Using Purrr::map2 to Loop Over Two Vectors of Column Names in Order to Conditionally Recode Multiple Columns into New Variables

Question

library(tidyverse)

Using the sample code below, I want to use "mutate" or "mutate_at" to recode multiple columns into new columns based on the value of another column. Basically, I would like to recode the variables ending is "s" (q25s,q26s,etc...), based on the value of the corresponding non "s" variable. So for example, if q25 = 1, then q25s will be recoded so that 1 = 0, 2=0, 3=0, 4=1, 5=1, and 88=Missing, and the new name will be q25_new. If q25 does not equal 1, then is should not be recoded and q25_new should just be NA.

However, in order to achieve this I'm attempting to use tidyverse to create named vectors of column names, and then use "mutate", "recode", and "if_else" together with purr::map2.

I'm thinking something like the code below should be possible? I can't quite get it to work...and I feel like I need to use "paste0" somewhere to name all the new column variable names that start with "_new".

cols1<-Df %>%select(q25:q29)
cols2<-Df %>% select(q25s:q29s)

Df<- Df %>% map2(Df[cols1],Df[cols2],
  ~if_else(.x==1, mutate_at(vars (.y),funs(recode(.,`1`=0,`2`=0,`3`=0,`4`=1,`5`=1),"NA"))))

Here is the sample code.

Here is the sample code:
q25<-c(2,1,88,2,1)
q26<-c(2,88,88,88,2)
q27<-c(2,2,1,1,1)
q28<-c(88,1,1,2,2)
q29<-c(1,1,1,2,2)
q25s<-c(3,5,88,4,1)
q26s<-c(4,4,5,5,1)
q27s<-c(3,3,4,1,4)
q28s<-c(4,5,88,1,3)
q29s<-c(88,88,3,4,4)
Df<-data.frame(q25,q26,q27,q28,q29,q25s,q26s,q27s,q28s,q29s)

Can you show your desired output? Do want to keep q..s if q.. is not equal to 1? — Psidom
If q.. is not equal to 1, then q..s should not be recoded, so I suppose q_new would then be NA. — Mike

Moody_Mudskipper Moody_Mudskipper · Accepted Answer · 2017-10-26T19:14:09

Would this work ?

map2(Df[1:5],Df[6:10], ~ if_else(.x==1, recode(.y,`1`=0,`2`=0,`3`=0,`4`=1,`5`=1,`88` = NA_real_),NA_real_)) %>%
  as.data.frame %>%
  rename_all(paste0,"_new") %>%
  cbind(Df,.)

#   q25 q26 q27 q28 q29 q25s q26s q27s q28s q29s q25_new q26_new q27_new q28_new q29_new
# 1   2   2   2  88   1    3    4    3    4   88       3       4       3       4      NA
# 2   1  88   2   1   1    5    4    3    5   88       1       4       3       1      NA
# 3  88  88   1   1   1   88    5    4   88    3      88       5       1      NA       0
# 4   2  88   1   2   2    4    5    1    1    4       4       5       0       1       4
# 5   1   2   1   2   2    1    1    4    3    4       0       1       1       3       4

OK in the end I couldn't resist the challenge, so here's the pretty much 100% tidy way to go at it (same output) :

library(tidyr)
Df %>%
  mutate(n=row_number()) %>%
  gather(key,value,-n) %>%
  mutate(key2 = ifelse(grepl("s",key),"s","x"),
         key=sub("s","",key)) %>%
  spread(key2,value) %>%
  mutate(`_new` = if_else(x==1, recode(s,`1`=0,`2`=0,`3`=0,`4`=1,`5`=1,`88` = NA_real_),Inf)) %>%
  gather(key3,value,s,x,`_new`) %>%
  unite(key,key,key3,sep="") %>%
  spread(key,value) %>%
  rename_all(~gsub("x","",.x)) %>%
  select(order(nchar(names(.))),-n)

Using Purrr::map2 to Loop Over Two Vectors of Column Names in Order to Conditionally Recode Multiple Columns into New Variables

1 Answers