How is naive Bayesian algorithm understood?
Naive Bayesian algorithm is an algorithm of generative formula
Our goal is to classify the current instance of X as that category, but the resulting formula is the P (ck/x)
In practical problems we usually know that P (Ck) is called a priori probability. We will also know the number of P (x/ck), the probability of this condition
So how do you ask P (ck/x)? The first is to convert the conditional probability distribution to P (ck,x) full distribution/p (x)
The full distribution is then converted to the inverse conditional probability P (Ck) p (x/ck), after which the P (X/CK) is expanded into an independent distribution P (X1*X2*X3...XN/CK)
Then the P (x) is converted to the perfect probability formula P (x) ===sum (P (x/ck) p (CK))
Finally, only the maximum value of P (ck/x) is required, only the k is required. How to ask, maximum likelihood estimation method.
2. In the use of the maximum likelihood estimation method, the probability of 0 can be calculated. What to do, the Laplace smoothing can be done, so that there is no probability of 0 in the case
This article from "Jane Answers Life" blog, declined reprint!
Naive Bayesian algorithm