Data Mining-Bayes theorem

Bayesian theorem is a kind of classification method of statistics

The simplest Bayesian classification method is called naive Bayesian classification method

An important condition of naive Bayesian method is that the effect of one attribute value on classification is independent of other attribute values, also known as class conditional independence

P (h| X) =p (x| h) P (h)/P (X) which has been known to seek the former. That is, a posteriori = likelihood x priori/evidence factor

Naive Bayesian method is easy to realize, and it can obtain better classification accuracy in most cases. Its disadvantage lies in its conditional independence hypothesis that if there are strong dependencies between the various attributes of the data, no good results will be achieved.

How do you handle dependency between attributes? Bayesian belief Network is introduced

Bayesian network is a direction-free graph, the node in the graph represents a random variable, which can correspond to an attribute in the actual data. The edges between nodes represent a direct dependency relationship between variables.

Bayesian Network Learning-----------

The variables in the Bayesian network can be observed or hidden in all or some of the trained tuples. The case of hidden data is also known as missing or incomplete data.

Data Mining-Bayes theorem