CODEWITHSUNDEEP
rsapply
Boro Dega
Boro Dega

Reputation: 393

understanding sapply function in r

This might be the most stupid question, but I do not seem to grasp the function sapply: Here is my issue

      Example: 
      d = matrix(1:10, 5,2)
      d[3] = NA

 #              [,1] [,2]
         #[1,]    1    6
         #[2,]    2    7
         #[3,]   NA    8
         #[4,]    4    9
         #[5,]    5   10

If I would like to calculate the row means using sapply function I would use something like this: 

       sapply(d,mean)
       #[1]  1  2 NA  4  5  6  7  8  9 10

Should it not give me the mean of the list of the elements? It just spits out the elements of my matrix rather then the mean.

When I use apply, I get the right answer: 

   apply(d,1,mean, na.rm=T)
   [1] 3.5 4.5 8.0 6.5 7.5

Can anyone bother giving me a very dummy explanation.Highly appreicated. Used the following links before asking the question. link 1 Link 2 Link 3

Upvotes: 0

Views: 16153

Answers (2)

David_B
David_B

Reputation: 926

sapply (and its friends, like lapply) require a list (or a data.frame, which is really a special kind of list) as input. But even if you had turned your matrix into a data frame, it wouldn't have given you row means, it would have given you column means. If you want to understand how these functions work, it might help to look at this function (copied from here: http://adv-r.had.co.nz/Functionals.html), which shows the essence of lapply using base R. (sapply works in the same way, it just tries to simplify the output to a vector rather than always returning a list.)

lapply2 <- function(x, f, ...) {
 out <- vector("list", length(x))
 for (i in seq_along(x)) {
   out[[i]] <- f(x[[i]], ...)
  }
  out
}

By the way, if the point of your question is to find the best way of calculating row means, rather than understanding sapply, then there is a function rowMeans that is the fastest way of doing it -- faster than using apply, for example.

Upvotes: 2

RockScience
RockScience

Reputation: 18580

have you read ?sapply. The function takes either a vector or an expression object. It cannot take a matrix. So what happens in your example is that the matrix d is considered as a vector of numeric:

sapply(as.numeric(d),mean)

Upvotes: 2

Related Questions