Understanding pedestrian behavior at unsignalized intersections becomes crucial for improving traffic safety. In this study, the crossing speeds of pedestrians at four unsignalized intersections in Izmir, Türkiye were investigated. Data were collected on weekdays for one hour each during peak and off-peak hours using unmanned aerial vehicles and cameras. The study incorporated pedestrian crossings executed on the designated crosswalks, as well as those occurring within 30 meters from the crosswalk. Two of the intersections did not have crosswalks, thus pedestrian crossings within the initial 40 meters from the roadway were considered. This study employed the eXtreme Gradient Boosting (XGBoost) machine learning algorithm to predict pedestrian crossing speed based on a comprehensive set of factors. Hyperparameter tuning was performed to optimize the model's performance. Various features, including the existence of a crosswalk, pedestrian age (young, adult, elderly), gender, group crossings, and load-carrying situations were considered in the model. The results of the study provide valuable insights for traffic management. This research contributes to the ongoing efforts to enhance pedestrian safety.

Pedestrian Crossing Speed, Unsignalized Intersections, XGBoost