Fundamentals of Statistics contains material of various lectures and courses of H. Lohninger on statistics, data analysis and here for more.

Exercise - Linear Discriminant Analysis

Suppose you are the manager of a bank and you have the problem of discriminating between genuine and counterfeit banknotes. On the left you see a banknote worth 1000 Austrian schillings (about 72.67 Euro). You are measuring several distances on the banknote and the width and height of it. Measuring these values of about 100 genuine and 100 counterfeit banknotes, you will finally end up with a table of 200 observations with several variables.

Use the data set FLURIEDW to set up a model which is based on linear discriminant analysis and is capable of discriminating between genuine and counterfeit money (this data set contains data on Swiss francs; it has been obtained courtesy of H. Riedwyl [Flury 1983]). Use  DataLab  to calculate the model.

Use the calculated model to find out which of the following bank notes are counterfeit:

        Length  Left    Right   Bottom   Top   Diagonal
BN1     215.1   130.0   129.8    9.1    10.2    141.5
BN2     214.7   130.7   130.8   11.2    11.2    139.4
BN3     214.3   129.9   129.9   10.2    11.5    139.6
BN4     214.7   130.0   129.4    7.8    10.0    141.2

You may directly load the data of these four banknotes into  DataLab  and apply the stored discriminating model.

Last Update: 2012-10-08