Spreading over a column in R

Question

So say I have a data frame like this:

data.frame(x = c(1,1,1,3,3,3),y = c(12,32,43,16,32,65))

and I want to transform it into a data frame like this:

data.frame(x = c(1, 3), y_1 =  c(12,16), y_2 =c(32, 32),y_3= c(43, 65))

basically spreading the y values for each unique x value. I've tried to do this using tidyr but can't quite see how it would work. Any ideas?

Thanks.

Answer 1

Here's a data.table solution:

library(data.table)

dat = as.data.table(df) # or setDT to convert in place

dat[, obs := paste0('y_', 1:.N), by=x]
dcast(dat, x ~ obs, value.var="y")

#   x y_1 y_2 y_3
#1: 1  12  32  43
#2: 3  16  32  65

This will work even if the number of rows is not the same for all x .

Answer 2

We can use aggregate , and then cSplit from splitstackshape package to coerce to data frame,

library(splitstackshape)
df1 <- aggregate(y ~ x, df, paste, collapse = ',')
df1 <- cSplit(df1, 'y', ',', direction = 'wide')
#   x y_1 y_2 y_3
#1: 1  12  32  43
#2: 3  16  32  65

Answer 3

The answer given by Sotos using aggregate is particularly elegant, but the following approach using reshape might also be instructive:

df <- data.frame(x = c(1,1,1,3,3,3),y = c(12,32,43,16,32,65))
df[,"time"] <- rep(1:3, 2)
wide_df <- reshape(df, direction="wide", timevar="time", idvar="x")

Answer 4

One option with dplyr/tidyr

library(dplyr)
library(tidyr)
df1 %>% 
    group_by(x) %>% 
    mutate(n = paste("y", row_number(), sep="_")) %>%
    spread(n,y)
#     x   y_1   y_2   y_3
#   (dbl) (dbl) (dbl) (dbl)
#1     1    12    32    43
#2     3    16    32    65