The following article describes an algorithm for constructing a decision tree classifier. It is a modified version of an algorithm in which one class was separated in every node of the tree. Now classes are divided into groups in every node. The proposed splitting method is based on Fisher criterion. The results of experiments on two real datasets are presented in comparison to the previous version of the algorithm.
KeywordsLinear Discriminant Analysis Decision Tree Classifier Fisher Criterion Sequential Classification Good Split
Unable to display preview. Download preview PDF.
- W.Malina: Two-Parameter Fisher Criterion. IEEE Trans. Systems, Man and Cybernetics, Vol. 31, No. 4, August 2001.Google Scholar
- A. Kolakowska, W. Malina: Fisher sequential classifier. 5`h Conference Neural Networks and Soft Computing, Zakopane 6–10 czerwca 2000, str.227–232.Google Scholar
- D. Mitchie, D. J. Spiegelhalter, C. C. Taylor: Machine Learning, Neural and Statistical Classification. 1994.Google Scholar
- Safavian S.R., Landgrebe D.: A Survey of Decision Tree Classifier Methodology. IEEE Trans. Systems, Man, and Cybernet., Vol. 21, No. 3, May/June 1991.Google Scholar