R help - Sep 2008 - library/function that estimates parameters of well known distributions from empirical data?

I found this a few months ago, but for the life of me I can't remember what
the function or package was, and I have had no luck finding it this week.

I have found, again, the functions for working with distributions like
Cauchy, F, normal, &c., and ks.test, but I have not found the functions for
estimating the distribution parameters given a vector of values.

What I need to do is estimate the distribution parameters for each candidate
distribution, and then test to see which gives the best fit to the data.

I want to examine the question, given this dataset (which may have thousands
of records), does the normal or cauchy distribution fit the data best, and
which what parameters.  It will not be known a priori whether or not the
most appropriate distribution is non-central, though we do know that often
(not always) values of medium size in absolute value are more often positive
than negative and that very large values are more often negative than
positive.

Could someone please give me a gentle reminder of the package and
function(s) I ought to be examining?

Thanks

Ted
-- 
View this message in context:
http://www.nabble.com/library-function-that-estimates-parameters-of-well-known-distributions-from-empirical-data--tp19323700p19323700.html
Sent from the R help mailing list archive at Nabble.com.

Ted Byers <r.ted.byers <at> gmail.com> writes:
> 
> 
> I found this a few months ago, but for the life of me I can't remember
what
> the function or package was, and I have had no luck finding it this week.
> 
> I have found, again, the functions for working with distributions like
> Cauchy, F, normal, &c., and ks.test, but I have not found the functions
for
> estimating the distribution parameters given a vector of values.
> 
  Look at the fitdistr function in the MASS package.  Consider
AIC comparisons for ranking the fits to these non-nested models.

  good luck
   Ben Bolker