CODEWITHSUNDEEP
rcombinations
RTrain3k
RTrain3k

Reputation: 867

Add conditions to expand grid in R?

Is there anyway to incorporate conditions/restrictions into expand.grid in R?

As an example, how would I prevent the combination c(1,1,1) and c(2,2,2) from appearing while the grid is being created? Is there a scalable solution that would work with many restrictions? My grid is large and my memory cannot handle it, e.g Error: cannot allocate vector of size 32.0 Gb, so subsetting after the grid is created is not an option.

expand.grid(A = rep(1:2),
            B = rep(1:3),
            C = rep(1:4))

Code below does not work:

expand.grid(A[!((A==1 & B==1 & C==1) & (A==2 & B==2 & C==2))] = rep(1:2),
            B[!((A==1 & B==1 & C==1) & (A==2 & B==2 & C==2))] = rep(1:3),
            C[!((A==1 & B==1 & C==1) & (A==2 & B==2 & C==2))] = rep(1:4))

as.data.frame(expand.grid(A = rep(1:2),
                          B = rep(1:3),
                          C = rep(1:4)))[!((A==1 & B==1 & C==1) & (A==2 & B==2 & C==2))]

Thanks!

Upvotes: 3

Views: 804

Answers (2)

Parfait
Parfait

Reputation: 107767

Simply assign the expand.grid data frame and then run needed logic which you can shorten by comparing equality across all columns (A==C not needed per the transitive property of equality):

d1 <-  expand.grid(A = rep(1:2),
                   B = rep(1:3),
                   C = rep(1:4))

d1 <-  with(d1, d1[!(A==B & B==C),])

Upvotes: 1

akrun
akrun

Reputation: 887901

If we create the dataset

d1 <-  expand.grid(A = rep(1:2),
        B = rep(1:3),
        C = rep(1:4))

Can use rowSums to check if the data is equal to the first column

d1[!!rowSums(d1 != d1[,1]),]

Upvotes: 1

Related Questions