R - svd() function - infinite or missing values in 'x'

Question

I am constantly getting this error. I am sure the matrix does not have any non-numeric entries. I also tried imputing the matrix, did not work.

Anyone know what the error might be?

fileUrl <- "https://dl.dropboxusercontent.com/u/76668273/kdd.csv";
download.file(fileUrl,destfile="./kdd.csv",method="curl");
kddtrain <- read.csv("kdd.csv");
kddnumeric <- kddtrain[,sapply(kddtrain,is.numeric)];
kddmatrix <- as.matrix(kddnumeric);
svd1 <- svd(scale(kddmatrix));

Snoopy

In future, a small reproducible example (and not something that requires downloading) is appreciated. – Prophet

Answer 1

You have columns composed of all zeroes. Using scale on a column of all zeroes returns a column composed of NaN. To solve this, remove columns where you have all zeroes (svd will not reveal anything new about them), or replace NaN columns with zero after using scale.

Reproducible example:

mat <- matrix(c(1,2,3,0,0,0,2,4,6,5,12,13),nrow = 3)
     # [,1] [,2] [,3] [,4]
# [1,]    1    0    2    5
# [2,]    2    0    4   12
# [3,]    3    0    6   13
scale(mat)
     # [,1] [,2] [,3]       [,4]
# [1,]   -1  NaN   -1 -1.1470787
# [2,]    0  NaN    0  0.4588315
# [3,]    1  NaN    1  0.6882472
# attr(,"scaled:center")
# [1]  2  0  4 10
# attr(,"scaled:scale")
# [1] 1.000000 0.000000 2.000000 4.358899
svd(mat) #fine
svd(scale(mat)) # not fine

Prophet

thanks a lot, perfect answer. a simple line kddmatrix <- kddmatrix[,colSums(kddmatrix != 0) != 0] before applying svd solved it. – Snoopy