Decision tree classifier's square brackets

Question

I am learning the decision tree classifier. The class here is whether or not a person will accept a loan offer (orange= not accept, blue=accept). I don't understand what do values inside the square brackets mean. I read in the tutorial saying that they represent the records per class. For example, total sample is 3000, I infer 2713 persons will not accept loan offer=> go to left node, 287 cases to go to right node. But the tree shows the left node value is 2363. And the left node with 2363 is already a leaf node which can not be splited further, why is there a bracket with 2326 and 37 values?

Do I misunderstand something? Could anyone explain what square brackets in a decision tree represent?

Answer 1

2363 have an income less than or equal to 110.5. For each node the numbers in square brackets are the “distribution” of yes/no loan uptake, so for example overall 2713 did not take out a loan. I think one way to look at it is actuals and predictions. So the model predicts loan non-uptake for income less than or equal 110.5, and it gets 2326 of these correct and the other 37 (who do take a loan) incorrect. Why does this node not get split further? It depends on the “splitting” rule, and there are numerous ways this is done. This may be done to avoid overfitting. If splitting a node further does not sufficiently increase homogeneity in the predicted values then it may be made a leaf node.