort of plot,  and is it
 > of any real use in statistical analysis?
 >
 > thanks.
 > Carl
-- 

Sent from my Cray XK6
"Quidvis recte factum, quamvis humile, praeclarum."