Data-driven spatial-temporal analysis of highway traffic volume considering weather and festival impacts

Artikel i vetenskaplig tidskrift, 2022

This paper aims to discover the relationships among the weather, holidays, and the traffic volume using multisource data from the Guangdong-Hong Kong-Macao Greater Bay Area (GBA) and to reveal the corresponding regional spatial–temporal traffic and migration patterns. Using accurate hourly weather and traffic volume data, this study examines the traffic volume from the origin to the destination county by considering traffic factors, weather factors, and temporal factors. A Random-effect regression model and a random forest model are established to analyze the above factors and identify the factors that contribute to the annual variation in traffic patterns. An RER + RF fusion prediction model based on ridge regression is proposed to predict the hourly traffic volume from origin to destination county, and is adopted in the spatial–temporal submodels. The results show that the impact of rainfall on traffic volume varies as the rainfall varies, and a rain-induced traffic pattern shift towards highway travel is found, which interacts with the negative effect of rainfall on highway traffic volumes. The Spring Festival holiday witnesses a V-shaped traffic volume curve during the study period. Some traffic pattern differences are also found in different spatial–temporal submodels. The RER + RF fusion model performs better in predicting in parent model and most of the spatial–temporal submodels, which validates the proposed model in predicting the traffic volume. The findings can provide transport agencies, urban planning agencies, and urban agglomeration travelers with valuable information for highway transport activity analysis considering the effects of weather and festival events.