Magnus
Magnus

Reputation: 760

how to avoid "No common size" error for separate_rows()-function

I'm working with data that looks something like this:

    AF:         AU:
1   MIT Duflo, Esther
2   NBER; NBER  Freeman, Richard B.; Gelber, Alexander M.
3   U MI; Cornell U; U VA   Bound, John; Lovenheim, Michael F.; Turner, Sarah
4   Harvard U; U Chicago    Fryer, Roland G., Jr.; Levitt, Steven D.
5   U OR; U CA, Davis; U British Columbia   Lindo, Jason M.; Sanders, Nicholas J.; Oreopoulos, Philip

I have two variables, AF: for affiliation and AU: for authors. Different authors and affiliations are separated with semicolon, I want to use the separate_rows-command and create somthing like this:

   AF:               AU:
    MIT               Duflo, Esther
    NBER              Freeman, Richard B.
    NBER              Gelber, Alexander M.
    U MI              Bound, John 
    Cornell U         Lovenheim, Michael F. 
    U VA              Turner, Sarah
    Harvard U;        Fryer, Roland G., Jr.
    U Chicago         Levitt, Steven D.
    U OR              Lindo, Jason M.
    U CA,             Davis Sanders, Nicholas J. 
    U British ColumbiaOreopoulos, Philip 

The standard version of separate_rows() generates an error message, probably since my data contains NAs:

authaf_spread<-separate_rows(authaf, 1:2, sep=";")
Error: All nested columns must have the same number of elements.

I downloaded and installed the develpment version, which just gives me another error message:

authaf_spread<-separate_rows(authaf, 1:2, sep=";")
Error: No common size for `AF:`, size 3, and `AU:`, size 4.
Call `rlang::last_error()` to see a backtrace

What does this mean and how do I circumvent this error?

If anyone's interested I'm attaching a link to the entire file:

https://www.dropbox.com/s/z456w7ll7v7o79z/authors_affiliations.csv?dl=0

Upvotes: 0

Views: 686

Answers (1)

rpolicastro
rpolicastro

Reputation: 1305

If you call separate_rows twice, it will work. I used str_trim from stringr to remove whitespace that appeared before and after the author names and affiliations, and drop_na from tidyr to remove rows that had NA for both columns.

# Loaded your .csv file as variable 'df'

authors <- df %>%
  separate_rows(AF., sep = ";") %>%
  separate_rows(AU., sep = ";") %>%
  mutate_all(~ str_trim(., side = "both")) %>%
  drop_na

# A tibble: 24,877 x 2
   AF.       AU.                  
   <chr>     <chr>                
 1 MIT       Duflo, Esther        
 2 NBER      Freeman, Richard B.  
 3 NBER      Gelber, Alexander M. 
 4 NBER      Freeman, Richard B.  
 5 NBER      Gelber, Alexander M. 
 6 U MI      Bound, John          
 7 U MI      Lovenheim, Michael F.
 8 U MI      Turner, Sarah        
 9 Cornell U Bound, John          
10 Cornell U Lovenheim, Michael F.
# … with 24,867 more rows

You can also remove rows that are duplicated with author and affiliation by using distinct.

authors %>% distinct(AF., AU.)

# A tibble: 5,873 x 2
   AF.       AU.                  
   <chr>     <chr>                
 1 MIT       Duflo, Esther        
 2 NBER      Freeman, Richard B.  
 3 NBER      Gelber, Alexander M. 
 4 U MI      Bound, John          
 5 U MI      Lovenheim, Michael F.
 6 U MI      Turner, Sarah        
 7 Cornell U Bound, John          
 8 Cornell U Lovenheim, Michael F.
 9 Cornell U Turner, Sarah        
10 U VA      Bound, John          
# … with 5,863 more rows

Upvotes: 2

Related Questions