Answers are grounded in this paper and cite their sources.
Ice formation on roads during the winter months poses a serious threat to traffic safety. In regions with harsh winters, adverse weather conditions can lead to accidents causing significant material and psychological losses. The need for effective icing prediction and warning systems is increasing in order to minimize such accidents. In this study, a mobile application was developed to predict road icing in advance and provide real-time alerts to drivers. It was hypothesized that automatically retrieving meteorological data via an API would reduce errors caused by manual entry and improve prediction accuracy. Temperature, relative humidity, dew point, wind speed, and snowfall data obtained from the Open-Meteo platform were processed and utilized for icing prediction. To enhance prediction accuracy, various machine learning algorithms such as Random Forest, SVM, and MLP were trained and their performance was compared. In the final model, a Voting Classifier-based ensemble method that combines the strengths of these algorithms was employed. The developed ensemble model outperformed individual models, achieving 93.20% accuracy, 92.16% macro recall, 93.20% weighted recall, 91.56% macro F1, and 93.25% weighted F1 scores. These results particularly improved the accurate prediction of high-risk icing classes. Finally, a user-friendly interface was designed to provide drivers with early warnings based on the predicted level of risk. The findings are expected to contribute to raising driver awareness of icing risks and enhancing road safety.
No takes yet. Share an insight, caveat, or question.
Nacak et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: