THEUSE OF CLUSTER ANALYSIS AND THE THEORY OF ARTIFICIAL NEURAL NETWORKS TO PREDICT THE EFFECTIVENESS OF TARGETED HUMAN ACTIVITY

Aim. Approbation of mathematical methods of cluster analysis and artificial neural networks to solve the problems of classification and prediction of visual images reproduction in subjects with different properties of the nervous system and different «physiological cost» of activity. Materials and Methods. Conducted a comprehensive study of the subjects of both sexes aged 18 to 20 years in the implementation of targeted activities for the reproduction of visual images using various complexes of physiological and psychophysiological techniques, including the use of multivariate statistics (cluster analysis and the theory of artificial neural networks). Results. The role of individual characteristics of the Central nervous system (psychodynamic properties and indicators of functional asymmetry), as well as «physiological value» (on the basis of mathematical analysis of the heart rhythm) in the formation of unequal effectiveness of purposeful activity of the subjects in the reproduction of visual images was studied. The use of cluster analysis allowed us to identify 2 relatively homogeneous groups of subjects characterized by certain performance indicators in the reproduction of visual images: cluster 1 - highly productive; cluster 2 – low-impact. The leading indicator of performance was the number of erroneous elections, the second significant indicator of the selected clusters was the number of correct elec-tions. In the analysis of the average ranks of indicators, using the technology of artificial neural networks, it was found that the greatest role in solving the classification problem was played by the indicators of mathematical analysis of the heart rhythm, the second place was taken by the indicators of functional asymmetry, the third – psychodynamic characteristics. Conclusion. The use of methods of multivariate statistics (cluster analysis and the theory of artificial neural networks), allow to rank the studied indicators, to identify the most significant and on their basis to build predictive models, allowing with high representativeness to foresee the success of the specific test subjects in the reproduction of visual images.