CODEWITHSUNDEEP
rfunctionvariables
Jeff Johnson
Jeff Johnson

Reputation: 53

How to pass variable name into R function

I know this has been explained a few times here in different scenarios, but I am having a hard time digesting this.

The following code works fine as long as it's not inside a function (see below).

df$season <- as.character(df$season) 
      temp <- model.matrix( ~ season - 1, data=df) 
      df <- cbind(df,temp)

BEFORE:

head(df[c(1,2)])
    datetime season
1 2011-01-01      1
2 2011-01-01      1
3 2011-01-01      1
4 2011-01-01      1
5 2011-01-01      1
6 2011-01-01      1

AFTER:

> head(df[c(1,2,13:16)])
    datetime season season1 season2 season3 season4
1 2011-01-01      1       1       0       0       0
2 2011-01-01      1       1       0       0       0
3 2011-01-01      1       1       0       0       0
4 2011-01-01      1       1       0       0       0
5 2011-01-01      1       1       0       0       0
6 2011-01-01      1       1       0       0       0

However, when I try to wrap it in a multi-use function:

binarize <- function(data, myvar) { 
  data$myvar <- as.character(data$myvar) 
  temp <- model.matrix( ~ myvar - 1, data=data) 
  data <- cbind(data,temp) 
}

it throws an error, undoubtedly because it cannot evaluate myvar or data (or both?): Error in $<-.data.frame(*tmp*, "myvar", value = character(0)) : replacement has 0 rows, data has 10886

I've tried experimenting with eval(substitute()) but still it's not working. My ideal end-state is that you start with a dataframe and a variable, have the function map all of the values for the selected variable into separate binary columns and append that to the original dataframe. Again, when it's not in a function it works perfectly.

Here's the dput data if it helps to reproduce.

