Conditional recoding of factor

Question

I need to recode a couple of factor variables, but I just keep on failing.

Suppose my data looks like this:

df <- data.frame(a  = c("1","2","","Other"),
                 b  = c("3","","Other","Other"),
                 v1 = c("no","no","yes","yes"),
                 v2 = c("no","yes","no","no"),
                 v3 = c("no","yes","yes","no"))
df$a <- as.character(df$a)
df$b <- as.character(df$b)

df

>       a     b  v1  v2  v3
> 1     1     3  no  no  no
> 2     2        no yes yes
> 3       Other yes  no yes
> 4 Other Other yes  no  no

I want

v1 to be "yes" if (a=="1" | b=="1"),

v2 to be "yes" if (a=="2" | b=="2") and

v3 to be "yes" if (a=="3" | b=="3").

So the pattern is:

v# to be "yes" if (a="#" | b="#").

I tried with R base using 2 loops, but it did not work:

 for(i in c("a","b")){
   for(j in as.character(1:3)){
   df[which(df[,i]==j),][,c(paste("v",j,sep=""))] <- "yes"
   }}

I would prefer to do this using dplyr::mutate, but don't know how...

You can use a single loop `df[paste0("v", 1:3)] <- lapply(as.character(1:3), function(i) ifelse(df$a==i & df$b==i, "yes", NA))` — akrun, May 11 '16 at 13:45
Is the pattern always the same? i.e. for v1, a=1|b=1 , for v2 a=2|b=2 etc...? — Sotos, May 11 '16 at 14:10
@Sotos Yes it is! It is always v# for a="#" | b="#". Note that a and b are character vectors. — chamaoskurumi, May 11 '16 at 14:18

score 1 · Accepted Answer · answered May 11 '16 at 16:16

library(data.table)
dt = as.data.table(df) # or convert in-place using setDT

for (i in 1:3) dt[a == i | b == i, paste0('v', i) := 'yes']
#       a     b  v1  v2  v3
#1:     1     3 yes  no yes
#2:     2        no yes yes
#3:       Other yes  no yes
#4: Other Other yes  no  no

score -1 · Answer 2 · answered May 11 '16 at 13:52

-1

In this case, since the number of variables is relatively small, a simple group of ifelse would solve the issue:

df$v1<- ifelse(df$a==1 |df$b==1,"yes","no")
df$v2<- ifelse(df$a==2 |df$b==2,"yes","no")
df$v3<- ifelse(df$a==3 |df$b==3,"yes","no")

answered May 11 '16 at 13:52

JMenezes

1,004
1
6
13

well in fact my real dataset has a much higher number of variables. I just made the example more compact. Hardcoding for each variable is not an option for me. – chamaoskurumi May 11 '16 at 14:01
Besides, your code overwrites df[3,3] from "yes" to "no", which is not what I want to do. I just want to update the the `v` variables with the "yes" on the explained condition. – chamaoskurumi May 11 '16 at 15:09

Conditional recoding of factor

2 Answers2