[R] About fitdistr

[Prof Brian Ripley]

On Wed, 11 Oct 2006, Caroline TRUNTZER wrote:

> Dear all,
> I would like to estimate the parameters of a mixture of two beta
> binomial distributions. I saw the function fitdistr from the MASS
> library could be a solution but I have some difficulties. Does anyone
> know how to specify to fitdistr another function as these which are
> predefined? Or is there any other solution to estimate the parameters?

>From the help:

  densfun: Either a character string or a function returning a density
           evaluated at its first argument.

Just use the 'or' version.

Alternatively, chapter 16 of MASS has several ways to fit a mixture of two 
normals, which you could generalize.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595