Rule-Based Estimation of Attribute Relevance
We consider estimation of relevance of attributes used for classification. This estimation takes into account the predictive capabilities of the attributes. To this end, we are using Bayesian confirmation measure. The estimation is based on analysis of rule classifiers in classification tests. The attribute relevance measure increases when more rules involving this attribute suggest a correct decision, or when more rules that do not involve this attribute suggest an incorrect decision in the classification test; otherwise, the attribute relevance measure is decreasing. This requirement is satisfied by a monotonic Bayesian confirmation measure. Usefulness of the presented measure is verified experimentally.
Keywordsattribute relevance Bayesian confirmation decision rule classification ensemble classifier
Unable to display preview. Download preview PDF.
- 1.Asuncion, A., Newman, D.: UCI machine learning repository (2007), http://www.ics.uci.edu/~mlearn/MLRepository.html
- 2.Błaszczyński, J., Greco, S., Słowiński, R.: Ordinal and non-ordinal classification using monotonic rules. In: 8th International Conference of Modeling and Simulation, MOSIM 2010 (May 2010)Google Scholar
- 3.Błaszczyński, J., Greco, S., Słowiński, R.: Inductive discovery of laws using monotonic rules. Engineering Applications of Artificial Intelligence (to appear)Google Scholar
- 4.Błaszczyński, J., Słowiński, R., Stefanowski, J.: Feature set-based consistency sampling in bagging ensembles. In: From Local Patterns To Global Models (LEGO), ECML/PKDD Workshop, pp. 19–35 (2009)Google Scholar
- 6.Błaszczyński, J., Słowiński, R., Szeląg, M.: Sequential covering rule induction algorithm for variable consistency rough set approaches. Information Sciences 181(5), 987–1002 (2011)Google Scholar
- 9.Greco, S., Słowiński, R., Pawlak, Z.: Can Bayesian confirmation measures be useful for rough set decision rules? Engineering Applications of Artificial Intelligence 17(4), 345–361 (2004)Google Scholar
- 11.Greco, S., Słowiński, R., Szczęch, I.: Properties of rule interestingness measures and alternative approaches to normalization of measures. IEEE Transactions on Knowledge and Data Engineering (to appear)Google Scholar
- 15.Robnik-Šikonja, M., Kononenko, I.: Explaining classifications for individual instances. IEEE Trans. on Knowl. and Data Eng. 20, 589–600 (2008)Google Scholar
- 16.Strumbelj, E., Kononenko, I.: An efficient explanation of individual classifications using game theory. J. Mach. Learn. Res. 11, 1–18 (2010)Google Scholar