XgBoost (Extreme Gradient Boosting) : Data Science Interview Questions MCQs

Topic: XgBoost (Extreme Gradient Boosting)

 

1. Does gradient boosted trees generally perform better than random forest?

A) Yes
B) No
C) Can't say

ANSWER= A) Explain:- Because Random forest build trees in parallel, while in boosting, trees are built sequentially.

 

2. Which out of random forests and boosting trees are more prone to overfitting?

A) Random Forest
B) Boosting Trees
C) Both
D) Can't say

ANSWER= B) Explain:- Both random forests and boosting trees are prone to overfitting, But boosting models are more prone.

 

3. Which of the following is true about Xgboost?

A) Like any other boosting method, XGB is sensitive to outliers
B) Unlike LightGBM, in XGB, one has to manually create dummy variable/label before feeding to the model
C) A or B both
D) None of these

ANSWER= C) Explain:-

 

4. Which of the following is true about weight of XGB leaf node ?

A) “leaf weight” can be said as the model’s predicted output associated with each leaf (exit) node.
B) “leaf weight” can be said as the actual associated with each leaf (exit) node
C) A or B depend on the situation
D) None of these

ANSWER= B) Explain:- “leaf weight” (derived while training) can be said as the model’s predicted output associated with each leaf (exit) node.

 

5.How change in max_depth parameter value impact the tree based model? ?

A) Higher the value, high likely to get overfit
B) Lower the value , high likely to overfit
C) Higher the value, No impact
D) Lower the value, no impact

ANSWER= A) Explain:- If the value of max_depth is increased, the model would be prone to overfit.

 

6.What does parameter n_estimator represent in Tree based model?

A) Number of feature
B) Number of parameter to be tuned
C) Number of ensembled trees
D) None of these

ANSWER= C) Explain:- Number of ensembled trees.

 

7. What is default value of max_depth?

A) 4
B) 5
C) 6
D) None of above

ANSWER=C) Explain:- Default value=6 and only if growing_policy=lossguide, max_value=0.

 

8. Which parameter is tuned for L1 and L2 regularization on the weights in Tree based Model?

A) Alpha
B) Lambda
C) Alpha, Lambda
D) None of these

ANSWER= C) Explain:- Alpha and Lambda parameters are tuned for L1 and L2 regularization on the weights. Encourages the small weights. Default=1..

 

9. Which parameter is tuned to decide how many data samples will be trained?

A) n_estimator
B) gamma
C) subsample
D) None of these

ANSWER= C) Explain:- How many data samples will be trained? Default=1 means 100% percentage, if it is set 0.5, 50% of data is chosen randomly.

 

10. What are the parameters used to control overfitting in Tree based model?

A) max_depth, subsample
B) min_child_weight, max_depth
C) max_depth, n_estimator
D) None of the above

ANSWER= B) Explain:- min_child_weight and max_depth parameter used to control overfitting. min_child_weight too high values can cause underfitting. Default=1.Whereas max_depth high values can cause overfitting