[R] Weibull distribution

Valentin Dimitrov vsdimitrov at yahoo.com
Fri Jul 21 17:03:07 CEST 2006


Dear Leaf,

I modified your code as follows:

gamma.fun <- function(mu,sd,start=100)    
 {
f.fn <- function(alpha) 
{abs(sd^2-mu^2/(gamma(1+1/alpha))^2*(gamma(1+2/alpha)-(gamma(1+1/alpha))^2))}
alpha <- optim(start, f.fn)
beta <- mu/gamma(1+1/alpha$par)
return(list=c(a=alpha$par,b=beta));
 }

Now it works properly.

First, I added an abs(). You tried to solve an
equation by means of the R-function optim(), which
finds a minimum. That's why you can find the solution
of f(x)=a through minimization of abs(f(x)-a).
Second, I deleted the optim-method BFGS from the
optim() function, because it is not appropriate in
this case. BFGS seeks to make the gradient (or here
the first derivative) zero and in your case f(x)
converges to a constant for big x, which means f'(x)
is approximately 0 for big x, which is why BFGS stops
almost immediately after the start value. The default
method of optim() ( Nelder and Mead ) is more
appropriate, since it does not need the first
derivative and works only with function values.

Best regards,
Valentin


--- Leaf Sun <leaflovesun at yahoo.ca> wrote:

> Hi all,
> 
> By its definition, the mean and variance of two-par.
> Weibull distribution are:
> 
>  
> 
>  
> 
>  (www.wikipedia.org)
> 
> 
> I was wondering, if given mean and sd. could we
> parameterize the distribution? I tried this in R.
> 
> gamma.fun <- function(mu,sd,start=100)    
> {
> f.fn <- function(alpha)
>
sd^2-mu^2/(gamma(1+1/alpha))^2*(gamma(1+2/alpha)-(gamma(1+1/alpha))^2)
> alpha <- optim(start, f.fn,method='BFGS')
> beta <- mu/gamma(1+1/alpha$par)
> return(list=c(a=alpha$par,b=beta));
> }
> 
> 
> But the problems come up here:
> 
> 1)  the return values of a and b are only related to
> the input mean, and nothing to do with the sd. For
> instance, when I apply a mean mu = 3 whatever I use
> sd=2, sd=4, the function returned the same scale and
> shape values.
> 
> > gamma.fun(3,4,10);
>        a        b 
> 5.112554 3.263178 
> 
> > gamma.fun(3,2,10);
>        a        b 
> 5.112554 3.263178 
> 
> 2) the start value determines the results: if I
> apply mean = 3, and sd=2, with a start of 10, it
> would return alpha close to 10, if I use a start =
> 100, it would return alpha close to 100.
> 
> > gamma.fun(3,2,10);
>        a        b 
> 5.112554 3.263178 
> 
> > gamma.fun(3,2,100);
>         a         b 
> 99.999971  3.017120 
> 
> Since I am not a statistician, I guess there must be
> some theoretical reasons wrong with this question.
> So I am looking forward to some correction and
> advice to solve these. Thanks a lot in advance!
> 
> Leaf
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> 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
>



More information about the R-help mailing list