CAIFormer: A Causal Informed Transformer for Multivariate Time Series Forecasting

Xingyu Zhang, Wenwen Qiang, Siyu Zhao, Zeen Song, Changwen Zheng

Keywords: Time Series Forecasting; Causal Analysis

Abstract: Multivariate time series forecasting (MTSF) is crucial across various domains but remains challenging due to three main difficulties: capturing the temporal patterns of each variable, modeling complex cross-variable dependencies, and eliminating spurious correlations. Most existing MTSF methods adopt an all-to-all paradigm that feeds all variable histories into a unified model to predict their future values without distinguishing their individual roles. However, this undifferentiated paradigm makes it difficult to identify variable-specific causal influences and often entangles causally relevant information with spurious correlations. To address this limitation, we propose an all-to-one forecasting paradigm that predicts each target variable separately. Specifically, we first construct a Structural Causal Model (SCM) from observational data and then, for each target variable, we partition the historical sequence into four sub-segments according to the inferred causal structure: endogenous, direct causal, collider causal, and spurious correlation. The prediction relies solely on the first three causally relevant sub-segments, while the spurious correlation sub-segment is excluded. Furthermore, we propose Causal Informed Transformer (CAIFormer), a novel forecasting model comprising three components: Endogenous Sub-segment Prediction Block (ESPB), Direct Causal Sub-segment Prediction Block (DCSPB), and Collider Causal Sub-segment Prediction Block (CCSPB), which model the endogenous, direct causal, and collider causal sub-segments, respectively. Their outputs are then combined to produce the final prediction. Extensive experiments and ablation studies on multiple benchmark datasets demonstrate the effectiveness of the CAIFormer.

Primary Area: learning on time series and dynamical systems

Submission Number: 3987