Nave Bayesian Classifiers DATA-SCIENCE-FOUNDATIONS Practice Questions (Page 5)

Questions 21–25

1. 21application · medium

   A data scientist is training a Bernoulli Naive Bayes model to classify emails as spam or not. In the training set, 40% of emails are spam. Among spam emails, the word 'discount' appears in 60% of them. Among non-spam emails, it appears in 10%. With Laplace smoothing (alpha=1) and a vocabulary size of 1000, what is the smoothed probability P('discount' appears | spam)?

   • A0.601
   • B0.600
   • CCannot be determined from the given information
   • D0.500
   Reveal Solution Discussion Select an answer first
2. 22expert · medium

   A data scientist is comparing Naive Bayes and logistic regression for a binary classification problem with 20 features. The features are known to be highly correlated. The dataset has 5,000 samples. The scientist observes that Naive Bayes has lower accuracy than logistic regression. Which explanation is most likely?

   • ALogistic regression is always more accurate than Naive Bayes.
   • BNaive Bayes cannot handle continuous features.
   • CLogistic regression uses a different optimization algorithm that is superior.
   • DNaive Bayes assumes conditional independence, which is violated by correlated features, leading to overconfident and often incorrect probability estimates.
   Reveal Solution Discussion Select an answer first
4. 23application · easy

   A team is building a classifier to predict whether a person has a certain medical condition based on continuous measurements like blood pressure and cholesterol level. They want to use Naive Bayes. Which variant should they choose?

   • AGaussian Naive Bayes
   • BBernoulli Naive Bayes
   • CComplement Naive Bayes
   • DMultinomial Naive Bayes
   Reveal Solution Discussion Select an answer first
5. 24application · medium

   A Naive Bayes model classifies loan applications as approved or rejected. For a particular applicant, the model outputs P(approved) = 0.45 and P(rejected) = 0.55. The applicant is rejected. Which statement best describes the interpretation?

   • AThe applicant has a 45% chance of being approved, so the model is uncertain and should not make a decision.
   • BThe applicant is definitely rejected because the probability is above 0.5.
   • CThe model is 55% accurate in its predictions.
   • DThe model estimates a 55% probability that the applicant should be rejected, so rejection is the more likely class, but the estimate is not certain.
   Reveal Solution Discussion Select an answer first
6. 25foundation · easy

   In Naive Bayes, what does a high posterior probability for a class indicate?

   • AThe model has been trained for many epochs
   • BThe model is confident that the instance belongs to that class
   • CThe instance has many features
   • DThe class has a low prior probability
   Reveal Solution Discussion Select an answer first