Create New Data Frame with Column Names from Unique Values in another Data Frame and Corresponding Values Assigned to Column

问题

I'm new to R, and I'm pretty sure this is something simple to accomplish, but I cannot figure out how to perform this action. I've tried the split function, utilizing a for loop, but cannot quite figure out how to get it right. As an example, this is what my original data frame looks like:

dat <- data.frame(col1 = c(rep("red", 4), rep("blue", 3)), col2 = c(1, 3, 2, 4, 7, 8, 9))

 col1 col2
  red    1
  red    3
  red    2
  red    4
 blue    7
 blue    8
 blue    9

I want to create new columns for each unique value in col1 and assign it's corressponding value in col2 to the new data frame. And this is how I want my new data frame:

I've gotten close with a list structure close to what I wanted, but I need a data frame to boxplot and dotplot the results. Any help would be appriciated. Thanks!

回答1:

I'm sure there's a more efficient solution, but here's one option

dat <- data.frame(col1 = c(rep("red", 4), rep("blue", 3)), col2 = c(1, 3, 2, 4, 7, 8, 9))
dat

  col1 col2
1  red    1
2  red    3
3  red    2
4  red    4
5 blue    7
6 blue    8
7 blue    9    

ust <- unstack(dat, form = col2 ~ col1)
res <- data.frame(sapply(ust, '[', 1:max(unlist(lapply(ust, length)))))
res
  blue red
1    7   1
2    8   3
3    9   2
4   NA   4

Edit: If you want the column order red then blue

res[, c("red", "blue")]
  red blue
1   1    7
2   3    8
3   2    9
4   4   NA

回答2:

Here's an Hadleyverse possible solution

library(tidyr)
library(dplyr)
dat %>%
  group_by(col1) %>%
  mutate(n = row_number()) %>%
  spread(col1, col2)
# Source: local data frame [4 x 3]
# 
#   n blue red
# 1 1    7   1
# 2 2    8   3
# 3 3    9   2
# 4 4   NA   4

Or using data.table

library(data.table)
dcast(setDT(dat)[, indx := 1:.N, by = col1], indx ~ col1, value.var = "col2")
#    indx blue red
# 1:    1    7   1
# 2:    2    8   3
# 3:    3    9   2
# 4:    4   NA   4

回答3:

Just to show another option using base R *applyand cbind

# split the data into list using col1 column
tmp.list   = lapply(split(dat, dat$col1), function(x) x$col2)

# identify the length of the biggest list
max.length = max(sapply(tmp.list, length))

# combine the list elements, while filling NA for the missing values
data.frame(do.call(cbind, 
  lapply(tmp.list, function(x) c(x, rep(NA, max.length - length(x))))
))

#  blue red
#1    7   1
#2    8   3
#3    9   2
#4   NA   4

来源：https://stackoverflow.com/questions/32236327/create-new-data-frame-with-column-names-from-unique-values-in-another-data-frame

标签

reshape