mincriterion with ctree2 model

[asardaes]

I believe that mincriterion shouldn't be set to 0 when using a ctree2 model, the user should be able to define it. I know ctree2 uses maxdepth as tuning parameter, but they are rather independent. Here's an example:

library(party)
data("iris")

# No restrictions on depth, mincriterion = 0.95
model_1 <- ctree(Species ~ ., data = iris, controls = ctree_control())
# Just to know the obtained depth (which is 4)
depth <- party:::maxdepth(model_1@tree)

model_2 <- ctree(Species ~ ., data = iris, 
                 controls = ctree_control(maxdepth = 4, mincriterion = 0.95))
model_3 <- ctree(Species ~ ., data = iris, 
                 controls = ctree_control(maxdepth = 4, mincriterion = 0))

plot(model_1)
plot(model_2)
plot(model_3)

You can see that models 1 and 2 are identical, but model 3 isn't.

[topepo]

Good point. I added the extra parameter to the model.