> dput(head(df,50))
structure(list(datetime = structure(c(14975, 14975, 14975, 14975, 
14975, 14975, 14975, 14975, 14975, 14975, 14975, 14975, 14975, 
14975, 14975, 14975, 14975, 14975, 14975, 14975, 14975, 14975, 
14975, 14975, 14976, 14976, 14976, 14976, 14976, 14976, 14976, 
14976, 14976, 14976, 14976, 14976, 14976, 14976, 14976, 14976, 
14976, 14976, 14976, 14976, 14976, 14976, 14976, 14977, 14977, 
14977), class = "Date"), season = c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L), holiday = c(0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L), workingday = c(0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 1L, 1L, 1L), weather = c(1L, 1L, 1L, 1L, 1L, 
2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 2L, 
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 2L, 3L, 2L, 2L, 2L, 2L, 2L, 
3L, 3L, 3L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L), temp = c(9.84, 
9.02, 9.02, 9.84, 9.84, 9.84, 9.02, 8.2, 9.84, 13.12, 15.58, 
14.76, 17.22, 18.86, 18.86, 18.04, 17.22, 18.04, 17.22, 17.22, 
16.4, 16.4, 16.4, 18.86, 18.86, 18.04, 17.22, 18.86, 18.86, 17.22, 
16.4, 16.4, 15.58, 14.76, 14.76, 14.76, 14.76, 14.76, 13.94, 
13.94, 13.94, 14.76, 13.12, 12.3, 10.66, 9.84, 9.02, 9.02, 8.2, 
6.56), atemp = c(14.395, 13.635, 13.635, 14.395, 14.395, 12.88, 
13.635, 12.88, 14.395, 17.425, 19.695, 16.665, 21.21, 22.725, 
22.725, 21.97, 21.21, 21.97, 21.21, 21.21, 20.455, 20.455, 20.455, 
22.725, 22.725, 21.97, 21.21, 22.725, 22.725, 21.21, 20.455, 
20.455, 19.695, 17.425, 16.665, 16.665, 17.425, 17.425, 16.665, 
16.665, 16.665, 16.665, 14.395, 13.635, 11.365, 10.605, 11.365, 
9.85, 8.335, 6.82), humidity = c(81L, 80L, 80L, 75L, 75L, 75L, 
80L, 86L, 75L, 76L, 76L, 81L, 77L, 72L, 72L, 77L, 82L, 82L, 88L, 
88L, 87L, 87L, 94L, 88L, 88L, 94L, 100L, 94L, 94L, 77L, 76L, 
71L, 76L, 81L, 71L, 66L, 66L, 76L, 81L, 71L, 57L, 46L, 42L, 39L, 
44L, 44L, 47L, 44L, 44L, 47L), windspeed = c(0, 0, 0, 0, 0, 6.0032, 
0, 0, 0, 0, 16.9979, 19.0012, 19.0012, 19.9995, 19.0012, 19.9995, 
19.9995, 19.0012, 16.9979, 16.9979, 16.9979, 12.998, 15.0013, 
19.9995, 19.9995, 16.9979, 19.0012, 12.998, 12.998, 19.9995, 
12.998, 15.0013, 15.0013, 15.0013, 16.9979, 19.9995, 8.9981, 
12.998, 11.0014, 11.0014, 12.998, 22.0028, 30.0026, 23.9994, 
22.0028, 19.9995, 11.0014, 23.9994, 27.9993, 26.0027), casual = c(3L, 
8L, 5L, 3L, 0L, 0L, 2L, 1L, 1L, 8L, 12L, 26L, 29L, 47L, 35L, 
40L, 41L, 15L, 9L, 6L, 11L, 3L, 11L, 15L, 4L, 1L, 1L, 2L, 2L, 
0L, 0L, 0L, 1L, 7L, 16L, 20L, 11L, 4L, 19L, 9L, 7L, 10L, 1L, 
5L, 11L, 0L, 0L, 0L, 0L, 0L), registered = c(13L, 32L, 27L, 10L, 
1L, 1L, 0L, 2L, 7L, 6L, 24L, 30L, 55L, 47L, 71L, 70L, 52L, 52L, 
26L, 31L, 25L, 31L, 17L, 24L, 13L, 16L, 8L, 4L, 1L, 2L, 1L, 8L, 
19L, 46L, 54L, 73L, 64L, 55L, 55L, 67L, 58L, 43L, 29L, 17L, 20L, 
9L, 8L, 5L, 2L, 1L), count = c(16L, 40L, 32L, 13L, 1L, 1L, 2L, 
3L, 8L, 14L, 36L, 56L, 84L, 94L, 106L, 110L, 93L, 67L, 35L, 37L, 
36L, 34L, 28L, 39L, 17L, 17L, 9L, 6L, 3L, 2L, 1L, 8L, 20L, 53L, 
70L, 93L, 75L, 59L, 74L, 76L, 65L, 53L, 30L, 22L, 31L, 9L, 8L, 
5L, 2L, 1L)), .Names = c("datetime", "season", "holiday", "workingday", 
"weather", "temp", "atemp", "humidity", "windspeed", "casual", 
"registered", "count"), row.names = c("1", "2", "3", "4", "5", 
"6", "7", "8", "9", "10", "11", "12", "13", "14", "15", "16", 
"17", "18", "19", "20", "21", "22", "23", "24", "25", "26", "27", 
"28", "29", "30", "31", "32", "33", "34", "35", "36", "37", "38", 
"39", "40", "41", "42", "43", "44", "45", "46", "47", "48", "49", 
"50"), class = "data.frame")

Thanks for the help in advance!

Upvotes: 0

Views: 253

Answers (1)

Rich Scriven
Rich Scriven

Reputation: 99321

You really shouldn't use data$column inside functions, data[[column]] is much safer. But I think you can change the function to something like this. Not sure if it gives you the right result but it gets the evaluation done.

binarize <- function(data, myvar) { 
    form <- substitute( ~ x - 1, list(x = as.name(myvar))) 
    temp <- model.matrix(eval(form), data = data) 
    cbind(data, temp)
}

or even simpler might be 

binarize <- function(data, myvar) {
    form <- as.formula(paste("~", myvar, "- 1"))
    temp <- model.matrix(form, data = data) 
    cbind(data, temp)
}

In either case, the function will need to be called with a character string for the myvar variable, i.e.

binarize(df, "season")

Upvotes: 2

Related Questions