Hi all,
I have a huge sparse matrix, it's 360K*210K. Right now I have all the
nonzero elements and their indices, so I want to transform this information
into a sparse matrix so that I could pass the sparse matrix into svm
function.
I was planning to use sparse_matrix<- as.matrix.csr(v,nrow=360,ncol=300k),
where v would be a vector with a length = 360K*210K and with nonzero
elements in their places.
However I could not even create a vector of that size: vector
<-mat.or.vec( 1.08e+11,1), Error in numeric(nr) : vector size specified is
too large. So I thought to transform only part of the data at a time: so I
tried smaller size vector. As I found out, apparently the longest length
you can create through mat.or.vec( ) function is 2*10^9, but that means I
will have to break the data into 54 parts and transform each of them into a
long vector and then transform the vector into sparse matrix and then rbind
the 54 matrices into one . But I was worried that even if I could transform
that data into 54 smaller sparse matrix, I wouldn't be able to rbind them
together because of its huge size. I don't want to go through the arduous
process only to find that they can't be stacked together...
So I would really love to know how to transform this information into a
sparse matrix ? And even if it can be done, can svm function in package
e1071 handle a sparse matrix of this size ?
Thanks!!
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org 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.
