CODEWITHSUNDEEP
rdplyrtidyr
Manuel R
Manuel R

Reputation: 4145

Using tidyr::complete with group_by

Does anyone know if tidyr::complete() supports grouping via group_by()?

To be precise: I have some data frame that looks like this

df <- data.frame(
  "ID"   = rep(1:2, each = 2),
  "Col1" = c("A", NA, "AA", NA),
  "Col2" = c("B", "C", "BB", "CC"))

Now i'd like to use complete() and group_by() to compute all possible combinations per group! 

df %>% 
 group_by(ID) %>% 
 complete(Col1, Col2)

  Error in .Call("dplyr_left_join_impl", PACKAGE = "dplyr", x, y, by_x,  : 
  negative length vectors are not allowed

This causes an error. However, using complete() without grouping works but thats not what i want.

df %>% 
 complete(Col1, Col2)

Questions:

  1. Have I done anything wrong, or does complete() simply not work with group_by?
  2. If so, how could I do this instead (preferably without using a loop)?

Upvotes: 5

Views: 1947

Answers (3)

Manuel R
Manuel R

Reputation: 4145

Just wanted to let everyone know that with the devolopment version of tidyr (version 0.3.1.9000 as of 13.01.2016) all tidyr verbs now respect grouping so a workaround using dplyr::do is not necessary anymore. I will edit my answer once the version is avaiable on CRAN.

Upvotes: 2

shadow
shadow

Reputation: 22323

You could do it using complete and group_by, but you have to use a do statement: 

df %>% 
 group_by(ID) %>% 
 do(complete(., Col1, Col2, fill = list(ID = .$ID)))

Upvotes: 5

akrun
akrun

Reputation: 887291

We could do this using data.table. Convert the 'data.frame' to 'data.table' (setDT(df)), and Cross Join (CJ) the unique elements of 'Col1' and 'Col2', grouped by 'ID'.

library(data.table)#v1.9.6+
setDT(df)[,CJ(Col1, Col2, unique=TRUE), by = ID]
#   ID V1 V2
#1:  1 NA  B
#2:  1 NA  C
#3:  1  A  B
#4:  1  A  C
#5:  2 NA BB
#6:  2 NA CC
#7:  2 AA BB
#8:  2 AA CC

Upvotes: 2

Related Questions