<ul><li>Multivariate time series (MTS) forecasting is crucial for various industries and research sectors.</li><li>Spatial-temporal graph neural networks (STGNNs) are popular for MTS forecasting, but struggle with computational complexity and identifying patterns in extensive historical datasets.</li><li>A new k-nearest neighbor MTS forecasting (kNN-MTS) framework is introduced, using a retrieval mechanism over a large datastore without additional training.</li><li>The hybrid spatial-temporal encoder (HSTEncoder) enhances kNN-MTS by capturing long-term temporal and short-term spatial-temporal dependencies, leading to improved forecasting performance on real-world datasets.</li></ul>

Nearest Neighbor Multivariate Time Series Forecasting

Discover more