level order for melt.array

[knokknok]

Since arrays row/column/... order can be meaningful, it would be nice to have an option in melt.array to be able to have the levels of factors generated from dimnames to be in the original array order.

[hadley]

That's the default with as.is = FALSE

[knokknok]

As I understand it, as.is = FALSE converts dimnames to factors. However the levels of the factors are the ordered strings.
What I was looking for is levels in the initial order of the row/col-names: factor(colnames(x), levels=unique(colnames(x)))

[hadley]

Ah, good point - I only checked with arrays that already had sorted names. It would be helpful if you could contribute a couple of examples of expected behaviour.

[knokknok]

A bit convoluted but:

data <- matrix(rnorm(40), ncol=4, dimnames=list(paste0("S", 1:10), paste0("1", c("", "0", "00", "000"), "g/ml")))
df <- melt(data, varnames=c("Sample", "Concentration"))
# default:
ggplot(df, aes(x=Concentration, y=value))+geom_point()
# ordered as in the matrix:
df$Concentration <- factor(df$Concentration, levels=colnames(data))
ggplot(df, aes(x=Concentration, y=value))+geom_point()

[hadley]

Here's a slightly simpler case:

library(reshape2)

data <- matrix(1:12, c(3, 4))
dimnames(data) <- list("x" = letters[3:1], y = LETTERS[4:1])

str(melt(data))