R help - Nov 2006 - goodness of fitting for a mixed distribution

Hi all!
  i have done a fitting of experimental data using optim function and
setting up my mixed distribution logliklehood (it is a linear
combination of a normal and a weibull distribution). How can i test
the goodness of fit? I want to be able to say that my distribution fit
the data at 90%, 95% or so on....

thanks in advance,
  nelson

Nelson,
I do not know of any formal hypothesis testing mechanism, but I recommend a
QQ plot regardless. Venables and Ripley (2002, 4th edition MASS) have a
detailed example for the mixed normal case (see page 440 in particular) that
could be adapted for your normal + weibull case. Or you could construct the
model quantile portion yourself.

Hope that helps,
Bill

-------------------------------
Bill Pikounis, PhD

Nonclinical Statistics
Centocor, Inc.
> -----Original Message-----
> From: r-help-bounces@stat.math.ethz.ch
> [mailto:r-help-bounces@stat.math.ethz.ch]On Behalf Of nelson -
> Sent: Monday, November 06, 2006 9:58 AM
> To: r-help@stat.math.ethz.ch
> Subject: [R] goodness of fitting for a mixed distribution
> 
> 
> Hi all!
>   i have done a fitting of experimental data using optim function and
> setting up my mixed distribution logliklehood (it is a linear
> combination of a normal and a weibull distribution). How can i test
> the goodness of fit? I want to be able to say that my distribution fit
> the data at 90%, 95% or so on....
> 
> thanks in advance,
>   nelson
> 
> ______________________________________________
> R-help@stat.math.ethz.ch 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.

	[[alternative HTML version deleted]]

Nelson,
I did a quick Google search on the phrase "goodness of fit mixture
models"
and found this thread. There seems to be some good info in there:

http://www.biostat.wustl.edu/archives/html/s-news/2000-07/msg00166.html

Hope that helps,
Bill

-------------------------------
Bill Pikounis, PhD
Nonclinical Statistics
Centocor, Inc.
> -----Original Message-----
> From: r-help-bounces at stat.math.ethz.ch
> [mailto:r-help-bounces at stat.math.ethz.ch]On Behalf Of nelson -
> Sent: Monday, November 06, 2006 11:20 AM
> To: r-help at stat.math.ethz.ch
> Subject: Re: [R] goodness of fitting for a mixed distribution
> 
> 
> Thanks,
>   i wonder if I can use a kolmogorov smirnov test or a similar test
> (but what test?). I use QQ-plot but the people i'm working to asked me
> numbers... :) and not sum of square error.
> 
> The told about chi-square test and Kolmogorov Smirnov Test...
> 
> I produce QQ- plots, but they doesn't like them... while i see theme
> very descriptive...
> 
> any advice? They only want to say "oh, my mixed distribution fits data
> at 90%".... :(
> 
> 
> nelson
> 
> ______________________________________________
> R-help at stat.math.ethz.ch 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.
>