Fill matrix from data frame

Question

I have a data frame which contains some data (Claims) and two additional columns, containing row and column information, for example:

Claim row Column

5000 2 10

etc.

I have a matrix that contains some data as well as NAs. I'd like to replace the NAs with the data from the data frame, using the row/column information. So 5000 would replace the NA in the matrix that is currently in row 2, column 10.

Is there an easy way to accomplish this with R?

Can you provide a [minimal reproducible example](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example)? and What have you tried? — Jilber Urbina, Sep 13 '13 at 09:05

score 2 · Answer 1 · edited Sep 13 '13 at 10:30

2

# If you have for example a matrix like this
m <- matrix(c(1:9), ncol=3, nrow=3)

# and information column such as:
info <- rbind(c(5000,2,2), c(8000,1,3))

# you can do like this:
m[info[,c(2:3)]] <- info[,1]

edited Sep 13 '13 at 10:30

Simon O'Hanlon

58,647
14
142
184

answered Sep 13 '13 at 10:30

score -1 · Answer 2 · answered Sep 13 '13 at 09:38

I think I've figured out a (bad) way of doing it: Loop through all the rows in the matrix, select all NA elements in the matrix and assign to the resulting vector a subset of the data frame where the row column matches the current row in the loop, i.e. something like this:

# Fill triangle
new_data <- data.frame(Claim = pred_claims, Column = cols, row = ro) # data frame 
full_triangle <- auto_incr # matrix to be filled

for (i in 2:10) {
  full_triangle[i,is.na(full_triangle[i,])] <- subset(new_data, row==i)$Claim
}

Fill matrix from data frame

2 Answers2