ЗАСТОСУВАННЯ ГРАФОВИХ НЕЙРОННИХ МЕРЕЖ ДЛЯ ПРОГНОЗУВАННЯ ЛОКАЛЬНИХ КЛІМАТИЧНИХ УМОВ

This work discusses the development of a system for predicting local climate indicators based on Graph Neural Networks. Data from Internet of Things sensors and known weather stations are integrated using open Application Programing Interfaces, providing a comprehensive view of meteorological conditions. The system considers the geographical structure of the area, optimizes forecast accuracy, and adapts to changes in climate conditions. The use of the latest smart sensor technologies and graph neural networks makes the developed system more efficient to improve the accuracy and accessibility of forecasts for users. The built-in interface allows visualizing the results and provides practical value for local use.