[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Octave-bug-tracker] [bug #35378] undefined variable in ols.m
From: |
Jordi Gutiérrez Hermoso |
Subject: |
Re: [Octave-bug-tracker] [bug #35378] undefined variable in ols.m |
Date: |
Thu, 26 Jan 2012 10:52:47 -0500 |
2012/1/26 Michael Creel <address@hidden>:
> I agree that using the rank is better. As a note, though, the maximum
> likelihood estimator divides by nr, with no degrees of freedom correction.
> The OLS estimator divides by (nr - rank). In most situations one would not
> use a regressor matrix of less than full rank, because the OLS parameter
> estimates are not unique in that case.
This is what the Moore-Penrose pseudoinverse is for. The ols
documentation says this.
- Jordi G. H.