1

I want to use PCA for a sparse matrix(which represents images),420000*1784 size ,about 2% non-zero entries,and the principal function tells me that there is a lot NA value in its egien values and egien matrix ,what is the reason for that? is there another function to deal with such situations ?Or I need to write a function to do this ?

suemi
  • 69
  • 5
  • Please make it reproductible ? ( == code + data ) – dickoa Aug 07 '13 at 05:29
  • Have you done any prior search on the subject? I found this http://stats.stackexchange.com/questions/35185/dimensionality-reduction-svd-or-pca-on-a-large-sparse-matrix that seems relevant. – Roman Luštrik Aug 07 '13 at 06:32
  • 1
    Besides [`irlba`](http://stackoverflow.com/questions/17352771/svd-of-very-large-matrix-in-r-program/17354248), you can also use `igraph::arpack`, as in [this question](http://stackoverflow.com/questions/16966748/computing-eigenvectors-of-a-sparse-matrix-in-r/). – Vincent Zoonekynd Aug 07 '13 at 08:20
  • 1
    Thank you for your help.And thanks to the links given by Roman Luštrik ,I know how to deal with my problem – suemi Aug 08 '13 at 04:49

0 Answers0