Question

1. What is the drawback of using the step_wise model in multiple linear regression?
2. How is feature importance addressed in decision trees?
3. Is there a guarantee that an ensemble method always outperforms a simple decision tree? Elaborate on your answer.

Answer 1

Answer :- In step wise model , when we follow it, it not process for one final result, it means step by step model constructed for process step with the multiple decision path. When we return back it gives us lot of ways and make confused which way is appropriate according to our decision or result.

Sometimes in step by step process we process more then two modules simultaneously according to its need or work.where we never reach in one final destination or decision.

But when we take decision tree as comparing process it gives us one destination point according to the information gaining. We calculate it according to majority of true values.

Decision tree construct on bases of data entropy calculation. Where the maximum possible entropies checked in each steps and processing accordingly. In decision tree we get majority data decision process destination.

In ensemble method we combine several decision tree to find the better possibility result. And accordingly we get different decision module by ensemble methods. No doubt it compares more decision tree and provide us a better modules methods. But according to me it not guaranty it's not always outperform the decision tree. Because it more focus on true value and combined more decision trees decision point according to information gathering and and gives the highest possibility. And sometimes it becomes not appropriate decision whatever we input accordingly.

Ensemble method is better than decision tree but not guaranteed it always outperforms.