Nave Bayesian Classifiers DATA-SCIENCE-FOUNDATIONS Practice Questions (Page 3)

Questions 11–15

1. 11expert · medium

   A company is deploying a Naive Bayes model for real-time fraud detection. The model has high precision but low recall. The business requirement is to minimize the number of fraudulent transactions that go undetected, while keeping the number of false alarms manageable. Which action is most appropriate?

   • ASwitch to a different model that always has higher recall.
   • BUse a confusion matrix to report accuracy instead of recall.
   • CRaise the classification threshold so that only high-confidence predictions are flagged.
   • DLower the classification threshold so that more transactions are flagged as fraudulent.
   Reveal Solution Discussion Select an answer first
2. 12foundation · medium

   In a Naive Bayes classifier, how is the class label assigned to a new unlabeled instance?

   • ABy computing the joint probability of the instance and each class and selecting the class with the highest joint probability.
   • BBy computing the likelihood of the instance under each class and selecting the class with the highest likelihood.
   • CBy computing the prior probability of each class and selecting the class with the highest prior.
   • DBy computing the posterior probability for each class and selecting the class with the highest probability.
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A company wants to classify short product reviews as positive or negative based on the presence or absence of a set of predefined keywords (e.g., 'excellent', 'poor', 'refund'). Each review is represented as a binary vector indicating which keywords appear. Which variant of Naive Bayes is most appropriate?

   • AComplement Naive Bayes
   • BMultinomial Naive Bayes
   • CBernoulli Naive Bayes
   • DGaussian Naive Bayes
   Reveal Solution Discussion Select an answer first
5. 14application · medium

   A Naive Bayes model is trained on a small dataset of emails. In the training data, the word 'urgent' never appears in any non-spam email. When classifying a new email that contains 'urgent', the model assigns a posterior probability of zero for the non-spam class. Which technique should be applied to avoid this zero-probability problem?

   • ADimensionality reduction
   • BLaplace smoothing
   • CFeature scaling
   • DIncreasing the training set size
   Reveal Solution Discussion Select an answer first
6. 15application · medium

   A team is building a Naive Bayes classifier to categorize news articles into topics (sports, politics, technology). Each article is represented as a vector of word counts. Which variant of Naive Bayes is most appropriate?

   • ABernoulli Naive Bayes
   • BGaussian Naive Bayes
   • CComplement Naive Bayes
   • DMultinomial Naive Bayes
   Reveal Solution Discussion Select an answer first