DA-SPS: A Dual-stage Network based on Singular Spectrum Analysis, Patching-strategy and Spearman-correlation for Multivariate Time-series Prediction

TL;DR

This work tackles multivariate time-series forecasting by addressing the varying influence of extraneous variables and by exploiting distinct temporal patterns within the target sequence. It introduces DA-SPS, a dual-stage network with TVPS (SSA-based decomposition plus LSTM and P-Conv-LSTM) for the target and EVPS (Spearman correlation filtering plus L-Attention) for correlated extraneous variables, with outputs fused for final prediction. Across four public datasets and a private laptop-board yield dataset, DA-SPS demonstrates state-of-the-art accuracy and robustness, supported by comprehensive ablations that highlight the importance of SSA, patch-based trend modeling, and correlation-aware feature selection. The approach offers practical impact for industrial and real-world forecasting tasks where extraneous information varies in relevance and temporal structure.

Abstract

Multivariate time-series forecasting, as a typical problem in the field of time series prediction, has a wide range of applications in weather forecasting, traffic flow prediction, and other scenarios. However, existing works do not effectively consider the impact of extraneous variables on the prediction of the target variable. On the other hand, they fail to fully extract complex sequence information based on various time patterns of the sequences. To address these drawbacks, we propose a DA-SPS model, which adopts different modules for feature extraction based on the information characteristics of different variables. DA-SPS mainly consists of two stages: the target variable processing stage (TVPS) and the extraneous variables processing stage (EVPS). In TVPS, the model first uses Singular Spectrum Analysis (SSA) to process the target variable sequence and then uses Long Short-Term Memory (LSTM) and P-Conv-LSTM which deploys a patching strategy to extract features from trend and seasonality components, respectively. In EVPS, the model filters extraneous variables that have a strong correlation with the target variate by using Spearman correlation analysis and further analyses them using the L-Attention module which consists of LSTM and attention mechanism. Finally, the results obtained by TVPS and EVPS are combined through weighted summation and linear mapping to produce the final prediction. The results on four public datasets demonstrate that the DA-SPS model outperforms existing state-of-the-art methods. Additionally, its performance in real-world scenarios is further validated using a private dataset collected by ourselves, which contains the test items' information on laptop motherboards.

Figures (7)

• Figure 1: The framework of the DA-SPS model.
• Figure 2: The structure of the P-Conv-LSTM.
• Figure 3: The structure of the L-Attention. Here, we assume that the selected variables are variables 2, 5, $\dots$, and $M$. The total number of the variables is $Q$.
• Figure 4: MAE curves for each model across Electricity, Solar, Traffic, and Exchange datasets for various prediction tasks.
• Figure 5: Prediction curves for different models in the $\operatorname{horizon}=6$ task.