How to transform a key/value string into separate columns?

问题

I've got a data.frame with key/value string column containing information about features and their values for a set of users. Something like this:

data<-data.frame(id=1:3,statid=c("s003e","s093u","s085t"),str=c("a:1,7:2","a:1,c:4","a:3,b:5,c:33"))
data
#   id statid          str
# 1  1  s003e      a:1,7:2
# 2  2  s093u      a:1,c:4
# 3  3  s085t a:3,b:5,c:33

What I'm trying to do is to create a data.frame containing column for every feature. Like this:

data_after<-data.frame(id=1:3,statid=c("s003e","s093u","s085t"),
                   a=c(1,1,3),b=c(0,0,5),c=c(0,4,33),"7"=c(2,0,0))
data_after
#   id statid a b  c X7
# 1  1  s003e 1 0  0  2
# 2  2  s093u 1 0  4  0
# 3  3  s085t 3 5 33  0

I was trying to use str_split from stringr package and then transform elements of created list to data.frames (later bind them using for example rbind.fill from plyr) but couldn't done it. Any help will be appreciated!

回答1:

You can use dplyr and tidyr:

library(dplyr); library(tidyr)
data %>% mutate(str = strsplit(str, ",")) %>% unnest(str) %>% 
         separate(str, into = c('var', 'val'), sep = ":") %>% spread(var, val, fill = 0)

#   id statid 7 a b  c
# 1  1  s003e 2 1 0  0
# 2  2  s093u 0 1 0  4
# 3  3  s085t 0 3 5 33

回答2:

We can use cSplit to do this in a cleaner way. Convert the data to 'long' format by splitting at ,, then do the split at : and dcast from 'long' to 'wide'

library(splitstackshape)
library(data.table)
dcast(cSplit(cSplit(data, "str", ",", "long"), "str", ":"), 
                    id+statid~str_1, value.var="str_2", fill = 0)
#   id statid 7 a b  c
#1:  1  s003e 2 1 0  0
#2:  2  s093u 0 1 0  4
#3:  3  s085t 0 3 5 33

来源：https://stackoverflow.com/questions/38144082/how-to-transform-a-key-value-string-into-separate-columns

标签

split

multiple-columns

key-value