CODEWITHSUNDEEP
rdummy-variable
user3285362
user3285362

Reputation: 31

R: How can I convert an ordered factor to dummy variables?

For example. A factor that with ordered levels

[1] 0 0 6 6 3 4 Levels: 0 < 1 < 2 < 3 < 4 < 5 < 6

should be converted to

ti0 ti1 ti2 ti3 ti4 ti5 ti6 1 0 0 0 0 0 0 1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 1 1 1 0 0

I have looked at packages like dummies and functions like model model.matrix but cant get to a solution.

Upvotes: 0

Views: 784

Answers (2)

JWilliman
JWilliman

Reputation: 3883

If your purpose is to create a variable with a custom contrast matrix, and in your case it looks like you are creating a cumulatively coded ordinal variable as described here, you can do the following.

x <- factor(c("0", "0", "6", "6", "3", "4"), levels = 0:6, ordered = TRUE)

# Create custom contrast function
contr.cum <- function(x, base = 1L) 
{
  dmns <- levels(x)
  n <- length(dmns)
  contr <- array(diag(n), dim = c(n,n), dimnames = list(dmns, dmns))
  contr[lower.tri(contr)] <- 1
  contr <- contr[, -base, drop = FALSE]
  contr
}

# Apply custom function to variable
contrasts(x) <- contr.cum(x)

# View model matrix
model.matrix(~x)

This 'customised' variable can be used directly in other functions (eg. regression equations) without the need to manually create dummy variables.

Upvotes: 0

Roman Luštrik
Roman Luštrik

Reputation: 70653

This appears to work.

x <- factor(c("0", "0", "6", "6", "3", "4"), levels = 0:6, ordered = TRUE)

out <- matrix(0, nrow = length(x), ncol = max(as.numeric(x)))

for (i in 1:length(x)) {
  out[i, 1:as.numeric(x[i])] <- 1
}
colnames(out) <- paste("ti", levels(x), sep = "")


     ti0 ti1 ti2 ti3 ti4 ti5 ti6
[1,]   1   0   0   0   0   0   0
[2,]   1   0   0   0   0   0   0
[3,]   1   1   1   1   1   1   1
[4,]   1   1   1   1   1   1   1
[5,]   1   1   1   1   0   0   0
[6,]   1   1   1   1   1   0   0

Upvotes: 1

Related Questions