Why do the Naïve Bayesian classifier implementations use the log of probability value rather than the pure probability value?

Why do the Naïve Bayesian classifier implementations use the log of probability value rather than the pure probability value?
A . To ensure the conditional independence of attribute values
B . To avoid numerical underflow errors in high dimensional problems
C . To obtain a more accurate estimate of the probabilities without the need for a Laplace smoothing
D . To invalidate the variables that are continuous

View Answer

Answer: B