Explainable AI to Enhance Demand Forecasting Model for Autonomous Taxis in Smart Cities

The rapid evolution of Autonomous Taxis (ATs) in smart cities requires robust and accurate demand forecasting models to optimize resource allocation and ensure smooth mobility services. Traditional forecasting methods often function as black-box models, limiting interpretability and trust in their predictions. This study focuses on enhancing demand forecasting for ATs by incorporating Explainable Artificial Intelligence (XAI) to improve the transparency and interpretability of predictions. Simulations conducted on demand forecasting features demonstrate that the incorporation of XAI not only provides clearer insights into the prediction process but also increases prediction accuracy. For evaluation, we utilize a realworld dataset comprising over 4,500 taxis in Bangkok, Thailand, highlighting the practical benefits of this approach. The results underscore the potential of XAI to provide actionable insights for urban planners and fleet managers, paving the way for smarter and more adaptive mobility systems in the era of autonomous transportation.