In addition to reducing traffic congestion, urban public transit systems are essential for supporting economic growth. However, delays often compromise their reliability. This study investigates the utility of machine learning (ML) models, enhanced with Explainable AI (XAI) techniques—specifically SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME)—to predict and interpret delays in public transportation. We employed Random Forest and kNearest Neighbors (kNN) models to identify and explain key features affecting delay predictions, such as time of day, day of the week, and route characteristics. SHAP provided consistent and robust global insights into feature importance, while LIME offered clear and straightforward local explanations for individual predictions. The combined use of SHAP and LIME enhances model transparency and trustworthiness, making the predictions more actionable for stakeholders. Our findings demonstrate that integrating XAI methods enhances the interpretability and practical application of delay predictions in public transportation.