CODEWITHSUNDEEP
rsorting
JmO
JmO

Reputation: 572

reorder specified columns in a data.frame ascending

I have one short question, how can I reorder a dataframe only for selected columnnames. I would need a universal solution here because I have to use it on changing amounts of V columns (everytime with V columns >100)

Example:

Consider I have the data:

dkk <- structure(list(A = 2L, X = 3L, C = 4L, D = 5L, Z = 6L, V1 = 5L, 
                     V6 = 5L, V4 = 5L, V5 = 5L, V3 = 2L, V2 = 2L), .Names = c("X", 
                                                                              "B", "C", "D", "Z", "V1", "V6", "V4", "V5", "V3", "V2"), 
                class = "data.frame", row.names = c(NA, -1L)) 

# X B C D Z V1 V6 V4 V5 V3 V2
  2 3 4 5 6  5  5  5  5  2  2

How can I reorder the columns with a V so that they are in a ascending order: 

# X B C D Z V1 V2 V3 V4 V5 V6
  2 3 4 5 6  5  2  2  5  5  5

Many thanks!!

Upvotes: 3

Views: 182

Answers (2)

akrun
akrun

Reputation: 887088

Here is a faster option with setcolorder from data.table

library(data.table)
i1 <- grep("V\\d+", names(dkk), value = TRUE)
cbind(dkk[setdiff(names(dkk), i1)], setcolorder(dkk[i1], order(i1))[])
#  A B C D Z V1 V2 V3 V4 V5 V6
#1 2 3 4 5 6  5  2  2  5  5  5

This becomes a bit complicated when the 'V' names are intermingled with other columns, for example suppose, we change the column names to

set.seed(24)
names(dkk) <- sample(names(dkk))
dkk
#   D C V6 Z V4 V1 B V2 V3 A V5
#1 2 3  4 5  6  5 5  5  5 2  2

Now, the option is to create a numeric index of those columns with 'V' ('i2'), extract the names ('i3') and assign the order of names and columns separately

i2 <- grep("^V\\d+", names(dkk))
i3 <- names(dkk)[i2]
names(dkk)[i2] <- sort(names(dkk)[i2])
dkk[i2] <- dkk[i2][order(i3)]

to get

dkk
#  D C V1 Z V2 V3 B V4 V5 A V6
#1 2 3  5 5  5  5 5  6  2 2  4

Update

There was one glitch in the above solution. It doesn't do the sorting correctly when we have column names with numbers greater than 9 i.e. 'V10', 'V11', etc. Suppose, our third column name is 'V100'

colnames(dkk)[3] <- "V100"   
dkk
#  D C V100 Z V4 V1 B V2 V3 A V5
#1 2 3    4 5  6  5 5  5  5 2  2

i2 <- grep("^V\\d+", names(dkk))
i3 <- names(dkk)[i2]

We can parse the number part with parse_number to assist in ordering

i4 <- readr::parse_number(i3)
names(dkk)[i2] <- i3[order(i4)]
dkk[i2] <- dkk[i2][order(i4)]
dkk
#  D C V1 Z V2 V3 B V4 V5 A V100
#1 2 3  5 5  5  5 5  6  2 2    4

data

dkk <- structure(list(A = 2L, B = 3L, C = 4L, D = 5L, E = 6L, V1 = 5L,
                 V6 = 5L, V4 = 5L, V5 = 5L, V3 = 2L, V2 = 2L), .Names = c("A",
                                                                          "B", "C", "D", "Z", "V1", "V6", "V4", "V5", "V3", "V2"),
            class = "data.frame", row.names = c(NA, -1L))

Upvotes: 4

Pierre Lapointe
Pierre Lapointe

Reputation: 16277

You can try this with order on colnames:

dkk[,order(colnames(dkk))]

  A B C D E V1 V2 V3 V4 V5 V6
  2 3 4 5 6  5  2  2  5  5  5

EDIT To order only columns that contain "V". Note: I included a Z column in the data set. Basically, I c the columns names that do not need to be sorted to the "V" columns that are sorted.

dkk <- structure(list(A = 2L, B = 3L, C = 4L, D = 5L, E = 6L, V1 = 5L,
                     V6 = 5L, V4 = 5L, V5 = 5L, V3 = 2L, V2 = 2L), .Names = c("A",
                                                                              "B", "C", "D", "Z", "V1", "V6", "V4", "V5", "V3", "V2"),
                class = "data.frame", row.names = c(NA, -1L))

cols <- c(colnames(dkk)[!grepl("V",names(dkk))],
colnames(dkk)[grepl("V",names(dkk))][order(colnames(dkk)[grepl("V",names(dkk))])])

dkk[,cols]

  A B C D Z V1 V2 V3 V4 V5 V6
1 2 3 4 5 6  5  2  2  5  5  5

Upvotes: 2

Related Questions