On 3/2/07, Bruno C. <[EMAIL PROTECTED]> wrote: > 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 >
Design an experiment. Start with a small matrix and keep increasing until it gets too big. You will want to control for other factors like what other objects are in memory. I don't think this is particularly more elegant than splitting the matrix into pieces that you know will fit into memory. And will certainly take much more time. Hadley ______________________________________________ [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.
