Naive Bayes how many parameters need to be estimated :
Given a dataset of N points in which each point has d features . Each feature can take V values.
X=<X1,X2…...Xd> and Xi=<V Discrete Values> to be classified into Y = <K Discrete Classes>
P(Y|X=x1,x2,x3...xd) is prop to P(Y).P(X=x1,x2,x3...xd|Y)
P(Y) needs K-1 parameters
P(X=x1,x2,x3...xd|Y) needs dVK – dK parameters => dk(V-1)
In total K-1 + dk(V-1) parameters needed.