bigml is for generalized regression model. I need to use lars ans svm. Anway I am perfectly able to train my model: my training matrix is not so big. The big matrix is the test one then I can split it by rows into several matrices without affecting the results. The point is that I wanted to split it in an elegant way, mainimizing the needed submatrices wrt memory. But I have the impression I am asking R too much :D
> Did you consider biglm package. I did not use it myself but from its > description it can be used for linear models on objects that do not > fit into memory. > > HTH > Petr > > > On 2 Mar 2007 at 13:12, Bruno C. wrote: > > Date sent: Fri, 2 Mar 2007 13:12:07 +0100 > From: "Bruno C." <[EMAIL PROTECTED]> > To: "petr.pikal" <[EMAIL PROTECTED]> > Copies to: r-help <[EMAIL PROTECTED]> > Subject: Re: [R] Reformulated matrices dimensions limitation > problem > > > You are right. and I am aware of that. > > > > This is what I need to do: > > load a regression model > > load the bigget posible matrix > > do prediction on this matrix > > > > the first and 3rd step will not need to much memory... > > > > and I would really appreciate this degree of introspection from R: it > > would be great if there would be a package that could simulate the > > amount of memory needed from a process But I don't pretend that much, > > this is why I am asking only about a function able to build a matrix > > matrix, with size based on memory available... > > > > > > > > > > > Hi > > > > > > creating a biggest possible matrix does not automaticaly mean you > > > can do some computation with it. > > > > > > So the size will depend partly on what you want to do with it. I > > > presume that you do not want only to create a matrix just for > > > pleasure to be able to. > > > > > > Cheers > > > Petr > > > > > > > > > On 2 Mar 2007 at 10:15, Bruno C. wrote: > > > > > > Date sent: Fri, 2 Mar 2007 10:15:33 +0100 > > > From: "Bruno C." <[EMAIL PROTECTED]> > > > To: "r-help" <[EMAIL PROTECTED]> > > > Subject: [R] Reformulated matrices dimensions limitation > > > problem > > > > > > > First I wanted to thank both Marc Schwartz Greg Snow and for their > > > > reply. > > > > > > > > Then I needed to add a level of complexity to the problem. > > > > I would be able to create the biggest possible matrix. > > > > > > > > In other way does it exist a method to ask smthing like the > > > > following : > > > > > > > > max number of rows for a matrix if column=x? > > > > > > > > Thank you > > > > > > > > > > > > > > > > > > > > > > > > ------------------------------------------------------ > > > > Passa a Infostrada. ADSL e Telefono senza limiti e senza canone > > > > Telecom http://click.libero.it/infostrada2marz07 > > > > > > > > ______________________________________________ > > > > [email protected] mailing list > > > > https://stat.ethz.ch/mailman/listinfo/r-help > > > > PLEASE do read the posting guide > > > > http://www.R-project.org/posting-guide.html and provide commented, > > > > minimal, self-contained, reproducible code. > > > > > > Petr Pikal > > > [EMAIL PROTECTED] > > > > > > > > > > > > ------------------------------------------------------ > > Passa a Infostrada. ADSL e Telefono senza limiti e senza canone > > Telecom http://click.libero.it/infostrada2marz07 > > > > ______________________________________________ > > [email protected] mailing list > > https://stat.ethz.ch/mailman/listinfo/r-help > > PLEASE do read the posting guide > > http://www.R-project.org/posting-guide.html and provide commented, > > minimal, self-contained, reproducible code. > > Petr Pikal > [EMAIL PROTECTED] > > ------------------------------------------------------ Passa a Infostrada. ADSL e Telefono senza limiti e senza canone Telecom http://click.libero.it/infostrada2marz07 ______________________________________________ [email protected] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
