CODEWITHSUNDEEP
rdataframe
user1830307
user1830307

Reputation:

Reorder a column in a dataframe to a specified order

I have a data frame, where every row has a value, and every block of 100 rows has an index (between 1 and 10). I would like to sort the index blocks in a particular order, but am unsure how to do that:

N=1000
value = runif(N, min=0, max=100)
index = rep(1:10, each=100)
DF=data.frame(value,index)
ord = c(1,4,6,3,7,9,8,2,5,10)

So basically, I would like the index column of DF to be ordered in blocks of the order specified in ord, instead of the index column of DF being ordered as 1,2,3,4,5,6,7,8,9,10.

Any advice would be appreciated!

Upvotes: 3

Views: 1968

Answers (3)

David Arenburg
David Arenburg

Reputation: 92282

You could simply convert index to factor and set the levels as in ord order and then sort the data as in

DF$index <- factor(DF$index, levels = ord)
DF[order(DF$index), ]

If you don't want to "alter your original data" you could simply create a separate index as in

indx <- factor(DF$index, levels = ord) 
DF[order(indx), ]

Additional otion is to order your data set by reference using setorder from the data.table package

library(data.table)
setorder(setDT(DF)[, index := factor(index, levels = ord)], index)

Upvotes: 4

First Name
First Name

Reputation: 23

Here is my proposal:

#Genreate the data    
N<-1000
value <- runif(N, min=0, max=100)
index <- rep(1:10, each=100)
DF<-data.frame(value,index)
ord <- c(1,4,6,3,7,9,8,2,5,10)

#Create a list with the data sorted by the provided order
newDF<-apply(matrix(ord,,ncol=1),1,function(x) DF[DF[,2]==x,])

#Unlist the list into a dataframe
do.call(rbind.data.frame,newDF)

Upvotes: 2

thelatemail
thelatemail

Reputation: 93813

Without altering your original data, you can use merge with sort=FALSE:

merge(data.frame(index=ord), DF, by="index", sort=FALSE)

#   ord    value
#1    1 37.29915
#2    1 30.09436
#3    4 18.05961
#4    4 46.73024
#5    6 93.15545
#6    6 69.33484
#7    3 70.92353
#8    3 81.63010
#9    7 22.23649
#10   7 32.36390
# etc etc

Upvotes: 1

Related Questions