How to pass a record to a decision tree?

Question

I'm working a small project. Lets say, I have a table of around 100K records containing columns like Age, Gender, Region, Life(nominal - range of days the products is used) etc., Here Life is a dependent variable and all others are independent variable.I created a decision tree out of the data available. Now my query is, suppose if I have one new record, I want to know in which terminal node that record falls after traversing the decision tree i.e., under which Life range does that record falls. For that, how can I pass that record to the decision tree and get a output?

Answer 1

predict(model,newdata)

Let's say your original data.frame had 4 columns as you list in your question. Your new record would need to be formated as a data.frame with the same columns names as your independent factors, e.g., newdata = data.frame(Age=15,Gender="Male",Region="Southwest") or whatever those values should be. Let's assume you've stored your model thusly model = rpart(Life~.,data=data,method="class") then predict(model,newdata) will return a vector of the probability that the new record belongs to each of the terminal classes. You then need to have some cutoff logic to determine which group you'll assign it to.