Introduction to Naive Bayes Classifiers Answers 1. What is the main principle behind the Naive Bayes classifier? A. To maximize the likelihood of the input data without any assumptions. B. To calculate posterior probabilities using Bayes' theorem. C. To perform classifications based solely on the training set. D. To minimize the number of features used in classification. Answer: To calculate posterior probabilities using Bayes' theorem. (B) The Naive Bayes classifier calculates posterior probabilities using Bayes' theorem. 2. Which of the following is NOT an advantage of Naive Bayes classifiers? A. Works well with high-dimensional data. B. Small memory requirements. C. Ensures complete accuracy in predictions. D. Simplicity and ease of implementation. Answer: Ensures complete accuracy in predictions. (C) Naive Bayes classifiers do not guarantee complete accuracy due to their assumptions. 3. In which scenario might the performance of a Naive Bayes classifier degrade? A. With datasets containing many irrelevant features. B. When applied to well-structured datasets. C. When features are mutually independent. D. In scenarios with limited training data. Answer: With datasets containing many irrelevant features. (A) Naive Bayes classifiers are sensitive to irrelevant features, which can degrade performance. 4. When dealing with feature dependence, what is one proposed solution? A. Only focus on datasets with fewer attributes. B. Use sophisticated models that consider feature dependence. C. Incorporate deep learning models exclusively. D. Avoid any form of feature engineering. Answer: Use sophisticated models that consider feature dependence. (B) Using more sophisticated models can help account for feature dependence in Naive Bayes. 5. Which evaluation metric is commonly used to assess the performance of a Naive Bayes classifier? A. F1-score. B. Training speed index. C. Randomness score. D. Memory efficiency ratio. Answer: F1-score. (A) The F1-score is used to measure the performance of classifiers like Naive Bayes. 6. What is the main assumption of Naive Bayes classifiers? A. The classes are normally distributed. B. Features are independent given the class label. C. All features are correlated given the class label. D. Each feature has an equal probability of occurring. Answer: Features are independent given the class label. (B) The naive assumption is that features are independent given the class label, simplifying calculations. 7. Which type of Naive Bayes classifier is appropriate for continuous features? A. Categorical Naive Bayes B. Multinomial Naive Bayes C. Gaussian Naive Bayes D. Bernoulli Naive Bayes Answer: Gaussian Naive Bayes (C) Gaussian Naive Bayes is used for continuous features and assumes a normal distribution. 8. In Bayes' theorem, what does P(A|B) represent? A. The prior probability of event A. B. The likelihood of event B given event A. C. The posterior probability of event A given event B. D. The prior probability of event B. Answer: The posterior probability of event A given event B. (C) P(A|B) is the posterior probability, which measures the probability of A after considering B. 9. What type of Naive Bayes classifier is best suited for text classification tasks? A. Bernoulli Naive Bayes B. Gaussian Naive Bayes C. Multinomial Naive Bayes D. Normal Naive Bayes Answer: Multinomial Naive Bayes (C) Multinomial Naive Bayes is particularly effective for text classification involving word counts. 10. What is estimated during the training process of Naive Bayes classifiers for each class? A. Only the prior probabilities. B. The variance only. C. Both prior and likelihood probabilities. D. Only the likelihood probabilities. Answer: Both prior and likelihood probabilities. (C) Both prior (P(class)) and likelihood (P(feature|class)) probabilities are estimated during training. 11. Which of the following best describes Bernoulli Naive Bayes? A. It is designed for multinomial distributions. B. It models features that follow a Gaussian distribution. C. It is used for binary features. D. It assumes independence of all features without labeling. Answer: It is used for binary features. (C) Bernoulli Naive Bayes is specifically designed for binary features, modeling their presence or absence. 12. What is usually calculated for continuous data during the training of Naive Bayes classifiers? A. Mean and variance. B. Minimum and maximum values. C. Mode and frequency. D. Count and total occurrences. Answer: Mean and variance. (A) For continuous data, the mean and variance are estimated from the training data. 13. What is the significance of the assumption of independence in Naive Bayes classifiers? A. It allows for a higher error rate. B. It ensures equal distribution of features. C. It models features as mutually exclusive. D. It reduces complexity and simplifies calculations. Answer: It reduces complexity and simplifies calculations. (D) The assumption of independence greatly simplifies computations in Naive Bayes classifiers.