CODEWITHSUNDEEP
rdata.table
Jakob
Jakob

Reputation: 1453

order a data.table based on a character column with a specific (not alphabetical) order in mind

I want to order a data table based on a character vector specified manually.

library(data.table)
DT = data.table(x=c("c","b","a"), y=1:3)

I can order it alphabetically with:

DT[order(x)]

but can I order it based on a character vector like:

preferred.order <- c("b","a","c")

with the goal being:

data.table(x=c("b","a","c"), y=c(2,1,3))

In reality, I have a data.table with collected outputs and the variable names in the first column. For presentation purposes, I want those variables in a specific (not alphabetical) order.

Upvotes: 9

Views: 4134

Answers (1)

lmo
lmo

Reputation: 38500

One possibility is to join on the preferred order:

DT[preferred.order, on="x"]
   x y
1: b 2
2: a 3
3: c 1

Note that this requires the preferred.order vector contains all elements in DT$x and has no duplicates.

As an alternative, you could create a factor variable of DT$x with the preferred ordering and then use setorder to order DT by reference.

DT[, xFac := factor(x, levels=preferred.order)]
setorder(DT, xFac)

which returns

DT
   x y xFac
1: b 2    b
2: a 3    a
3: c 1    c

Which method is preferable will vary on the use-case.

Upvotes: 12

Related Questions