FUZZY ENSEMBLE OF DECISION TREES FOR THE COMPUTER SYSTEMS STATE IDENTIFICATION

Abstract

The object of the research is the process of identifying the state of computer systems. The subject of the research is the methods of fuzzy ensemble decision trees with multidimensional decision nodes for identifying the state of computer systems. The goal of the research is to develop and evaluate the effectiveness of a fuzzy ensemble of decision trees to improve the accuracy of identifying computer system states under conditions of uncertainty, noise, and incomplete data. Methods used: machine learning methods, data preprocessing techniques, ensemble classifiers, stacking approaches, methods for feature selection and combination of computer system attributes. Results obtained: the effectiveness of both classical and newly developed methods for identifying the state of computer systems under complex conditions, including data imbalance and the presence of anomalous states, was investigated. A comprehensive approach using Fuzzy Stacking with MDT was proposed, providing high accuracy and stability of classification. The best results were achieved with the stacking approach, which combines base classifiers and fuzzy decision trees, minimizing both type I and type II errors and achieving high generalization ability (MCC, F1-score, TS, LN(DOR)). Conclusions. Based on the results of the study, an improved approach for identifying the state of computer systems is proposed, which combines the stacking method with Fuzzy MDT and feature selection optimization. The integrated use of these methods significantly enhances classification accuracy, result stability, and model robustness to data imbalance, while ensuring high-quality classification even in the presence of new anomalous states.