Que3. You want to improve model generalization by combining predictions from multiple models, each trained on bootstrapped versions of the dataset. Which technique should you use?
a) Decision tree pruning b) Ridge regression c) Recursive feature elimination d) Bootstrap aggregation (bagging)
Que4. While constructing a binary decision tree, you encounter a discrete-valued attribute A. How does the decision tree handle dataset splitting in this case?
a) By forming two branches: A ≤ split_point and A > split_point. b) By creating one branch for each distinct value of A. A ∈ SA, where SA c) By applying a test A ∈ SA, where SA is a subset of values of A. d) None of the above
Que6. The primary goal of tree pruning in decision tree algorithms is to avoid overfitting. How is this achieved?
a) Improve performance by growing the tree and avoiding underfitting b) Increase complexity by expanding the tree and avoiding overfitting c) Prevent overfitting by simplifying the tree and reducing unnecessary branches d) Enhance accuracy by deepening the tree and reducing training bias
Que7. In the post-pruning of a decision tree, the leaf node is assigned the most frequent class label among the subtree being replaced. Is this statement true or false?
Que8. The bagging method improves predictive accuracy by using multiple models. How are these models handled?
a) Weights are assigned randomly b) All models receive equal weight c) Models are weighted based on their performance d) More recent models are given more importance
Que10. Random Forests are widely used in machine learning. Which of the following statements about them is incorrect?
a) Random forests use bagging and random selection of features at each node to train decision trees. b) A Random Forest model is built from many decision trees, and each tree is trained using different random samples of data and features. c) The number of features chosen at each split is a critical factor in determining the success of the Random Forest. d) Random Forests’ accuracy is determined by the individual decision trees’ accuracy and their mutual dependence.
Que12. In classification tasks with imbalanced data, techniques like oversampling and undersampling are used. What is a common way to handle such imbalance?
a) Oversampling the minority class and undersampling the majority class b) Undersampling the minority class and oversampling the majority class c) Oversampling and random sampling of data d) Using SMOTE for oversampling and undersampling the majority class
Que13. Imagine you’re analyzing the purchase behavior of customers on a popular online store during a seasonal sale. You want to assess the Gini indices for customer actions after splitting by the “Purchase Category” feature.
Node 1 (left child): Out of 30 customers, 15 added items to the cart but didn’t purchase (“No Purchase”) & 15 completed their purchase (“Purchase”).
Node 2 (right child): Out of 70 customers, 30 abandoned their cart (“No Purchase”) & 40 went ahead and purchased the items (“Purchase”).
Which option has the correct Gini indices for the child nodes?
a) Gini index for Node 1: 0.500, Gini index for Node 2: 0.428 b) Gini index for Node 1: 0.375, Gini index for Node 2: 0.370 c) Gini index for Node 1: 0.400, Gini index for Node 2: 0.375 d) Gini index for Node 1: 0.400, Gini index for Node 2: 0.500
Que14. Decision trees and linear regression models have different advantages. In a customer satisfaction prediction model, which of the following is an advantage of decision trees over linear regression?
a) Decision trees are less affected by outliers. b) Decision trees are easier to explain and interpret. c) Decision trees require the creation of dummy variables for qualitative predictors. d) Decision trees are more accurate in predicting continuous outcomes.
Que15. When constructing a decision tree for classifying customers based on their purchase behaviors, different heuristics are used to select the best split. Which of the following heuristics are typically used?
a) Information gain, Gain ratio, Gini impurity b) Information gain, Gini impurity, Residual sum of squares c) Gain ratio, Residual sum of squares, Entropy d) Precision, Gini impurity, Information gain
