Create multiple columns in R based on other column

Question

I have 2 columns in data frame , please refer to below

no  value
1   A_0.9
1   B_0.8
1   C_0.7
1   D_0.7
2   B_0.9
2   D_0.8
2   A_0.7
2   C_0.7

I want to create new data frame as below

no  value1  value2  value3  value4
1   A_0.9   B_0.8   C_0.7   D_0.7
2   B_0.9   D_0.8   A_0.7   C_0.7

ie: for each unique value in column "no" there will be multiple columns created using data in column "value"

Answer 1

t(unstack(df, value ~ no))
#   [,1]    [,2]    [,3]    [,4]   
#X1 "A_0.9" "B_0.8" "C_0.7" "D_0.7"
#X2 "B_0.9" "D_0.8" "A_0.7" "C_0.7"

To tidy the above output to fit your data,

 library(dplyr)
 df1 <- as.data.frame(t(unstack(df, value ~ no)))
 names(df1)[-1] <- paste0('value', 2:ncol(df1)-1)
 rownames(df1) <- NULL
 df1 <- add_rownames(df1, 'no')  #from dplyr package
 #    no value1 value2 value3 value4
 #  (chr) (fctr) (fctr) (fctr) (fctr)
 #1     1  A_0.9  B_0.8  C_0.7  D_0.7
 #2     2  B_0.9  D_0.8  A_0.7  C_0.7

Answer 2

Using data.table , we can create a sequence per unique value by no with rleid() , and consequently use it to dcast() the data to wide format.

library(data.table)
dcast(setDT(df)[, nr := rleid(value),by = no], no ~ nr)
#  no     1     2     3     4
#1  1 A_0.9 B_0.8 C_0.7 D_0.7
#2  2 B_0.9 D_0.8 A_0.7 C_0.7

Or with the dev version (1.9.7) of data.table , the following is possible, thanks @Arun!

dcast(setDT(df), no ~ rowid(no, prefix = 'value'))
#   no value1 value2 value3 value4
#1:  1  A_0.9  B_0.8  C_0.7  D_0.7
#2:  2  B_0.9  D_0.8  A_0.7  C_0.7

Answer 3

I would use reshape library, which wraps a nice set of data manipulation functions. Example to accomplish your task:

n = c(1,1,1,1,2,2,2,2)
x = c('A', 'B', 'C', 'D', 'A', 'B', 'C', 'D')
# Just to create the column names you showed in the example
columns = rep(paste("value", 1:4, sep=""), 2)
data = data.frame(n, columns, x)
cast(data, n~columns)