CODEWITHSUNDEEP
rfunction
Carlo
Carlo

Reputation: 161

error object not found when calling glmulti within a function

I'm having trouble to use the package glmulti within my own function.

The code below is a simplified example that reproduces the error: Error: object 'poplrun' not found

Which is the data.frame created within the function. In the second example, it doesn't find the argument l.

I think the problem is related to the environment where glmulti is called. I found this post

Trouble passing on an argument to function within own function

and tried to use do.call with substitute(poplrun) or as.name("poplrun") but clearly I'm missing something because it didn't work. I have also found this post Object not found error when passing model formula to another function

and tried to identify the environment within the formula (as you can see in my first example) but that one too doesn't work. I would really appreciate any help with this since it is now two days that I'm trying to solve this puzzle...

Thanks heaps!

Example 1

    set.seed(5)
    df1<-data.frame(Scenario=rep(LETTERS[1:2], each=10), 
                   Iteration=rep(1:10, 2), V1=rnorm(n=20, mean=0.5, sd=0.1))
    LookUpT<-data.frame(Scenario=rep(LETTERS[1:5]), SV1=1:5, SV2=6:10 )
    InteractRun<- function (
      param="V1" , 
      SVs=c("SV1", "SV2"),
      ic="aic",
      l=1 
      ) {
         poplrun<-df1
         require(plyr)
         poplrun<- join(poplrun, LookUpT, by = 'Scenario', type="left")
         xs<-paste(SVs, collapse="*") 
         .env<-environment() 
         formula<-as.formula(paste0(param, "~", xs), env=.env)
         require(betareg)
         require(glmulti)
         cand<-glmulti(formula, data=poplrun, method="d", level=l, 
         fitfunc=betareg,  na.action=na.omit)
         print(cand)
       }
       InteractRun()

Example 2

    set.seed(5)
    df1<-data.frame(Scenario=rep(LETTERS[1:2], each=10), Iteration=rep(1:10, 2), 
            V1=round(rnorm(n=20, mean=20, sd=2))) 
    LookUpT<-data.frame(Scenario=rep(LETTERS[1:5]), SV1=1:5, SV2=6:10 ) 
    InteractRun<- function (
    param="V1" , 
    SVs=c("SV1", "SV2"), 
    fam="poisson",
    ic="aic",
    l=1 
    ) {
       poplrun<-df1
       require(plyr)
       poplrun<- join(poplrun, LookUpT, by = 'Scenario', type="left")
       xs<-paste(SVs, collapse="*") 
       formula<-as.formula(paste0(param, "~", xs)) # set up formula to be used  
       glm1<-glm(data=poplrun, formula, family=fam, na.action=na.omit)
       require(glmulti)
       cand<-glmulti(glm1, method="d", level=l,  na.action=na.omit)
       print(cand)
      }
     InteractRun()

Upvotes: 2

Views: 1263

Answers (1)

Carlo
Carlo

Reputation: 161

I'm going to answer my own question... After a while, I figured out how to tackle the problem. The correct call for glmulti is using do.call but there is no need for either substitute() or as.name(). In the example 1 the call should be: cand <- do.call("glmulti", list(formula, data=poplrun, method="d", level=l, fitfunc=betareg, na.action=na.omit)) and in example 2: cand <- do.call("glmulti", list(glm1, method="d", level=l, na.action=na.omit))

Upvotes: 5

Related Questions