I have a long data set. How do I convert to wide format while using a rank function to limit the number of columns created?

I have a data set in a long format. Some records only have one manufacturer, some records are duplicated because there are multiple manufacturers associated with each record.

For example, I have the following data:

id      manufacturer
111     AAA
222     AAA
222     BBB
222     CCC
222     DDD
333     CCC
333     DDD
444     EEE

Using dplyr, I added a rank so I am able to count the number of manufacturers.

df %>% 
  select(id, manufacturer) %>% 
  group_by(id) %>% 
  mutate(rank = rank(manufacturer, ties.method = "first"))

To get:

id      manufacturer    rank
111     AAA             1
222     AAA             1
222     BBB             2
222     CCC             3
222     DDD             4
333     CCC             1
333     DDD             2
444     EEE             1

I've tried dcast which gets me close, but I'm not quite there.

dcast(df, id ~ rank)

When I use dcast I get:

id     1    2    3    4
111    1    NA   NA   NA
222    1    2    3    4
333    1    2    NA   NA
444    1    NA   NA   NA

What I am trying to get is:

id     manuf_1    manuf_2    manuf_3    manuf_4
111    AAA                                
222    AAA        BBB        CCC        DDD
333    CCC        DDD        
444    EEE

Solution

You just need to set the value.var, and mutate rank to your desired format before dcast

df %>% 
  dplyr::mutate(rank = paste0('manuf_', rank)) %>% 
  reshape2::dcast(id ~ rank, value.var = 'manufacturer')
#    id manuf_1 manuf_2 manuf_3 manuf_4
# 1 111     AAA    <NA>    <NA>    <NA>
# 2 222     AAA     BBB     CCC     DDD
# 3 333     CCC     DDD    <NA>    <NA>
# 4 444     EEE    <NA>    <NA>    <NA>

data used

df <- data.table::fread('
id      manufacturer    rank
111     AAA             1
222     AAA             1
222     BBB             2
222     CCC             3
222     DDD             4
333     CCC             1
333     DDD             2
444     EEE             1
')