A temporal investigation of crash severity factors in worker-involved work zone crashes: Random parameters and machine learning approaches

In the context of work zone safety, worker presence and its impact on crash severity has been less explored. Moreover, there is a lack research contributing factors by time-of-day. To accomplish this, first mixed logit model was used to determine statistically significant their effects. Significant in both models included work-zone-specific characteristics crash-specific characteristics, where environmental were only daytime model. addition, results from parameter transferability test provided evidence that nighttime crashes need be modeled separately. Further, explore nonlinear relationship between levels time-of-day, as well compare effects variables assess prediction performance, Support Vector Machines (SVM) trained Cuckoo Search (CS) algorithm utilized. Opening SVM black-box, variable analysis also performed. addition identified models, impacts vehicle-level characteristics. The illustrated termination area most critical for crashes, this location highest increase severe injury likelihood. summary, study demonstrate separately time-of-day with high level confidence. Furthermore, show CS-SVM provide better performance compared models.