18

I have created a dataframe "killers" with 3 variables. The data are numeric though there exist NA values throughout.

My goal is to calculate the mean on each of the 3 variables. 

sapply(killers, function(x) median)

This returns:

$heartattack
function (x, na.rm = FALSE) 
UseMethod("median")
<bytecode: 0x103748108>
<environment: namespace:stats>

I know that the na.rm argument is a means to ignore NA values. Since na.rm = FALSE exists in what was returned by R, one presumes that there is a way to set this to TRUE within the line of code above. I tried a few variations:

sapply(killers, na.rm=TRUE function(x) median)
sapply(killers, function(x) median, na.rm=TRUE)
sapply(killers, function(x) median(na.rm=TRUE))

I'm not sure if I'm close or if this is going to involve nesting functions, as per other similar (though ultimately not helpful in this instance that I can see) posts on the topic on SO. e.g. How to pass na.rm as argument to tapply?, Ignore NA's in sapply function

Of course, I could just calculate the mean on each vector that was used to create killers, but surely if what I'm asking is possible then that is better.

Community
  • 1
  • 1
Doug Fir
  • 19,971
  • 47
  • 169
  • 299
  • 3
    `?sapply` says "...: optional arguments to 'FUN'." and there are examples of how to pass additional arguments to functions in the *Examples* section. – Joshua Ulrich Jan 22 '13 at 16:38
  • @JoshuaUlrich not meant to be that offensive. – user1317221_G Jan 22 '13 at 16:41
  • 2
    You asked `sapply` to return the function `median` and that's exactly what it did. Functions are objects. (What you did not do was tell it to return `median(x)`.) – IRTFM Jan 22 '13 at 17:55

1 Answers1

42

Just do:

sapply(killers, median, na.rm = TRUE)

An alternative would be (based on your code)

sapply(killers, function(x) median(x, na.rm=TRUE))
Jilber Urbina
  • 58,147
  • 10
  • 114
  • 138