'If error' using data table in 'R'

Question

I am trying to replace NA's in a data set using a 'key'. The 'key' represents similarity based on some factors.

For example in the dataset below, the NA in row 5, can be replaced by 1, 0 or 1 on from rows 1, 2 and 6 respectively.

The code I have below works fine if aren't errors. For example the NA on row 3 errors because based on the key there are no similar rows. I would like to introduce something to say 'If error' (like in excel) replace that NA(not all the others) using a different key, which I will specify.

> s3ITR
  Index Country Age Time Charity
1     1  France  30   40       1
2     2  France  30   40       0
3     3  France  30   50      NA
4     4 Germany  30   40       0
5     5  France  30   40      NA
6     6  France  30   40       1
7     7 Germany  30   40       1
8     8 Germany  30   40       1

DT <- data.table(s3ITR, key = "Country,Age,Time")
DT[, Impute := sample(na.omit(Charity), length(Charity), replace = T), by = key(DT)] 
DT[!is.na(Charity), Impute := Charity]
DT[order(Index)]
S3<-mydf
S3$Charity <- DT$Impute

I'm new to 'R', any ideas ?

You could impute rows with unique Country/Age/Time first and than use your code for the other NA values. — Roland, Feb 10 '14 at 09:53
Please check this: http://stackoverflow.com/questions/13180057/replacing-control-character-in-sed — Konstantinos, Mar 05 '14 at 13:54

'If error' using data table in 'R'

0 Answers0