Introduction to Decision Trees Answers 1. What is the main purpose of pre-pruning in decision trees? A. To increase the number of branches in the tree. B. To enhance the accuracy of the model. C. To prevent overfitting by stopping tree growth early. D. To improve recall by adding more nodes. Answer: To prevent overfitting by stopping tree growth early. (C) Pre-pruning is a method used to prevent the tree from growing too complex and overfitting the training data. 2. Which evaluation metric considers both false positives and false negatives? A. F1-score B. Precision C. Recall D. Accuracy Answer: F1-score (A) The F1-score is the harmonic mean of precision and recall, making it sensitive to both false positives and false negatives. 3. What method can be used to handle missing values in a dataset? A. Ignoring all instances with missing data B. Feature scaling C. Pruning unnecessary branches D. Imputation by replacing missing values with estimated values Answer: Imputation by replacing missing values with estimated values (D) Imputation involves replacing missing values with estimated values to make the dataset complete for analysis. 4. In what scenario can feature scaling improve a decision tree's performance? A. When using split criteria that require numerical limits. B. When ensuring a balanced tree structure. C. In scenarios where all features are categorical. D. When decisions are made purely based on bias. Answer: When using split criteria that require numerical limits. (A) Feature scaling can enhance performance when using certain split criteria that are sensitive to the scale of features. 5. Which application is NOT typically associated with decision trees? A. Credit risk assessment B. Medical diagnosis C. Image classification D. Performance monitoring Answer: Performance monitoring (D) Performance monitoring is not a typical application of decision trees, which are instead used in predictive analytics like medical diagnosis and credit risk assessment. 6. What do decision trees use as a model representation? A. A list structure B. A tree-like graph C. A circular diagram D. A linear graph Answer: A tree-like graph (B) Decision trees are represented in a tree-like format where nodes represent features, branches represent decision rules, and leaf nodes represent predictions. 7. Which criteria are commonly used for splitting the data in decision trees? A. Correlation and covariance B. Gini impurity and entropy C. Standard deviation and range D. Mean squared error and variance Answer: Gini impurity and entropy (B) Gini impurity and entropy are two primary criteria employed to determine the best feature for data splitting in decision trees. 8. What is a potential downside of using decision trees? A. They are inherently simple and lack interpretability B. They can be prone to overfitting C. They require extensive pre-processing of data D. They only work with categorical data Answer: They can be prone to overfitting (B) Decision trees can easily overfit the training data, especially when allowed to grow deeply without constraints. 9. What is the purpose of pruning in decision trees? A. To improve computational efficiency B. To enhance the decision-making process C. To increase the complexity of the tree D. To reduce overfitting Answer: To reduce overfitting (D) Pruning is a technique employed to reduce overfitting by removing sections of the tree that provide little power in predicting target values. 10. What do classification trees predict? A. Numerical target variables B. Continuous numerical distributions C. Categorical target variables D. Random sample selections Answer: Categorical target variables (C) Classification trees are utilized to predict categorical outcomes, as opposed to regression trees, which predict continuous values. 11. Which statement best describes the root node of a decision tree? A. It represents a branch of decision rules. B. It represents the final prediction. C. It represents a feature used for splitting. D. It represents the entire dataset. Answer: It represents the entire dataset. (D) The root node is the topmost node in the tree structure and signifies the entire dataset from which decisions will be made. 12. Why are decision trees considered non-parametric? A. They must always fit a specific model type. B. They generate predictions through linear regression. C. They require a large amount of data. D. They do not make assumptions about the data's distribution. Answer: They do not make assumptions about the data's distribution. (D) Decision trees are classified as non-parametric because they do not assume any underlying distribution for the data during the modeling process. 13. Which of the following is NOT a characteristic of decision trees? A. They can only handle categorical input features. B. They require minimal pre-processing of data. C. They are capable of analyzing feature importance. D. They can visualize decision rules easily. Answer: They can only handle categorical input features. (A) Decision trees can handle both numerical and categorical data, making them versatile in modeling different types of datasets.