When I use the same data set to estimate a tree and a logistic regression model. Is there a way of comparing the resulting probabilities? In my view, there is not. Because, for logistic regression, I look at the effect a particular feature has on the outcome holding all the features constant (for example). But for classification trees, I only get conditional probabilities dependent on features further up the tree. Is this a correct reasoning?
Asked
Active
Viewed 50 times
1

kjetil b halvorsen
- 63,378
- 26
- 142
- 467

Patrick Balada
- 609
- 9
- 16
-
1There is no reason you cannot compare them, as a way of helping understand the diference between the models. – kjetil b halvorsen Sep 07 '17 at 10:58
-
The following could be relevant: https://stats.stackexchange.com/questions/164048/can-a-random-forest-be-used-for-feature-selection-in-multiple-linear-regression/164068#164068 – kjetil b halvorsen Sep 07 '17 at 11:09