How does a regression tree split the y variable? Is it just a case of even chunking of the range or are the chunks of variable size?
Asked
Active
Viewed 34 times
1
-
there needs to be a decision-tree tag, but I cant create it – Trajan Feb 24 '21 at 17:49
-
'cart' covers decision trees – mkt Feb 24 '21 at 19:03
-
When/why does a tree need to split/chunk the target variable? – Ben Reiniger Feb 24 '21 at 20:11
1 Answers
1
The target variable is not split, features are. And, this is typically done via a greedy search over the features and possible thresholds. In the end, usually, the target variable of the samples in the leaves are averaged for outputting a prediction.
gunes
- 49,700
- 3
- 39
- 75