Search code examples
renvironmentforecasting

Object not found within R function

I have a couple of tslm functions (from the forecast R package) inside a bigger function. I cannot seem to get them both to find their relevant objects. Including an explicit environment fixed one instance and broke the other. A minimal example follows:

library(forecast)
data(gas)

testlm <- function(x)
{
e<-new.env()
e$x<-x
   tslm(e$x~trend)->e$z
return(e$z)
}

testlm(gas)

This throws the following error:

 Error in eval(expr, envir, enclos) : object 'e' not found

I also tried to explicitly give the formula the correct environment by setting

e$f<-as.formula("e$x~trend+season",env=e)

but got the exact same error.

P.S. The other error seems harder to reproduce, but getting this to work with the explicit environment should be enough.

Solution

  • Finally figured it out! To avoid messing with tslm or lm's environments you can just put the data as a separate parameter. Changing the tslm line to

    tslm(data ~ trend,data=e$x)

    works normally.