Using label as value while having negative values-CodePudding

im trying to use the labels from a dataset as the values instead. The dataset was imported from a .sav file using haven. I tried (and suceeded with different data) with this:

df %>% 
  mutate(state = get_labels(state)[state])

Now i have two problems:

This does not work for the example below, apparently (if i got that right) because some of the values are negative? Is there a way around that? I get the error

Error in `mutate()`:
! Problem while computing `vote = get_labels(vote)[vote]`.
Caused by error in `get_labels(vote)[vote]`:
! only 0's may be mixed with negative subscripts

Some of the columns in the dataframe are not labelled. Is there a way to only execute this if they happen to have one and else leave them be? I tried the above code on a unlabelled column and it was deleted...

Many thanks in advance, and i hope my example below is reproducible!

Example data:
structure(c(215, 1, -97, 4, -97, 5, -97, -98, 6, -97, -97, 5, 
-99, -97, 7, -98, 1, -97, 1, 1), labels = c(`keine Angabe` = -99, 
`weiss nicht` = -98, `trifft nicht zu` = -97, `nicht in Auswahlgesamtheit` = -94, 
`ungueltig waehlen` = -83, `CDU/CSU` = 1, SPD = 4, FDP = 5, GRUENE = 6, 
`DIE LINKE` = 7, BP = 126, `DIE GRAUEN` = 149, `Die PARTEI` = 151, 
`DIE VIOLETTEN` = 152, DVU = 168, FAMILIE = 171, `FREIE WAEHLER` = 180, 
NPD = 206, oedp = 209, PBC = 214, PIRATEN = 215, RENTNER = 224, 
REP = 225, RRP = 226, Tierschutzpartei = 237, Volksabstimmung = 249, 
AfD = 322, BGE = 338, DiB = 349, `andere Partei` = 801, `keine Partei; keiner Partei` = 808, 
`Interview abgebrochen` = -93, `Fehler in Daten` = -92, Mehrfachnennungen = -73, 
DKP = 156, MLPD = 202, SSW = 234, NM = 326, LKR = 331, `V-Partei³` = 344, 
Tierschutzallianz = 345, `Die Humanisten` = 350, `Buendnis C` = 351, 
`Die Grauen` = 361, Volt = 364, dieBasis = 372, `Team Todenhoefer` = 373, 
BUENDNIS21 = 374, BUERGERBEWEGUNG = 378, Sonstige = 800), label = "Wahlabsicht: BTW, Zweitstimme (Version B)", class = c("haven_labelled", 
"vctrs_vctr", "double"))

CodePudding user response：

If all you are trying to do is to use labels instead of the values, use unlabelled() from labelled package. You can pass your entire dataframe at once.

I transformed the dataframe in tibble because it is more organized this way. But this step is not needed.

Packages need

library(dplyr)
library(labelled)

Solution

df %>% as_tibble() %>% unlabelled()

Output

# # A tibble: 20 × 1
# value          
# <fct>          
# 1 PIRATEN        
# 2 CDU/CSU        
# 3 trifft nicht zu
# 4 SPD            
# 5 trifft nicht zu
# 6 FDP            
# 7 trifft nicht zu
# 8 weiss nicht    
# 9 GRUENE         
# 10 trifft nicht zu
# 11 trifft nicht zu
# 12 FDP            
# 13 keine Angabe   
# 14 trifft nicht zu
# 15 DIE LINKE      
# 16 weiss nicht    
# 17 CDU/CSU        
# 18 trifft nicht zu
# 19 CDU/CSU        
# 20 CDU/CSU