Explainable AI for Big Data Analytics in Urban Mobility Forecasting

Urban mobility systems produce immense volumes of heterogeneous data from diverse sources such as GPS logs, traffic sensors, weather feeds, and ride-hailing platforms. While these data offer significant potential for enhancing traffic forecasting and policy decisions, conventional AI models used for such analytics often rely on complex, opaque architectures that lack interpretability, limiting their effectiveness in real-world governance. To address this challenge, we propose an Explainable AI (XAI) framework tailored for big data analytics in urban mobility forecasting. Our approach integrates a hybrid LSTM-XGBoost model for spatiotemporal prediction, complemented by Shapley Additive Explanations (SHAP), to provide transparent, feature-level insights. This framework offers dual advantages: high predictive accuracy and actionable interpretability, enabling stakeholders to understand, validate, and trust model decisions. Experiments on largescale datasets, including New York City's TLC trip records, demonstrate strong forecasting performance (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$R^{2}=0.91$</tex>) and clear visualization of key factors such as peak-hour surges and weather-driven demand fluctuations. The proposed solution contributes to the deployment of transparent and trustworthy AI in smart city applications.