Fisher Kernel Discriminant Analysis Under Indefinite Kernel Function

Abstract

In this research, Fisher's Indefinite discriminant analysis method was used on data representing three groups, the first with size (135), the second with size (120), and the third with size (55) by people with hepatitis A type, B, and c, which is the most common type of hepatitis. The virus was common, and the infection of the three diseases was considered as a response variable, and the value was given (1) for the first type, (2) for the second type, and (3) for the third type, and the independent variables that were relied upon in agreement with specialized doctors are (7) independent variable, gender, age, blood sugar level, mother's infection with viral hepatitis, previous blood transfusions or surgeries, injections of medicines and drugs, and the percentage of white blood cells in the blood. It was concluded that Fisher's unbounded discriminant analysis method has an efficient performance because it recorded the lowest rate of classification error, and it was half the probability of error.