JP Morgan Interview Questions | Decision Tree
Question
Do you think 50 small decision trees are better than a large one? Why?
in progress
0
Machine Learning
4 years
4 Answers
973 views
Great Grand Master 0
Answers ( 4 )
yes. this is the idea of bagging because multiple decision trees trained on different data will capture different variances each time hence the resultant model will improved performance than all the the individual units
yes to an extent. Though increasing the number of decision trees will help in our predictive model by taking the average of all the results. But if you increase too much there can be a problem of overfitting. So we need to be careful in increasing the number of trees
Yes it is an ensemble method that takes many weak decision trees to make strong learner, also are more robust accurate.
Yes, that’s the rationale behind ensemble methods like Bagging, Boosting, RandomForests.
Building more trees helps us in reducing the variance and the model generalizes well over
the training set. How you tune the hyperparametres while building these trees is also
very important.