Tree-Based Learning DY0-001 Practice Questions (Page 1)

Questions 1–5

1. 1application · hard

   A data scientist is building a decision tree and is deciding between using Gini impurity and information gain as the splitting criterion. The dataset has many categorical features with high cardinality. Which consideration is most relevant when choosing between these criteria?

   • AInformation gain tends to favor features with many categories, which can lead to overfitting; Gini impurity is less biased toward high-cardinality features.
   • BGini impurity is always better than information gain for categorical features.
   • CGini impurity is only applicable to binary classification problems.
   • DInformation gain is computationally faster than Gini impurity.
   Reveal Solution Discussion Select an answer first
2. 2foundation · easy

   In a decision tree, which component represents the outcome of a series of decisions and does not have any child nodes?

   • AInternal node
   • BBranch
   • CLeaf node
   • DRoot node
   Reveal Solution Discussion Select an answer first
4. 3application · medium

   A data scientist is tuning a random forest model. The current model has high variance, and the scientist wants to reduce overfitting. Which hyperparameter adjustment is most likely to help?

   • ADecrease the number of trees in the forest.
   • BIncrease the minimum samples per leaf.
   • CDecrease the minimum samples per leaf.
   • DIncrease the maximum depth of the trees.
   Reveal Solution Discussion Select an answer first
5. 4foundation · easy

   What is the primary role of bootstrap aggregation (bagging) in ensemble learning?

   • ATo eliminate the need for feature selection
   • BTo reduce variance and improve model stability by averaging predictions from models trained on bootstrap samples
   • CTo reduce bias by sequentially focusing on misclassified samples
   • DTo increase the depth of individual trees
   Reveal Solution Discussion Select an answer first
6. 5application · medium

   A data scientist trains a decision tree on a training dataset and achieves 99% accuracy, but only 72% accuracy on a held-out validation set. The tree has many deep branches with small leaf sizes. Which action is most appropriate to improve validation performance?

   • AUse a different splitting criterion, such as information gain, to automatically prevent overfitting.
   • BRemove the validation set and retrain on all data to improve accuracy.
   • CPrune the tree by removing branches that provide little predictive power.
   • DIncrease the maximum depth of the tree to capture more complex patterns.
   Reveal Solution Discussion Select an answer first