Page 1 of 1
t-distribution
Posted: Fri Sep 02, 2011 1:38 pm
by jolly
I have to generate errors which are distributed as a Student t-distribution and I use function @rtdist(v). How I should choose value of degrees "v"?
Re: t-distribution
Posted: Fri Sep 02, 2011 1:51 pm
by startz
That would depend on what you're trying to do...
Re: t-distribution
Posted: Fri Sep 02, 2011 1:57 pm
by jolly
I'm creating a regresion, and it's is said that errors e~t(v).and nothing about "v" ..
Re: t-distribution
Posted: Fri Sep 02, 2011 2:04 pm
by startz
Then there's no way to know what value to use, other than it has to be positive. If the value is large, say 120+ you're essentially getting normal errors. Since whoever made up the assignment specified t-, they probably want a smallish integer.