CODEWITHSUNDEEP
rdataframeconditional-statementslapply
laksa
laksa

Reputation: 39

R conditional lapply?

I have a data frame such as

a = c(2,NA,3,4)
b = c(NA,3,NA,NA)
c= c(5,NA,7,9)
test = data.frame(a,b,c)

> test
   a  b  c
1  2 NA  5
2 NA  3 NA
3  3 NA  7
4  4 NA  9

I would like to fill in only NA values in test$b with the average of test$a and test$c for that row. The result should be

   a  b    c
1  2  3.5  5
2 NA  3    NA
3  3  5    7
4  4  6.5  9

I have tried the apply family but haven't gotten anywhere. Would like to avoid a for loop because I am told I should try to avoid for loops.

In English I want to say, 

if test$b[i] == NA, test$b[i] = (test$a[i] + test$b[i])/2
else leave test$b[i] as it is.

I'm sure this kind of question has been answered many times but I can't find (or recognise) something analogous. Thanks in advance.

Upvotes: 1

Views: 500

Answers (1)

akrun
akrun

Reputation: 887851

You can create a logical row index ('indx') for the elements that are 'NA' in the 'b' column. Use that to replace the NA values in 'b' by taking the `rowMeans of the columns other than 'b'. (Modified based on comments from @thelatemail)

indx <- is.na(test$b)
test$b[indx] <- rowMeans(test[indx,], na.rm=TRUE)

test
#   a   b  c
#1  2 3.5  5
#2 NA 3.0 NA
#3  3 5.0  7
#4  4 6.5  9

Upvotes: 5

Related Questions