Reuse an object created in a function in R

Question

I have searched SO and found a similar question, but I still can't get it to work.

I am rather new to R programming, so I appreciate this help.

Suppose I have a data.frame:

df <- data.frame(State = c("NJ", "MA", "CA", "CA", "MA"), 
                 Sales = c(15, 100, 30, 56, 60), 
                 Price = c(34, 52, 21, NA, 20), 
                 stringsAsFactors = FALSE)

Now I want a function that creates a new data.frame that just has a given state... So I wrote:

state_function <- function(state) {
        #Subset the rows in the state from all rows
        new_df <- df[df[,"State"] == state, ]
         ## I will do many other things inside the function ##
        summary(new_df)
        return(new_df)
}

state_function ("NJ")

But now I want to re-use this data.frame, new_df , outside of this function ..(and I want to generalize this example) .. How can I do this? In general, I want to use a object or data.frame that I create in a function, outside of a function.

Answer 1

Assign the value to a variable when running the function:

new_df <- state_function("NJ")

Answer 2

I think you want this:

state_function <- function(state) {

   new_df <- subset(df, State == state)
   print( summary(new_df) )
   new_df

}

new_df <- state_function("CA")

Note that we can use the subset function to clean up the first line of the body. Also the summary line in the question does nothing without print. Automatic printing does not occur within functions. Also the return is not really needed. Finally assign the output of the function to a variable.

Answer 3

You can also use assign to assign a global variable within a function. Be careful though, as y will be overwritten each time you run the function.

state.function <- function(x, state)
{
    assign('y', subset(x, x$State == state), envir = .GlobalEnv)
    return(y)
}

> state.function(df, "CA")
#   State Sales Price
# 3    CA    30    21
# 4    CA    56    NA
> y
#   State Sales Price
# 3    CA    30    21
# 4    CA    56    NA

This works well if you're going through data quickly. If you remove return(y) there will be no output to the function, but the variable y will exist in the global environment. This is sometimes useful.

And if you hate quoting things, like I do

state.function <- function(x, state)
{
    z <- deparse(substitute(state))
    assign('y', subset(x, x$State == z), envir = .GlobalEnv)
}

> state.function(df, CA)
> y
#   State Sales Price
# 3    CA    30    21
# 4    CA    56    NA