Advertisement

Determining attribute relevance in decision trees

  • Mirsad Hadzikadic
  • Ben F. Bohren
Communications Session 6B Learning and Discovery Systems
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1325)

Abstract

Concept formation, an artificial intelligence classification technique, has been used successfully by many researchers in predicting outcomes of new objects based on a decision tree built from previously seen objects. All systems based on concept formation are capable of providing outcome predictions. INC2.5, a concept formation system, goes further by (a) implementing an algorithm that identifies relevant attributes and (b) administering a test that measures system's predictive ability based on the reduced attribute set. These capabilities are important to users attempting to prove a specific feature's contribution to an outcome. This paper focuses on the algorithm for analyzing attribute relevance as opposed to the classification and prediction techniques that have been explained in previous publications.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Bohren, B.F. and Hadzikadic, M. (1994), Turning Medical Data into Decision-Support Knowledge. Proceedings of the 18th SCAMC, 735–739.Google Scholar
  2. 2.
    Fisher, D. H. Knowledge Acquisition Via Incremental Conceptual Clustering. In Machine Learning, 2, 2 (1987) 139–172.Google Scholar
  3. 3.
    Hadzikadic, M., Automated Design of Diagnostic Systems. Artificial Intelligence in Medicine Journal, 4 (1992a) 329–342.Google Scholar
  4. 4.
    Hadzikadic, M. Prediction Performance as a Function of the Representation Language in Concept Formation Systems. Proceedings of the Fourteenth Annual Conference of the Cognitive Science Society, 850–854, Bloomington, Indiana, July 29–August 1, 1992b.Google Scholar
  5. 5.
    Hadzikadic, M., Bohren, B.F., “Learning to Predict: INC2.5,” IEEE Transactions on Knowledge and Data Engineering. 9, 1 (1997) 168–173.Google Scholar
  6. 6.
    Hadzikadic, M., Bohren, B., Hakenewerth, A., Norton, J., Mehta, B., Andrews, C. “Concept Formation vs. Logistic Regression: Predicting Death in Trauma Patients.” Artificial Intelligence in Medicine Journal 8 (1996) 493–504.Google Scholar
  7. 7.
    Hanson, S. J. and Bauer, M. Conceptual Clustering, Categorization, and Polymorphy. In Machine Learning, 3, 4 (1989) 343–372.Google Scholar
  8. 8.
    Kolodner, J. L. Retrieval and Organizational Strategies in Conceptual Memory: A Computer Model, Lawrence Erlbaum Associated, Publishers, London, 1984.Google Scholar
  9. 9.
    Lebowitz, M. Experiments with Incremental Concept Formation: UNIMEM. In Machine Learning, 2, 2 (1987) 103–138.Google Scholar
  10. 10.
    Michalski, R.S., and Stepp, R.E. Learning From Observation: Conceptual Clustering. In Machine Learning: An Artificial Intelligence Approach, R.S. Michalski, J.G. Carbonell, and T. M. Mitchell (eds.), Morgan Kaufinann Publishers, Inc., Lao Altos, CA, 1983.Google Scholar
  11. 11.
    Tversky, A. Features of Similarity. Psychological Review, 84 (1977) 327–352.Google Scholar
  12. 12.
    Mitchell, T. M. Machine Learning. McGraw-Hill, 1997.Google Scholar
  13. 13.
    Quinlan, J.R. C4.5: Programs for Machine Learning. San Mateo, CA: Morgan Kauffmann, 1993.Google Scholar
  14. 14.
    Mingers, J. An Empirical Comparison of Selection Measures for Decision-tree induction. Machine Learning, 3 (4) 319–342, 1989.Google Scholar
  15. 15.
    Buntine, W. and Niblett, T. A Further Comparison of Splitting Rules for Decisiontree Induction. Machine Learning, 8, 75–86, 1992.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1997

Authors and Affiliations

  • Mirsad Hadzikadic
    • 1
  • Ben F. Bohren
    • 2
  1. 1.Carolinas HealthCare System and University of North Carolina at CharlotteUSA
  2. 2.Speech Systems Inc.USA

Personalised recommendations