How to use vars in lm in an own function?

Question

I want to write a function, which calculates a linear regression based on the input.

I can build the function, but when I call it (e.g. myregression(i1,i2) it will result in an error)

myregression <- function(input1, input2) {
   model <- lm(data = trainData, example ~ input1 + input2)
}

How can I use the input in the function lm?

You may need `paste0("example ~", input1, " + ", input2)` and use `myregression("i1", "i2")` — akrun, Apr 22 '19 at 04:36
Another approach assuming that `input1` and `input2` are character strings giving the names of columns in `trainData` is `lm(trainData[c("example", input1, input2)])` . — G. Grothendieck, Apr 22 '19 at 04:39

akrun · Accepted Answer · 2019-04-22T04:46:00.133

1

Inside the function, we can use paste to create the formula

myregression <- function(input1, input2) {
    model <- lm(data = trainData, paste0("example ~", input1, " + ", input2))
     }

Or another option is reformulate

myregression <- function(input1, input2) {
      model <- lm(data = trainData, reformulate(c(input1, input2), "example"))
  }

and call the function as

myregression("i1", "i2")

edited Apr 22 '19 at 04:46

answered Apr 22 '19 at 04:39

akrun

1 Answers1