An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20090104/6fc01c52/attachment-0001.pl>
Customized LDA (MASS) object plot
3 messages · Rodrigo Aluizio, Brian Ripley
Hmm: what do you want to do that a customized panel function cannot do? plot.lda does not accept type="n" precisely because it does support a panel function to plot within the axes it has set up. Your claim that 'they don?t let me personalize them' is simply untrue. If you want to see the source code, just type 'MASS:::plot.lda'. Futhermore, MASS is support software for a book, and the book does have examples of customized LDA plots!
On Sun, 4 Jan 2009, Rodrigo Aluizio wrote:
Hi R experts. I performed an Linear Discriminant Analysis (lda) and now I want to plot the first two axes (LDA1 and LDA2). Well MASS package have the plot.lda and pairs.lda to do that. But, they don?t let me personalize them, once they don?t accept the type=?n? plot. So I start looking at the lda object properties, trying to found out my groups (n=4) axis coordinates, curiously I couldn?t. There is only the variables axis coordinates and some analysis results. If I can reach this information, I?ll be able to create a customized graph using the normal plot(), points() and par(). So the question is: How could I reach these axis coordinates inside my object (supposing they are there somewhere, once the plot.lda and pairs.lda are able to plot them)? Thanks in advance.
___________________________________ MSc. <mailto:r.aluizio at gmail.com> Rodrigo Aluizio Centro de Estudos do Mar/UFPR Laborat?rio de Micropaleontologia Avenida Beira Mar s/n - CEP 83255-000 Pontal do Paran? - PR - BRASIL Fone: (0**41) 3455-1496 ramal 217 Fax: (0**41) 3455-1105 [[alternative HTML version deleted]]
Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UK Fax: +44 1865 272595
Thanks for the enlightenment Prof. Ripley. I guess I've to start learning quite more about panel function. Once I'm pointed to the right way I'll probably be able to do what I want. Sorry for the mistake, I totally misses the panel function possibilities. Rodrigo. -----Mensagem original----- De: Prof Brian Ripley [mailto:ripley at stats.ox.ac.uk] Enviada em: domingo, 4 de janeiro de 2009 11:15 Para: Rodrigo Aluizio Cc: R Help Assunto: Re: [R] Customized LDA (MASS) object plot Hmm: what do you want to do that a customized panel function cannot do? plot.lda does not accept type="n" precisely because it does support a panel function to plot within the axes it has set up. Your claim that 'they don?t let me personalize them' is simply untrue. If you want to see the source code, just type 'MASS:::plot.lda'. Futhermore, MASS is support software for a book, and the book does have examples of customized LDA plots!
On Sun, 4 Jan 2009, Rodrigo Aluizio wrote:
Hi R experts. I performed an Linear Discriminant Analysis (lda) and now I want to plot
the
first two axes (LDA1 and LDA2). Well MASS package have the plot.lda and pairs.lda to do that. But, they don?t let me personalize them, once they don?t accept the type=?n? plot. So I start looking at the lda object properties, trying to found out my groups (n=4) axis coordinates,
curiously
I couldn?t. There is only the variables axis coordinates and some analysis results. If I can reach this information, I?ll be able to create a customized graph using the normal plot(), points() and par(). So the question is: How could I reach these axis coordinates inside my object (supposing they are there somewhere, once the plot.lda and
pairs.lda
are able to plot them)? Thanks in advance.
___________________________________ MSc. <mailto:r.aluizio at gmail.com> Rodrigo Aluizio Centro de Estudos do Mar/UFPR Laborat?rio de Micropaleontologia Avenida Beira Mar s/n - CEP 83255-000 Pontal do Paran? - PR - BRASIL Fone: (0**41) 3455-1496 ramal 217 Fax: (0**41) 3455-1105 [[alternative HTML version deleted]]
Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UK Fax: +44 1865 272595