14

I have a data frame, and I want to do some calculation with existing columns and create new column in my data set which is a combination of existing... I can do this easily outside function... but if I wrap the code witin function, the changes I made (inside functions) are not visible outside function... i.e. the new column doesn't exist...

I would appreciate sample code to do this...

user318247
  • 665
  • 3
  • 7
  • 13
  • It would help to know what software you're using, and to see a minimal example where things go wrong. – caracal Jun 15 '11 at 13:53
  • 2
    if it is about R you need to return new values out of functions either by `return(data_frame_I_work_with)` or `return(column_I_made)` – Dmitrij Celov Jun 15 '11 at 14:05

2 Answers2

13

I'll assume it is about R... R does not pass arguments by reference (environments and reference classes (S5) are an exception, but this is out of the current range of abstraction). Thus, when you write

addThree<-function(x){
 x<-x+3
}
4->y
addThree(y)

y is still 4 at the end of code, because inside the function, x is the fresh copy of ys value, not the y itself (again, not exactly, but those are higher order details).

Thus, you must adapt to R's pass-by-copy scheme and return the altered value and assign it back to your variable (using old wording, there are no procedures in R):

addThree<-function(x){
 return(x+3)
}
4->y
addThree(y)->y
#y is now 7

Don't worry, this works smoothly for even more complex objects because R is garbage-collected and has lazy evaluation.

BTW, you can omit return if you want to return the last value produced in function, i.e. addThree's definition may look like this:

addThree<-function(x) x+3
mbq
  • 18,510
  • 6
  • 49
  • 72
2

the best approach is to use mutate() from dplyr library. Example:

addcol = function(dat){
    dat1 = mutate(dat, x2=x1*2)
    return(dat1)
}

dat is a data frame with a column named "x1". Use this function addcol(), the new dataset now has a new column named "x2" which is twice the value of "x1", assuming x1 is numeric.

YakovL
  • 7,557
  • 12
  • 62
  • 102
zephyr
  • 21
  • 1