Nave Bayesian Classifiers DATA-SCIENCE-FOUNDATIONS Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   What is the key assumption of the Naive Bayes classifier?

   • AFeatures are mutually exclusive and exhaustive
   • BFeatures are normally distributed within each class
   • CFeatures are conditionally independent given the class label
   • DThe class prior probabilities are always equal
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   A Naive Bayes model classifies a patient as having a disease (D) or not (N) based on symptoms A and B. For a new patient, the model outputs P(D|A,B) = 0.85 and P(N|A,B) = 0.15. Which interpretation is correct?

   • AThe model is 85% accurate overall.
   • BThe patient definitely has the disease because the probability is above 0.5.
   • CThe model predicts the patient has the disease, with an estimated 85% probability, but this is only as reliable as the training data and assumptions.
   • DThe symptoms A and B are independent of each other.
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A team is using Naive Bayes to classify customer reviews as positive or negative. They notice the model performs poorly on reviews that contain both 'not good' and 'great' in the same sentence. Which property of Naive Bayes most directly explains this difficulty?

   • AThe model uses a Gaussian distribution for word counts, which is inappropriate for discrete text features.
   • BThe model cannot handle more than two classes, so it fails when reviews have mixed sentiment.
   • CThe conditional independence assumption treats 'not' and 'good' as independent, so it cannot capture the negation that flips sentiment.
   • DThe model assigns zero probability to unseen word combinations, causing the entire review to be misclassified.
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A Naive Bayes classifier is built to predict customer churn. The confusion matrix on a test set of 200 customers is: 120 true negatives, 30 false positives, 20 false negatives, and 30 true positives. What is the recall of the model?

   • A0.50
   • B0.60
   • C0.75
   • D0.15
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A data scientist is training a Multinomial Naive Bayes model for text classification. In the training set, the word 'warranty' appears 0 times in the 'electronics' class and 5 times in the 'other' class. Without smoothing, what would be the impact on classifying a new document that contains 'warranty'?

   • AThe model would assign a very small but non-zero probability to the 'electronics' class.
   • BThe model would still classify correctly because other words in the document provide evidence.
   • CThe posterior probability for the 'electronics' class would be zero, so the document would be classified as 'other' regardless of other evidence.
   • DThe model would ignore the word 'warranty' because it is unseen in that class.
   Reveal Solution Discussion Select an answer first