Introduction to Naive Bayes Algorithm Answers 1. What is a common disadvantage of the Naive Bayes algorithm? A. It is not suitable for text classification tasks. B. It requires feature scaling before application. C. It performs well with skewed datasets. D. It assumes independence among features. Answer: It assumes independence among features. (D) The assumption of feature independence is often violated in real-world data, affecting accuracy. 2. Which of the following applications is NOT typically associated with Naive Bayes? A. Image classification B. Spam filtering C. Logistic Regression D. Fraud detection Answer: Logistic Regression (C) Logistic Regression is a separate algorithm and not an application of Naive Bayes. 3. What metric is NOT commonly used to evaluate model performance? A. Probability score B. AUC C. F1-score D. Precision Answer: Probability score (A) Probability score is not a standard metric for evaluating model performance, unlike AUC, F1-score, and Precision. 4. What is a suitable strategy to handle missing values in a dataset used for Naive Bayes? A. Delete all rows with missing data. B. Scale the feature values before imputing. C. Ignore features that contain missing values. D. Impute values using Random Forest. Answer: Ignore features that contain missing values. (C) Ignoring features with missing values is one common strategy, while scaling is not required for Naive Bayes. 5. Which statement about feature scaling in Naive Bayes is correct? A. Feature scaling is not required for the algorithm. B. Feature scaling affects the calculation of probabilities. C. Features must be normalized for proper functioning. D. Feature scaling is crucial for accuracy. Answer: Feature scaling is not required for the algorithm. (A) Feature scaling is not necessary for Naive Bayes since probabilities are calculated independently for each feature. 6. What is the primary assumption made by the Naive Bayes algorithm regarding features? A. Features are conditionally independent given the class label. B. Features are discrete and binary only. C. Features have equal probabilities. D. Features are correlated with each other. Answer: Features are conditionally independent given the class label. (A) The Naive Bayes algorithm assumes that the features are conditionally independent, which simplifies its calculations. 7. In Bayes' Theorem, what does P(A|B) represent? A. Prior probability of event B. B. Prior probability of event A. C. Posterior probability of event A given event B. D. Likelihood of event B given event A. Answer: Posterior probability of event A given event B. (C) P(A|B) is the posterior probability of event A given event B. 8. Which type of Naive Bayes is most appropriate for discrete features like word counts? A. Gaussian Naive Bayes. B. Bernoulli Naive Bayes. C. Categorical Naive Bayes. D. Multinomial Naive Bayes. Answer: Multinomial Naive Bayes. (D) Multinomial Naive Bayes is suitable for modeling discrete features, such as word counts in text classification. 9. What does the Naive Bayes algorithm predict? A. The class with the least amount of features. B. The class with the lowest prior probability. C. The class label that maximizes the posterior probability. D. The average of the feature probabilities. Answer: The class label that maximizes the posterior probability. (C) Naive Bayes predicts the class label that maximizes the posterior probability given the features. 10. Which of the following is a characteristic of Gaussian Naive Bayes? A. It models binary features only. B. It assumes features follow a normal distribution within each class. C. It is best for continuous time series data. D. It requires large amounts of labeled data for training. Answer: It assumes features follow a normal distribution within each class. (B) Gaussian Naive Bayes is used for continuous features and assumes a normal distribution of those features within each class. 11. What is one key advantage of using the Naive Bayes algorithm? A. It can only handle low-dimensional data. B. It requires extensive computational resources. C. It is computationally efficient, especially for large datasets. D. It is always the most accurate classification method available. Answer: It is computationally efficient, especially for large datasets. (C) Naive Bayes is computationally efficient, making it suitable for large datasets and high-dimensional data. 12. How does the Naive Bayes algorithm handle feature probabilities? A. It calculates likelihoods by multiplying individual feature probabilities. B. It assumes equal feature probabilities for prediction. C. It utilizes the joint probability of all features. D. It disregards feature probabilities entirely. Answer: It calculates likelihoods by multiplying individual feature probabilities. (A) The algorithm calculates the likelihoods for each class by multiplying the individual feature probabilities. 13. Which component is NOT part of Bayes' Theorem? A. P(B|A) B. P(A) C. P(A|B) D. P(C|A) Answer: P(C|A) (D) P(C|A) is not part of the components specified in Bayes' Theorem, which includes P(A|B), P(B|A), P(A), and P(B).