Nested Spatio-Temporal Time Series Forecasting

In three linesNested spatio-temporal forecasting framework coupling macro-level regional trends with micro-level historical observations. Uses spectral clustering to construct semantically coherent regions, filtering systematic noise while preserving trends. Progressive coarse-to-fine predictor integrates features to anticipate dynamic anomalies. Outperforms state-of-the-art baselines on high-dimensional datasets.

Summary generated by Claude — human-verified