Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: RE: RE: Fitting a t distribution

From   "Nick Cox" <>
To   <>
Subject   st: RE: RE: RE: Fitting a t distribution
Date   Thu, 14 Apr 2005 22:25:32 +0100

In your first email you clearly identify that the parameters are df, 
scale and shift. You are not obliged to specifying starting
values in an -ml- program, if that is your question. 

I am not aware that any program pre-exists to do 
what you want. But a program like -gammafit- gives
you a template that just needs modifying for your 


Indranil Majumdar
> I am trying to fit the t-distribution to my data. Hence, I 
> don't know the
> parameters to start with because that's what I want to estimate. I was
> wondering if something is available for the t along the lines 
> of 'gammafit',
> 'betafit' available on the ssc website.

Nick Cox
> I am puzzled by this question. But yes, 
> if you wish to maximize a likelihood function, 
> -ml- is exactly the right command. And Al 
> Feiveson was exactly right in pointing you 
> towards -tden()-. 

Indranil Majumdar
> > I am trying to maximize a log-likelihood function involving 3 
> > parameters (in
> > particular, it is the log-likelihood of the t distribution 
> > with a shift and
> > scale parameter). The documentation for the -ml model- 
> > command requires an
> > equation to be specified. In my case I am not sure what that 
> > equation ought
> > to be. My objective is to 'search' for the optimal values of the 3
> > parameters - degrees of freedom, shift parameter and scale 
> > parameter that
> > maximize the log-likelihood function.
> > Is the -ml- command appropriate for this purpose? Could 
> > somebody clarify?

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index