Grey Stochastic TOPSIS‐Based Ensemble Learning for Risk Assessment on Road Traffic Emergencies

ABSTRACT The rapid growth of the Online‐to‐Offline food delivery industry has increased the load on road transportation and posed significant challenges in managing road traffic risks. The existing road traffic risk assessment encounters two primary challenges. First, due to the inherent randomness and uncertainty of emergencies, it is difficult for decision‐makers to accurately predict the risk. Second, the scarcity and incompleteness of accident‐related data exacerbate the risk assessment process. To address these two issues, an uncertainty‐based ensemble learning framework for assessing the risks of road traffic emergencies is put forth. Initially, to resolve inconsistencies in linguistic evaluations across multiple criteria, a stochastic assessment mechanism is designed. Subsequently, to mitigate the effects of data loss and/or incompleteness, a grey stochastic model is introduced. Thereafter, a grey stochastic Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method is developed within the failure mode and effects analysis framework to estimate the probability of encountering the highest risk for road traffic emergencies. Ultimately, by incorporating the gradient boosting tree algorithm, we explore the impact of uncertainties on the probability of incurring the most severe risks for road traffic emergencies. A case study associated with full‐category food delivery within school nutrition programs is utilized to demonstrate the effectiveness and merits of the proposed method.