Displaying 1 result from an estimated 1 matches for "bincombinations".
2005 Dec 15
1
millions of comparisons, speed wanted
...number of rows. I am searching for:
- ways to improve my code (if possible)
- ideas: create a C code for the slow parts of the code? use MySQL? other
ways?
As a toy example, having an input matrix called "input", my algorithm looks
like this:
## code start
ncolumns <- 6
input <- bincombinations(ncolumns) # from package e1071
# subset, let's say 97% of rows
input <- input[sample(2^ncolumns, round(2^ncolumns*0.97, 0), ]
minimized <- 1
while (sum(minimized) > 0) {
minimized <- logical(nrow(input))
to.be.compared <- combn2(1:nrow(input)) # from package combinat...