Transform (shuffle) just 2 Fields in a Dataframe

Question

I have a data frame with firstname, lastname and I want to permutate them but ONLY for the rows that have values. There are many null fields and I don't want reorder them so that there is ever a firstname value without a lastname value. Ex:

number<- c(1,2,3,4,5)
firstname<- c('','Eddie','Edward','','Edurardo')
lastname <- c('','Vedder', 'Van Halen', '', 'Norton')
permtest <- data.frame(number,firstname,lastname)
permtest
  number firstname  lastname
1      1                    
2      2     Eddie    Vedder
3      3    Edward Van Halen
4      4                    
5      5  Eduardo    Norton

Expected results would be:

Eddie Norton
Edward Vedder
Eduardo Van Halen

But not:

Eddie _____

or:

______ Van Halen

I tried the transform function but it didn't work:

permtest2 <- transform( permtest2, firstname = sample(firstname,lastname) )

score 1 · Accepted Answer · answered Dec 10 '18 at 19:57

1

What helps is shuffling only the nonempty entries:

permtest$lastname[permtest$lastname != ''] <- sample(permtest$lastname[permtest$lastname != ''])
permtest
#   number firstname  lastname
# 1      1                    
# 2      2     Eddie Van Halen
# 3      3    Edward    Vedder
# 4      4                    
# 5      5  Edurardo    Norton

answered Dec 10 '18 at 19:57

Julius Vainora

47,421
9
90
102

Awesome! Worked as expected. Thanks! – Jerid C. Fortney Dec 12 '18 at 14:52
I had to add this first, though: `dat[is.na(dat)] = ' '` – Jerid C. Fortney Dec 12 '18 at 16:18

score 0 · Answer 2 · answered Dec 10 '18 at 19:57

0

One of many ways to do this:

permtest[permtest == ''] = NA

library(dplyr)

permtest %>% 
  dplyr::filter(!is.na(lastname), !is.na(firstname)) %>% 
  dplyr::mutate(val = paste(firstname, lastname)) %>% 
  dplyr::pull(val)

[1] "Eddie Vedder"     "Edward Van Halen" "Edurardo Norton"

answered Dec 10 '18 at 19:57

astrofunkswag

2,608
12
25

I'm sure this is a valid answer too, but I received this (probably unrelated) error: Error: 'pull' is not an exported object from 'namespace:dplyr' – Jerid C. Fortney Dec 12 '18 at 14:51
Did you mean `dat[is.na(dat)] = ''`? That's what I used along with Julius' code (above)' after getting an error and, together, it seems to work. – Jerid C. Fortney Dec 12 '18 at 16:14
`pull` is definitely a [dplyr function](https://www.rdocumentation.org/packages/dplyr/versions/0.7.8/topics/pull). I'm not sure why that error would be occuring. – astrofunkswag Dec 12 '18 at 18:38
That first line of code converts `' '` to `NA` so I can filter it out with the `!is.na` command. The comment you added does the reverse, and converts `NA` to `' '`, but you don't initially have any `NA` in your dataframe so that won't do anything – astrofunkswag Dec 12 '18 at 18:44

score 0 · Answer 3 · answered Dec 10 '18 at 20:06

0

Using tidyverse you could do

library(tidyverse)
library(stringr)

permtest2 <- permtest %>% mutate(Nfname = str_length(firstname)) %>% filter(Nfname > 0) %>% mutate(lastname = sample(lastname, size = length(lastname))) %>% select(-Nfname)

answered Dec 10 '18 at 20:06

Derek Corcoran

3,930
2
25
54

I'm new at R, but I feel like this changes the dimensions of permtest2. I still need it to be 5 rows in length (keep the rows with blanks). – Jerid C. Fortney Dec 12 '18 at 13:54

Transform (shuffle) just 2 Fields in a Dataframe

3 Answers3