Please use this identifier to cite or link to this item:
Weihs, Claus
Luebke, Karsten
Year of Publication: 
Series/Report no.: 
Technical Report 2005,41
Linear Discriminant Analysis (LDA) performs well for classifica- tion of business phases – even though the premises of an LDA are not met. As the variables are highly correlated there are numerical as well as interpretational shortcomings. By transforming the classification problem to a regression setting both problems can be addressed by a computer-intensive prediction oriented method which also improves the classification performance.
Document Type: 
Working Paper

Files in This Item:
111.76 kB

Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.