Neural network and discriminant methods for classification with a case study

Predictor variables based classification and organisation of objects into t wo or more classes are fundamental to all branches of biosciences. Traditio nally, discriminant analysis has been used for classification. In the prese nt paper, we have made a comparative study of the three methods viz. discri minant analysis, logistic regression model and neural network for classific ation. A case study where 66 patients of meningitis were clinically evaluat ed for three months for deciding whether to discharge them or to continue t heir treatment, has been presented. The neural network method has been foun d to be superior to others.