Introduction

Linear Discriminant Analysis (LDA) [1,2] is a supervised technique used for classification. Running this method allows the user to make a classification rule based Fisher's Sample Linear Discriminants, [1], on a training data set. That is a data set where the classes are known a priori.

This classification rule can further be used on new data sets to predict their class. This can be done using the LinearDiscrimPrediction node.

The motivation behind the Fisher linear discriminant analysis is the need to obtain a reasonable representation of the populations that only involves a few linear combinations of the observations. This helps to separate the populations.



Bjørn Kåre Alsberg 2006-04-06