Bayes Flashcards
What is naive bayes?
opposite to 1R uses all attributes, and it makes a strong assumption that all attributes are independent of each other. It allows us to calculate the probability of the data just by using multiplication of probabilities
What is priori?
probability of an event BEFORE evidence has been seen
What is posteriori?
probability of an even AFTER evidence has been seen
What is bayes theorem?
if you know a conditional probability, then you can calculate the opposite probability just by using the same kind of calculation
e.g. P(H | E) = P(E | H) . P(A)
/P(B)
E: Evidence
H: Hypothesis
Why does naive bayes make a strong assumption?
To make things easier to calculate
What is the strong assumption naive bayes makes?
That all attributes are independent of each other, meaning that if we know info about one attributes we assume that it doesn’t give any further information about other attributes
What is a conditional probabilities?
If we know the conditional probability then we can calculate the opposite of it using bayes rule
What is joint probabilities?
Multiple events like P(A and B), the two events being A and B.
How do we use Naive Bayes for classification?
Whats the probability of the class given an instance, to apply bayes rule we need to estimate the P(E | H) probability of the evidence given the hypothesis
What is the naive assumption made for classification?
We assume that attributes are independent when considering a particular class value H
What is the “zero frequency problem”?
when an attribute doesnt occur in every class, which results in the probability being zero
How to fix the “zero frequency problem”?
add 1 to the count for every attribute value-class combination
