朴素贝叶斯分类器

Given the Y as the output such as classified labels， and the X as the input feature vector consisted of a series of properties, the Bayesian classifier proposes a rule that a map from the $X$ to $Y$.

$$Y =[y_1 , y_2 ,\cdots ,y_k ],X=[x_1,x_2,\cdots,x_n]$$
which $y_k$ represents a label and $x_i$ denotes a property of input data.

Bayesian theorem is defined as following:

$$P(Y \lvert X) = \frac {P(X \lvert Y) * P(Y)} {P(X)} \ =\frac{P(X \lvert Y) * P(Y)} {\sum _{i=0}^{i=k}P(X \lvert Y=y_i)*P(y_i)}$$

Naive Bayesian theorem that assumes the elements of $X$ are conditional independence is defined as following:

$$P(X\lvert Y) = P(x_1,x_2,\cdots ,x_n \lvert Y) \\ =\prod_{i=1}^{n}P(x_i,\lvert Y)$$
Combined the two formulas above, the Naive Bayesian Classifier can be applied to classify task by using the big data.