A method of entropy-predictive analysis of the integrated environmental risk index of municipal organic waste

• Дата: 2026
• Автори: Тригуба, А. М., Коваль, Н. Я., Тригуба, І. Л., Фірман, І. Р., Фамуляк, В. Ю.
• Формат: Стаття
• Мова: Українська
• Опубліковано: Інститут проблем реєстрації інформації НАН України 2026
• Теми: municipal organic waste; environmental risk; multifactor data; entropy analysis; integrated index; machine learning; decision support; forecasting
• Онлайн доступ: https://drsp.ipri.kiev.ua/article/view/358598
• Назва журналу: Data Recording, Storage & Processing

In this study, it is proposed a method of entropy-predictive analysis of the integrated environmental risk index of municipal organic waste under conditions of multifactor uncertainty in the urban environment. First, the key characteristics of the municipal organic waste management system are outlined as a weakly formalized domain influenced by a combination of social-demographic, geospatial, infrastructural, technological, environmental, and temporal factors. It is shown that conventional approaches to waste assessment are not sufficiently effective when the task requires simultaneous consideration of heterogeneous indicators, nonlinear dependencies, and spatial variability of risk formation. Second, the investigation substantiates the structure of a multifactor data array used for environmental risk assessment and forecasting. Special attention is paid to entropy-based evaluation of feature significance, which makes it possible to determine the actual contribution of individual indicators to the formation of the target variable and to reduce the influence of redundant or weakly informative parameters. On this basis, the method integrates procedures of data normalization, entropy ranking of features, weight determination, and construction of an integrated environmental risk index. Third, the predictive component of the proposed method is considered in the context of machine learning application. The integrated risk index is used as a target variable for forecasting, while the input space is formed from ranked multifactor indicators. It is analyzed the comparative performance of several predictive models and show that the integration of entropy-based feature processing with machine learning improves the explanatory ability of the models and increases the reliability of the obtained forecasts for decision-making purposes. The scientific contribution of the study lies in combining entropy-based feature significance assessment, integrated risk indexing, and predictive modeling into a unified analytical workflow oriented toward municipal waste management. The practical value of the method consists in its applicability for identifying territories with elevated environmental risk, prioritizing infrastructure interventions, optimizing collection and processing strategies, and supporting evidence-based management decisions in municipal organic waste systems. Tabl.: 3. Fig.: 4. Refs: 35 titles.