Question

Machine Learning : Please provide explanation with the answer

How many parameters do we need to perform boosting? Write the name of the parameters next to your selection b) 2 d) 4 How many parameters do we need to perform random forest? Write the name of the param eters next to your selection. b) 2 d) 4 What is random in random forest algorithm? a) at each split a random sample of m predictors is chosen as split candidates b) at each terminal a random selection of response is done c) the parameter m is selected randomly d) nothing is random in random forest algorithm Which of the following statements is TRUE about the comparison of cross validation (CV) error for bagging, random forest and boosting? a) Boosting always have a better CV error if the depth is chosen properly b) Bagging always have a better CV error c) Random forest always have a best CV error than bagging an boosting if the m is chosen properly d) Performance comparison of CV error is not possible in general

Answer 1

1.C = 3

Those are

1. Adaptive Boost
2. Gradient Boosting
3. XGBoost.

2.D = 4

THose are

1. max_features
2. mn_samples_leaf
3. min_samples_split
4. max_depth

3.A = Split the candidates

Random Forest is the special type of the case Where each split can be random that is the sample of m predictors is chosen to split the candidates.

4.B = Have Better CV error

Here Boosting always have the Better CV error.