Abstract

The article is dedicated to the area of an ensemble of classifiers, in particular, the issues related to the rough set theory were used to define the base classifiers. The new proposed method for defining base classifiers use the method of reducts search executed by a genetic algorithm. This algorithm allows to define the number of reducts that will be used. Based on selected reducts sub-tables are defined. For each sub-table a modified k-nearest neighbors algorithm is used and the decision vector is determined. The majority voting method is used to fuse decision vectors. Experimental results showed that the proposed approach, in most cases, gives better results than other well-known ensembles of classifiers. Moreover, it was noticed that increasing the number of base classifiers usually improves classification accuracy, but only to a certain level.

Recommended Citation

Przybyła-Kasperek, M. (2021). Ensemble of Classifiers Based on Genetic Reducts and K-Nearest Neighbors Classifier for Data with Non Missing Values. In E. Insfran, F. González, S. Abrahão, M. Fernández, C. Barry, H. Linger, M. Lang, & C. Schneider (Eds.), Information Systems Development: Crossing Boundaries between Development and Operations (DevOps) in Information Systems (ISD2021 Proceedings). Valencia, Spain: Universitat Politècnica de València.

Paper Type

Full Paper

Share

COinS
 

Ensemble of Classifiers Based on Genetic Reducts and K-Nearest Neighbors Classifier for Data with Non Missing Values

The article is dedicated to the area of an ensemble of classifiers, in particular, the issues related to the rough set theory were used to define the base classifiers. The new proposed method for defining base classifiers use the method of reducts search executed by a genetic algorithm. This algorithm allows to define the number of reducts that will be used. Based on selected reducts sub-tables are defined. For each sub-table a modified k-nearest neighbors algorithm is used and the decision vector is determined. The majority voting method is used to fuse decision vectors. Experimental results showed that the proposed approach, in most cases, gives better results than other well-known ensembles of classifiers. Moreover, it was noticed that increasing the number of base classifiers usually improves classification accuracy, but only to a certain level.