Decision Tree Algorithms Overview Answers 1. What metric evaluates the percentage of correctly predicted positive instances among all predicted positives? A. Accuracy B. Specificity C. Recall D. Precision Answer: Precision (D) Precision focuses specifically on the positive predictions to evaluate their accuracy. 2. Which metric is specifically used to assess the homogeneity of data within a node in decision tree algorithms? A. Gini impurity B. AUC C. F1-score D. Accuracy Answer: Gini impurity (A) Gini impurity is commonly employed in decision trees to evaluate the purity of splits. 3. What is the F1-score primarily used for in the evaluation of models? A. Combining precision and recall into a single metric B. Determining the number of positive instances C. Evaluating overall model performance D. Measuring accuracy of predictions Answer: Combining precision and recall into a single metric (A) The F1-score provides a balance between precision and recall, making it a combined measure. 4. Which metric indicates a model's ability to distinguish between positive and negative classes? A. Specificity B. Gini impurity C. AUC D. Recall Answer: AUC (C) AUC (Area Under the ROC Curve) is specifically designed to evaluate how well a model can differentiate between classes. 5. What does specificity measure in a classification model? A. Positive instances out of predicted positives B. Correctly predicted negative instances out of actual negatives C. Proportion of actual positives among all instances D. Accuracy of correctly classified instances Answer: Correctly predicted negative instances out of actual negatives (B) Specificity evaluates the proportion of true negatives that are correctly identified by the model. 6. Which decision tree algorithm is specifically designed for both classification and regression tasks? A. C4.5 B. QUEST C. CHAID D. ID3 Answer: C4.5 (A) C4.5 improves upon ID3 and can handle both classification and regression tasks. 7. What does CART primarily aim to maximize within each subset after the data is split? A. Homogeneity within subsets B. Data variability C. Diversity among samples D. Entropy across nodes Answer: Homogeneity within subsets (A) CART aims to maximize homogeneity within each subset to improve classification or regression accuracy. 8. Which algorithm is not suitable for datasets with continuous variables? A. C4.5 B. ID3 C. CHAID D. C5.0 Answer: ID3 (B) ID3 primarily focuses on classification tasks and is not suitable for continuous variables. 9. What is a notable feature of the C5.0 algorithm compared to C4.5? A. It is less efficient in training B. It requires categorical attributes only C. It is significantly faster in training time D. It handles missing values Answer: It is significantly faster in training time (C) C5.0 provides increased efficiency in training time compared to C4.5. 10. Which decision tree algorithm uses gain ratio to better select the best split at a node? A. C4.5 B. QUEST C. ID3 D. CART Answer: C4.5 (A) C4.5 uses gain ratio, which improves upon information gain, to determine the best split. 11. Which algorithm employs statistical inference to construct decision trees? A. C5.0 B. CHAID C. Conditional Inference Trees (CIT) D. C4.5 Answer: Conditional Inference Trees (CIT) (C) Conditional Inference Trees (CIT) uses statistical inference to ensure reliable splits in decision trees. 12. What is the purpose of using Gini impurity or entropy in CART? A. To handle missing values B. To measure the efficiency of data storage C. To determine the best split at a node D. To optimize training speed Answer: To determine the best split at a node (C) CART uses Gini impurity or entropy to measure the impurity of a node and determine the best split. 13. What characteristic distinguishes QUEST from other decision tree algorithms? A. It uses a basic approach for splitting nodes B. It performs poorly on large datasets C. It primarily focuses on continuous variables D. It requires only categorical data Answer: It primarily focuses on continuous variables (C) QUEST focuses on continuous variables and is designed to perform well with large datasets.