CLEAR-GNN: connectivity-learning enhanced adaptive regression using streamlined GNN

Traffic forecasting is an essential component in intelligent transportation systems (ITS) to ensure effective traffic flow management, real-time decision-making, and congestion mitigation. The existing model's spatiotemporal dependencies and scalability issues motivate the study to develop effective traffic forecasting systems to ensure robust solutions in real-time applications. The dependency complexities reduce the suboptimal prediction accuracy and maximize the computational cost while observing the traffic features. The research difficulties are overcome by introducing the Streamlined Graph Neural Networks (SGNN) that predict the traffic flow and maximize the prediction accuracy. The SGNN method combines sparse graph analysis, factorized graph learning, and temporal message passing to improve scalability and predictive accuracy. The network traffic interactions are optimized by observing node connectivity, and the network parameters are optimized using the memetic optimization algorithm. The optimization process balances the forecasting precision and computational complexity. The system's efficiency is evaluated using experimental results, reducing the 17.5% error rate and 32.6% enhancement in prediction accuracy. The effective utilization of the traffic forecasting process is applicable in real-time adaptive transportation control, smart-city planning, and traffic management. Thus, the process produces robust and scalable solutions in intelligent transportation systems.