Correlated Time Series Forecasting

Correlated time series forecasting aims to predict future values of multiple interconnected time series, leveraging their temporal dynamics and spatial relationships. Recent research emphasizes automated model design, employing graph neural networks and recurrent architectures like GRUs, often incorporating techniques like pre-training and efficient operator stacking to improve accuracy and scalability. This field is crucial for applications across various domains, including finance, traffic management, and cyber-physical systems, where accurate predictions of interconnected processes are essential for optimization and control. The current focus is on developing lightweight, efficient models that can achieve state-of-the-art accuracy while handling large-scale datasets.