Advertisement

On the Behavior of Splitting Criteria for Classification Trees

  • Roberta Siciliano
  • Francesco Mola
Part of the Studies in Classification, Data Analysis, and Knowledge Organization book series (STUDIES CLASS)

Summary

In the framework of classification trees, the behavior of splitting criteria is investigated through a simulation study and applications on a real data set. Some enphasis is appointed to the strength of the dependency among variables, the choice of the splitting rule, the role played by the type of predictors, the stability of the classification rule. Alternative splitting criteria and new splitting rules are also proposed to deal with the computational effort of splitting procedures in large data sets.

Keywords

Classification Tree Terminal Node Gini Index Classification Rule Misclassification Rate 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. Breiman L., Friedman J.H., Olshen R.A., Stone C.S.. (1984): Classification and Regression Trees, Belmont C.A. Wadsworth.MATHGoogle Scholar
  2. Celeux, G. and Lechevallier Y. (1982): Methodes de segmentation non parametriques, Revue de statistique appliquées, 4, 39–53.Google Scholar
  3. Ciampi, A. and Thiffault, J. (1987): Recursive Partition and Amalgamation (RECPAM) for Censored Survival Data: Criteria for tree selector, Statistical Software Newsletter, 2, vol. 14. 78–81.Google Scholar
  4. Hosmer. D. W. and Lemeshow, S. (1990): Applied Logistic Regression, J. Wiley, New York.Google Scholar
  5. lingers, J. (1988): An empirical comparison of selection measures for decision tree induction..tlachine learning, 3. 319–342.Google Scholar
  6. Mola. F. (1993): Aspetti metodologici e computazionali delle tecniche di segmentazione binaria. Un contributo basato su funz.ioni di predizione. PhD dissertation, University of Naples.Google Scholar
  7. Mola, F. and Siciliano, R. (1992): A Two-Stage Predictive Splitting Algorithm in Binary Segmentation, Computational Statistics, Dodge, Y. and Whittaker, J. (eds.), 1, 179–184, (Compstat ‘82 Proceedings), Physica Verlag.Google Scholar
  8. Mola, F. and Siciliano, R. (199–1): Alternative Strategies and CATANOVA Testing in Two-Stage Binary Segmentation. New Approaches in Classification and Data Analysis,Diday. E. et al. (eds.),316–323, Springer Verlag.Google Scholar
  9. Mola. F. and Siciliano. R. (1997a): A Fast Splitting Procedure for Classification Trees, Statistics and Computing (to appear).Google Scholar
  10. Mola, F. and Siciliano, R. (19976): Visualizing Data in Tree-Structured Classification, Proceedings of IFCS-96: Data Science, Classification and Related Methods. (Hayashi, C. et al.,eds.), Springer Verlag, Tokyo.Google Scholar
  11. Mola, F., Klaschka, J. and Siciliano, R. (1996): Logistic Classification Trees, COMPST.4T 96 Proceedings (A. Prat. ed. ), Physica Verlag.Google Scholar
  12. Taylor, P.C. and Silverman, B.W. (1993): Block Diagrams and Splitting Criteria for Classification Trees. Statistics and Computing. 3, 147–161.CrossRefGoogle Scholar

Copyright information

© Springer Japan 1998

Authors and Affiliations

  • Roberta Siciliano
    • 1
  • Francesco Mola
    • 1
  1. 1.Dipartimento di Matematica e StatisticaUniversità di Napoli Federico IINaplesItaly

Personalised recommendations