How can I go from wide to long while coupling numbered columns with one another?

问题

I have a dataset that looks like this:

phrase      wo1sp     wo2sp     wo3sp     wo1sc     wo2sc     wo3sc
hello       dan       mark      todd      10        5         4
hello       mark      dan       chris     8         9         4
goodbye     mark      dan       kev       2         4         10
what        kev       dan       mark      4         5         5

And I'd like to change it to something like this:

phrase      sp      sc
hello       dan     10 
hello       mark    5
hello       todd    4
hello       mark    8
hello       dan     9
hello       chris   4
goodbye     mark    2
goodbye     dan     4
goodbye     kev     10
what        kev     4
what        dan     5
what        mark    5

So, I know the first thing to do here is group_by(phrase). What I'm not sure about is how to associate sp1 with sc1, sp2 with sc2, etc. and make those each into their own rows. I've seen some vaguely similar things using reshape and tidy, but they don't depend on there being coupled columns. I would basically just like to collapse the numbers in the column names.

I have a request: When you answer, would you mind explaining what the code itself does? Many of the things I've searched on StackExchange present a seemingly esoteric solution with no explanation of what's going on.

回答1:

library("tidyverse")

test_set = tribble(~phrase,      ~wo1sp,     ~wo2sp,     ~wo3sp,     ~wo1sc,     ~wo2sc,     ~wo3sc,
                   "hello",       "dan",       "mark",      "todd",      10,        5,         4,
                   "goodbye",     "mark",      "dan",       "kev",       2,         4,         10,
                   "what",        "kev",       "dan",       "mark",      4,         5,         5)

test_set %>% 
  gather(key = col, value = val, -phrase) %>% 
  separate(col = col, into = c("num", "suffix"), sep = 3) %>% 
  spread(key = suffix, value = val) %>% 
  mutate(sc = as.numeric(sc)) %>% 
  select(-num)

Edit: I guess it's not necessary to split col into three columns, could just do sep = 3

来源：https://stackoverflow.com/questions/53840788/how-can-i-go-from-wide-to-long-while-coupling-numbered-columns-with-one-another

标签

dplyr

reshape