transposeBigData: Transpose a big matrix or data frame
Description
This transpose command partitions a big matrix (or data frame) into blocks and applies the t() function to
each block separately.
Usage
transposeBigData(x, blocksize = 20000)
Arguments
x
a matrix or data frame
blocksize
a positive integer larger than 1, which determines the block size. Default is 20k.
Value
A matrix or data frame (depending on the input x ) which is the transpose of x.
Details
Assume you have a very large matrix with say 500k columns. In this case, the standard transpose function of
R t() can take a long time. Solution: Split the original matrix into sub-matrices by dividing the
columns into blocks. Next apply t() to each sub-matrix. The same holds if the large matrix contains
a large number of rows. The function transposeBigData automatically checks whether the large matrix
contains more rows or more columns. If the number of columns is larger than or equal to the number of rows
then the block wise splitting will be applied to columns otherwise to the rows.
References
Any linear algebra book will explain the transpose.
# NOT RUN {x=data.frame(matrix(1:10000,nrow=4,ncol=2500))
dimnames(x)[[2]]=paste("Y",1:2500,sep="")
xTranspose=transposeBigData(x)
x[1:4,1:4]
xTranspose[1:4,1:4]
# }