Cannot create tree plot using C5.0

[Georg]

Hello,

I'm trying to fit a model using C5.0. My code is the following:

set.seed(123)
library(C50)
library(caret)
control <- trainControl(method = "repeatedcv", number = 10, repeats = 3)
fit_train <- train(vowel ~ F1 + F2 + F3 + duration, data = train, method = "C5.0", trControl = control)
plot(fit_train)

The problem is that I want to plot the tree, but instead of that, I get a plot of boosting iterations when I use the plot() function (see the figure below).

Any suggestions on how to plot the tree?

[fcas80]

Does this work?

plot(fit_train$finalModel)
text(fit_train$finalModel)

[Georg]

No, I get this error:
Error in data.frame(eval(parse(text = paste(obj$call)[xspot])), eval(parse(text = paste(obj$call)[yspot])), : arguments imply differing number of rows: 880, 220, 0

[nirgrahamuk]

something seems wrong with the caret C50 integration ...
I think you can try an approach of getting the parameters of the best fits from the caret return, and then use that to fit a standalone c50model.

c5model <- C5.0(vowel ~ F1 + F2 + F3 + duration,
                       data = train,
                       trials = fit_train$bestTune$trials, 
                       rules = FALSE,
                       control = C5.0Control(winnow = fit_train$bestTune$winnow))
plot(c5model)

[Georg]

This seems to work, thanks!

[nirgrahamuk]

note that I think you can only plot tree based C50s, not the rule type C50's, hence the rules=FALSE choice I showed