CODEWITHSUNDEEP
rdata.table
Wet Feet
Wet Feet

Reputation: 4555

Change column position of data.table

I am learning to use the data.table package. One of the things I am trying to do is move the last column ("x") to the first column. Here is how I do it for a data frame:

DF <- cbind(x, DF[1:ncol(DF)]) #Rearrange columns

I read up on setcolorder and tried this, but I get an error

setcolorder(DT, c("x", 1: (length(DT)-1) ) )

Does anyone know a better solution?

Upvotes: 13

Views: 10706

Answers (2)

user2925400
user2925400

Reputation: 11

I think the following syntax will work if you want x in the first column

DF <- cbind（x,DF)

Or if you want it in any column like n

DF.combine <- cbind(DF[,1:(n-1)], x, DF[,n:ncol(DF)])# n should >1

Upvotes: -1

A5C1D2H2I1M1N2O1R2T1
A5C1D2H2I1M1N2O1R2T1

Reputation: 193517

Option 1

Maybe you can use setdiff:

DT <- data.table(A = 1:2, B = 3:4, X = 5:6)
DT
#    A B X
# 1: 1 3 5
# 2: 2 4 6
setcolorder(DT, c("X", setdiff(names(DT), "X")))
DT
#    X A B
# 1: 5 1 3
# 2: 6 2 4

Option 2

Using a modified version of your approach:

setcolorder(DT, c("X", names(DT)[1:(length(DT)-1)]))

or 

setcolorder(DT, c(length(DT), 1:(length(DT)-1)))

Why the error in your approach? You were trying to include both column names and numeric column indices. Use one or the other, but not both.

Option 3

I've written a function called moveme (which for the time being you can find at this Gist or at my blog). You enter a string of "move" commands, separated by semicolons. This lets you shuffle around your columns pretty flexibly:

DT <- data.table(matrix(1:20, ncol = 10, dimnames = list(NULL, LETTERS[1:10])))
DT
#    A B C D  E  F  G  H  I  J
# 1: 1 3 5 7  9 11 13 15 17 19
# 2: 2 4 6 8 10 12 14 16 18 20

setcolorder(DT, moveme(names(DT), "E, F first; B last; H, I before G"))
# DT
#     E  F A C D  H  I  G  J B
# 1:  9 11 1 5 7 15 17 13 19 3
# 2: 10 12 2 6 8 16 18 14 20 4

Upvotes: 32

Related Questions