Abstract
Problem. The article is devoted to the development of a methodology for proactive analysis of road traffic accidents (RTAs) in the Republic of Kazakhstan (RK). Traditional retrospective approaches do not provide sufficient effectiveness in preventing incidents under conditions of annually. increasing accident rates and significant socio-economic losses, which exceed USD 7 billion per year. Goal. The aim of this study is to provide a theoretical justification for a proactive analysis methodology based on machine learning (ML) models. Methodology. The proposed approach is grounded in the integration of Big Data obtained from the national digital platform TOR (Traffic Operational Response) and the application of predictive ML models such as Random Forest and XGBoost. Originality. The scientific novelty lies in the synthesis of ML models and GIS-based analysis to create a dynamic proactive model for RTA risk assessment, adapted for the first time to the specific data environment of Kazakhstan. The developed framework enables the prediction of both the probability and the severity of RTAs on specific road segments using dynamic influencing factors. Results.The results can be utilized by road infrastructure agencies and law enforcement bodies in Kazakhstan for targeted patrolling and proactive interventions. Practical value. It is recommended to integrate the ML module directly into the TOR platform and to establish standardized interagency data exchange procedures.