#discriminant-analysis

Discriminant analysis gives one classification function per class, and a point goes to the class whose function is largest. An analyst first assumes three equally likely classes $C_1$, $C_2$, $C_3$, then learns that $P(C_1) = 2P(C_2) = 2P(C_3)$. After recomputing with these priors, the value of $C_1$'s classification function is larger at every data point.