Multiple Classifier Systems Based on Interpretable Linear Classifiers
Multiple classifier systems fall into two types: classifier combination systems and classifier choice systems. The former aggregate component systems to produce an overall classification, while the latter choose between component systems to decide which classification rule to use. We illustrate each type applied in a real context where practical constraints limit the type of base classifier which can be used. In particular, our context – that of credit scoring – favours the use of simple interpretable, especially linear, forms. Simple measures of classification performance are just one way of measuring the suitability of classification rules in this context.
Keywordslogistic regression perceptron support vector machines product models
Unable to display preview. Download preview PDF.
- Kelly, M.G., Hand, D.J., Adams, N.M.: Defining the goals to optimise data mining performance. Proceedings of the Fourth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, ed. Agrawal, R., Stolorz, P., Piatetsky-Shapiro, G., Menlo Park: AAAI Press (1998) 234–238Google Scholar
- Kelly, M.G., Hand, D.J.: Credit scoring with uncertain class definitions. IMA Journal of Mathematics Applied in Business and Industry 10 (1998) 331–345Google Scholar
- von Winterfeld, D., Edwards, D.: Costs and payoffs in perceptual research. Psychological Bulletin 91 (1982) 213–217Google Scholar
- Kittler, J., Duin, R.P.W.: Combining classifiers. Proceedings of the International Conference on Pattern Recognition, IEEE (1996) 897–900Google Scholar
- Hand, D.J., Henley, W.E.: Statistical classification methods in consumer credit scoring: a review. Journal of the Royal Statistical Society, Series A 169 (1997) 523–541Google Scholar
- Hand, D.J., Yu, K.: Idiot’s Bayes–not so stupid after all? To appear in International Statistical Review (2001)Google Scholar
- Hand, D.J.: Modelling consumer credit risk. Submitted to IMA Journal of Management Mathematics (2000)Google Scholar
- Hand, D.J., Kelly, M.G.: Superscorecards. Technical Report, Department of Mathematics, Imperial College, London.Google Scholar
- Hand, D.J.: How and why to choose a better scorecard. Technical Report, Department of Mathematics, Imperial College, London.Google Scholar
- Scott, M.J.: PARCEL: feature selection in variable cost domains. Doctoral dissertation, Engineering Department, Cambridge University, UK (1999)Google Scholar
- Provost, F.J., Fawcett, T.: Robust classification systems for imprecise environments. Proceedings of the Fifteenth National Conference on Artificial Intelligence, Madison, WI: AAAI press (1998) 706–713Google Scholar
- Holmes, C.C., Adams, N.M.: A probabilistic nearest neighbor method for statistical pattern recognition. Journal of the Royal Statistical Society, Series, B (2001) in pressGoogle Scholar