Spatial-Temporal Identity: A Simple yet Effective Baseline for Multivariate Time Series Forecasting
This addresses the problem of complex models with limited gains in multivariate time series forecasting for applications like traffic or weather prediction, offering a more efficient solution.
The paper tackled the indistinguishability of samples in multivariate time series forecasting by proposing a simple baseline using spatial and temporal identity information with MLPs, achieving state-of-the-art performance and efficiency.
Multivariate Time Series (MTS) forecasting plays a vital role in a wide range of applications. Recently, Spatial-Temporal Graph Neural Networks (STGNNs) have become increasingly popular MTS forecasting methods due to their state-of-the-art performance. However, recent works are becoming more sophisticated with limited performance improvements. This phenomenon motivates us to explore the critical factors of MTS forecasting and design a model that is as powerful as STGNNs, but more concise and efficient. In this paper, we identify the indistinguishability of samples in both spatial and temporal dimensions as a key bottleneck, and propose a simple yet effective baseline for MTS forecasting by attaching Spatial and Temporal IDentity information (STID), which achieves the best performance and efficiency simultaneously based on simple Multi-Layer Perceptrons (MLPs). These results suggest that we can design efficient and effective models as long as they solve the indistinguishability of samples, without being limited to STGNNs.