CODEWITHSUNDEEP
r
igmoreira
igmoreira

Reputation: 35

R - object formula not found inside a function

I created a function to roll apply an exponentially weighted least-squares using the dynlm package. Here is the code:

residualization<-function(df,formula_ref, size){
  rollapply(df,
            width=size,
            FUN = ewma_regression,
            formula_ref = formula_ref,
            by.column=FALSE, align="right")
}

ewma_regression<-function(x,formula_ref) {
    n<-nrow(x)
    weights <- 0.06*0.94^(seq(n-1,0,by=-1))
    t <- dynlm(formula=as.formula(formula_ref), data = as.zoo(x),weights = weights)
    return(t$residuals) 
}

However when I run this code on my dataset, it shows the problem:

Error in as.formula(formula_ref) : object 'formula_ref' not found

When I try to debug it, inside the environment of the function, the variable formula_ref does exist! However even inside the debug mode, I cannot run the dynlm regression even if I try to set formula_ref to a temporary formula object.

Can anyone help me out? I know it might be a silly mistake but I can't find out!

A reproducible example would be:

dates<-seq.Date(from=as.Date("2010-01-01"), length.out = 1000, by="day")
teste1<-data.frame(x=rnorm(1000),y=rnorm(1000)*5) 
teste2<-xts(teste1,order.by = dates)

formula.test<- y ~ x + I(x^2)

teste3<-residualization(df=teste2,formula_ref = formula.test, size=100)

Upvotes: 0

Views: 459

Answers (1)

mnist
mnist

Reputation: 6956

You can just wrap y ~ x + I(x^2) in quotation marks ("y ~ x + I(x^2)").

Upvotes: 1

Related Questions