Software system for different types of data classification based on the ensemble algorithms

This article describes the structure and functional content of a developed software system of different types of data classification based on the ensemble algorithms. Also the article describes the developed heterogeneous ensemble classification algorithm implemented in the software system. A distinctive feature of the algorithm is the iterative use of single (basic) classifiers on the initial training sample and inclusion in the ensemble only those classifiers whose relative error does not exceed a predetermined threshold. The software system was tested on real medical data. The accuracy of data classification using the basic classifiers and heterogeneous ensemble algorithm was compared. Test results showed the effectiveness (increase classification accuracy) of heterogeneous ensemble algorithm compared with the single classifiers.