is.data.frame(data) object ... not found in function context

Question

Have a strange problem using package tree cv.tree function from user defined function in R:

func <- function(train) {
    classification.tree <- tree(log10(perf) ~ syct+mmin+mmax+cach+chmin+chmax, train, split = "gini")
    cv.tree(classification.tree, ,FUN=prune.tree, K = 4)
    return (classification.tree)
}
data(cpus, package="MASS")
result <- func(cpus)
plot(result)

This produces the error:

Error in is.data.frame(data) : object 'train' not found 
16 is.data.frame(data) 
15 model.frame.default(formula = log10(perf) ~ syct + mmin + mmax + 
    cach + chmin + chmax, data = train, subset = 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",  ... 
14 eval(expr, envir, enclos) 
13 eval(expr, p) 
12 eval.parent(m) 
11 tree(formula = log10(perf) ~ syct + mmin + mmax + cach + chmin + 
    chmax, data = train, split = "gini", subset = 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",  ... 
10 eval(expr, envir, enclos) 
9 eval(oc) 
8 model.frame.tree(object) 
7 model.frame(object) 
6 cv.tree(classification.tree, , FUN = prune.tree, K = 4) at .active-rstudio-document#4
5 func(cpus) at .active-rstudio-document#9
4 eval(expr, envir, enclos) 
3 eval(ei, envir) 
2 withVisible(eval(ei, envir)) 
1 source("~/.active-rstudio-document", echo = TRUE) 

At the same time if I invoke same code right from the script it works perfectly fine:

data(cpus, package="MASS")
classification.tree <- tree(log10(perf) ~ syct+mmin+mmax+cach+chmin+chmax, cpus, split = "gini")
cv.tree(classification.tree, ,FUN=prune.tree, K = 4)
plot(classification.tree)

What am I missing?

Answer 1

The crash is in the cv.tree() call. update: cv.tree calls model.frame , and inside that function there's an eval, but the variable train does not exist in that function's environment.

I'll keep digging.... If I go deep into debug mode for model.frame and change the data list element of the 'object' from 'train' to 'cpus', then eval finds the object and executes.

Annnnd: I'm back where I started. It is an environment and lazy evaluation problem.
The fix is to use force :

func <- function(train) {
    force(train)
    classification.tree <- tree(log10(perf) ~ syct+mmin+mmax+cach+chmin+chmax, train, split = "gini")
    cv.tree(classification.tree, FUN=prune.tree, K = 4)
    return (classification.tree)
}

This makes "train" exist in the environments available to cv.tree and the functions it calls. Environments can get weird :-) ; this is one example thereof.