How does predict with ensemble regression works?

This might be a silly question...iam using bagging with regression trees. I built for instance 10 different trees, then I average them. when I then use predict to predict my y value, how does it work? does it somehow average the fitted y values?!

thanks for your help!!

Answer 1

[Shashank Prasanna]

For classification the output of the ensemble is the class that gets majority vote from all the learners. For regression the results are averaged.

http://www.mathworks.com/help/stats/ensemble-methods.html#bsw8at7

http://en.wikipedia.org/wiki/Bootstrap_aggregating

[Tania]

So for instance, if i have 10 observations and I built 5 different trees. i take all their predicted y values of the different trees and average them,yes?

[Machine Learning]

Short answer is yes.

However if you want a more accurate y you have to look into boosting.