Abstract
In many supervised classification problems there is ambiguity about the definitions of the classes. Sometimes many alternative but similar definitions could equally be adopted. We propose taking advantage of this by choosing that particular definition which optimises some additional criterion. In particular, one can choose that definition which leads to greatest predictive classification accuracy, so that any action taken on the basis of the predicted classes is most reliable.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Adams, N. M., Hand, D. J., and Li, H. G.: A simulation study of indirect prognostic Classification. In Proceedings in Computational Statistics, COMPSTAT-98, ed. R. Payne and P. Green, Heidelberg, Physica-Verlag, (1998) 149–154.
Hand, D. J.: Construction and Assessment of Classification Rules. Chichester: Wiley (1997).
Hand, D. J., Henley, W. E.: Statistical Classification methods in consumer credit scoring: a review. Journal of the Royal Statistical Society, Series A, 160 (1997) 523–541.
Hand, D. J., Li, H. G., and Adams, N. M.: Supervised Classification with structured class definitions. Technical Report (1998a).
Hand, D. J., Oliver, J. J. and Lunn, A. D.: Discriminant analysis when the classes arise from a continuum. Pattern Recognition, 31 (1998b), 641–650.
Kelly, M. G. and Hand, D. J.: Credit scoring with uncertain class definitions. Presented at Credit Scoring and Credit Control V, Edinburgh, September (1997).
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1999 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Kelly, M.G., Hand, D.J., Adams, N.M. (1999). Supervised Classification Problems: How to Be Both Judge and Jury. In: Hand, D.J., Kok, J.N., Berthold, M.R. (eds) Advances in Intelligent Data Analysis. IDA 1999. Lecture Notes in Computer Science, vol 1642. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-48412-4_20
Download citation
DOI: https://doi.org/10.1007/3-540-48412-4_20
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-66332-4
Online ISBN: 978-3-540-48412-7
eBook Packages: Springer Book Archive