Knowledge aggregation in decision-making process with C-fuzzy random forest using OWA operators.

The idea of knowledge aggregation contained in C-fuzzy decision tree nodes with OWA operators during the C-fuzzy random forest decision-making process is presented in this paper. C-fuzzy random forest is a new kind of ensemble classifier which consists of C-fuzzy decision trees. There are proposed three kinds of OWA operators for the given problem, called Local OWA, global OWA for each tree in the forest and global OWA for the whole forest. Weights of OWA operators are optimized using a genetic algorithm. In order to evaluate the created classifier, experiments were performed using ten datasets. The classifier was checked in comparison with C4.5 rev. 8 decision tree and single C-fuzzy decision tree. The influence of randomness and proposed OWA operators on the classification accuracy was tested. [ABSTRACT FROM AUTHOR]