Impact of road geometry on school-area traffic congestion using regression and machine learning analysis: lessons from six Saudi cities

Abstract

• Physical road geometric elements contribute measurably to school areas congestion intensity. • The Machine Learning showed superior accuracy compared with the Regression method. • The Extreme Gradient Boosting outperforms the Regression model. • Travel Time Index outperforms the Planning Time Index and Buffer Time Index in capturing delays. • The key geometric factors are the road category, number of lanes per direction, and on-street parking availability. Traffic congestion in school vicinities has appeared as a context-specific condition posing recurring operational challenges, particularly in car-dependent urban areas. This study examines the impact of road geometric characteristics on traffic congestion near schools in six major Saudi cities during student drop-off and pick-up periods. A total of 242 road segments were analyzed using Travel Time Reliability (TTR) metrics, including Travel Time Index (TTI), Planning Time Index (PTI), and Buffer Time Index (BTI), which were derived from Google Maps API over 20 daily time intervals. The Multiple Regression (MR) and Machine Learning (ML) methods were employed to model the impact of road geometric elements such as road classification, number of lanes per direction, traffic directions, and the presence of median and on-street parking on congestion levels. The analysis yields nuanced perspectives on the ML accuracy compared with the regression method, underscoring its potential for quantifying the importance of road geometric elements on the traffic condition. The Extreme Gradient Boosting (XGBoost) outperforms the Regression model across all indices with R 2 value of 71% for the TTI model. The TTI outperforms the PTI and BTI in capturing operational delays. The key geometric factors that show a strong influence are the road category and the number of lanes per direction. Fuel consumption and CO 2 emissions were estimated, revealing a positive correlation between travel time indicators and environmental implications. The findings emphasize the importance of incorporating geometric design into school zone traffic management for enhancing mobility and sustainability in car-dependent cities.