How to choose the number of splits in a tree

r
decision_trees
#1

I am currently solving one classification problem using CART in R, while doing it I build a tree .I want to know how to know the number of splits in a tree.
My tree.

f6.png1366×650 8.7 KB

#2

hello @sid100158,

If you see the above image for rpart.control you can see that there are several options like minsplit = 20, minbucket etc.These parameters can be used to change the number of splits indirectly.For example:

rpart.fit <- rpart(quality_bin ~ .,data = training)
rpart.plot::rpart.plot(rpart.fit)

gives
.

And

#Rpart with control:
rpart.cntrl <- rpart.control(minsplit = 100)
rpart.cntrl.fit <- rpart(quality_bin ~ .,data = training,control = rpart.cntrl)
rpart.plot::rpart.plot(rpart.cntrl.fit)

will give you:

If you notice the number of splits is different here.
Similarly the other parameters in rpart.control can be used.

Hope this helps!!