APPLICATION OF SYSTEM ANALYSIS AND MACHINE LEARNING METHODS IN THE LOGINOM ENVIRONMENT FOR CLUSTERING ENVIRONMENTAL MONITORING DATA OF THE MEGACITY

Krynetskaya Anastasia Dmitrievna, Trushina Veronika Igorevna

The paper considers the use of system analysis and machine learning tools implemented in the domestic Loginom analytical platform for the study of environmental monitoring data for atmospheric air in Moscow. The data obtained from the open data portal of the Moscow Government, containing average monthly concentrations of pollutants measured by automatic air pollution control stations, was downloaded, preprocessed, filtered and clustered. The correlation between the concentrations of pollutants and the types of observation zones (highways, residential areas, natural areas, etc.) was estimated. Cluster analysis methods and the Pearson correlation coefficient were used to test the hypothesis that the spatial distribution of pollutants depends on the nature of anthropogenic load. It is shown that the Loginom platform allows you to implement a full cycle of analytical data processing without programming, ensuring reproducibility and visualization of results. The work confirms the possibility of using Russian data mining tools for environmental monitoring tasks.

Keywords:system analysis, machine learning, Loginom, clustering, Pearson correlation, air pollution, environmental monitoring, data processing.