Using Fisher idea to transform the original dataset to a one-dimensional datase
Design two different classifiers based on Fisher idea:
One technique for threshold selection is to use the average of the transformed means of the classes, i.e. wTm1 + wTm2=2 where m1 and m2 is the means of the classes in the original spaceand w is the Fisher direction
another thereshold can be p1μ1 + p2μ2 where p1 and p2 are the probabilities of different classes