Feature Selection Using Non Linear Feature Relation Index
In this paper we propose a dependence measure for a pair of features. This measure aims at identifying redundant features where the relationship between the features is characterized by higher degree polynomials. An algorithm is also proposed to make effective use of this dependence measure for the feature selection. Neither the calculation of dependence measure, nor the algorithm need the class values of the observations. So they can be used for clustering as well as classification.
KeywordsFeature Selection Dependence Measure Feature Subset Feature Selection Method Feature Selection Algorithm
- 1.Liu, H., Motoda, H.: Computational Methods of Feature Selection. Chapman Hall/CRCGoogle Scholar
- 2.Dash, M., Liu, H.: Feature Selection for Clustering. In: Proceedings of the 4th Pacific-Asia Conference on Knowledge Discovery and Data Mining, Current Issues and New Applications, April 18-20, 2000, pp. 110–121 (2000)Google Scholar
- 8.Asuncion, A., Newman, D.J.: UCI Machine Learning Repository, http://www.ics.uci.edu/~mlearn/MLRepository.html
- 9.Pal, S.K., Mitra, P.: Patter Recognition Algorithms for Data Mining. ChapMan and Hall/CRCGoogle Scholar