From: Eddy G. <eddy_lolo_at_hotmail.com>

Date: Tue 14 Jun 2005 - 20:15:53 EST

R-help@stat.math.ethz.ch mailing list

https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html Received on Tue Jun 14 20:49:07 2005

Date: Tue 14 Jun 2005 - 20:15:53 EST

Laurent 14 juin 12:02 afficher les options
De : "Laurent" <eddy_l...@hotmail.com> - Rechercher les messages de cet
auteur

Date : Tue, 14 Jun 2005 03:02:37 -0700

Local : Mar 14 juin 2005 12:02

Objet : bs() function of the splines package
Répondre | Répondre à l'auteur | Transférer | Imprimer | Message individuel
| Afficher l'original | Retirer | Signaler un cas d'utilisation abusive

Hello,

I'm implementing a function using non uniform B-Splines. Looking at the code of the bs() function, I realized that if the intercept was set to TRUE, the behavior of the function was the following (df is the number of degrees of freedom that I believe can be interpreted as the number of control points):

- Compute df- ord + 1 _internal_ knots (ord is the order of the spline)
- Add ord times the boundaries values at each extrem of the knots vector.
- Compute the design matrix on this vector.

This results in a design matrix with df+1 columns. The bs function then _removes_ the first column of the matrix (resulting as expected in a matrix with df columns).

I'm a bit confused, does anyone see a mathematical justification to this manipulation?

In this case, is it licit tu use df- ord + 2 internal knots and remove the last columns too?

Thanks a lot,

Laurent

R-help@stat.math.ethz.ch mailing list

https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html Received on Tue Jun 14 20:49:07 2005

*
This archive was generated by hypermail 2.1.8
: Fri 03 Mar 2006 - 03:32:34 EST
*