Spatial-Temporal Identity: A Simple yet Effective Baseline for Multivariate Time Series Forecasting

TL;DR

This paper introduces a simple yet effective baseline for multivariate time series forecasting by incorporating spatial and temporal identity information, achieving high performance with efficiency comparable to complex graph neural networks.

Contribution

The paper proposes a novel baseline model using Spatial and Temporal IDentity (STID) with MLPs, addressing sample indistinguishability and challenging the necessity of complex STGNNs.

Findings

STID achieves state-of-the-art performance.

The model is more concise and efficient.

Sample indistinguishability is a key factor in MTS forecasting.

Abstract

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.