How to combine rowSums and ifelse with mutate - r

Question

I need to combine rowSums and ifelse in order to create a new variable. My data looks like this:
boss var1 var2 var3 newvar
1 NA NA 3 NA
1 2 3 3 8
2 NA NA NA 0
2 NA NA NA 0
2 NA NA NA 0
1 1 NA 2 3
if boss==1, and there's more than one missing value in var1 to var3, newvar should be NA, otherwise, it should be the result of var1+var2+var3
If boss==2, newvar should be automatically 0.
So far, I have been able to solve parts of the problem using dplyr:
mutate(newvar=rowSums(.[,2:4],na.rm=TRUE) +
ifelse(rowSums(is.na(.[,2:4]))>1 & boss==2,NA,0))
mutate(newvar=ifelse(boss==2,0,NA)
However, I'm struggling to combine the two. Any help is much appreciated.

Answer 1

Here is one option with case_when where we create an index ('i1') which computes the number of NA elements in the row. The index is used in the case_when to create logical conditions to assign the values
df %>%
mutate(i1 = rowSums(is.na(.[-1]))) %>%
mutate(newvar = case_when(i1 > 1 & boss==1 ~ NA_integer_,
boss==2 ~ 0L,
i1 <=1 & boss != 2~ as.integer(rowSums(.[2:4], na.rm = TRUE)))) %>%
select(-i1)
# boss var1 var2 var3 newvar
#1 1 NA NA 3 NA
#2 1 2 3 3 8
#3 2 NA NA NA 0
#4 2 NA NA NA 0
#5 2 NA NA NA 0
#6 1 1 NA 2 3
In base R, this can be done with creating index and without using any ifelse
i1 <- df$boss != 2
tmp <- i1 * df[-1]
df$newvar <- NA^(rowSums(is.na(tmp)) > 1 & i1) * rowSums(tmp, na.rm = TRUE)
df$newvar
#[1] NA 8 0 0 0 3
data
df <- structure(list(boss = c(1L, 1L, 2L, 2L, 2L, 1L), var1 = c(NA,
2L, NA, NA, NA, 1L), var2 = c(NA, 3L, NA, NA, NA, NA), var3 = c(3L,
3L, NA, NA, NA, 2L)), .Names = c("boss", "var1", "var2", "var3"
), row.names = c(NA, -6L), class = "data.frame")

Answer 2

A solution in base-R using apply can be as:
df$newvar <- apply(df,1, function(x){
#retVal = NA
if(x["boss"]==2){
0
} else if(sum(is.na(x[-1])) > 1){
NA
} else{
sum(x[-1], na.rm = TRUE)
}
})
# boss var1 var2 var3 newvar
# 1 1 NA NA 3 NA
# 2 1 2 3 3 8
# 3 2 NA NA NA 0
# 4 2 NA NA NA 0
# 5 2 NA NA NA 0
# 6 1 1 NA 2 3
Data:
df <- read.table(text =
"boss var1 var2 var3
1 NA NA 3
1 2 3 3
2 NA NA NA
2 NA NA NA
2 NA NA NA
1 1 NA 2",
header = TRUE, stringsAsFactors = FALSE)