A Combination Model for Time Series Prediction using LSTM via Extracting Dynamic Features Based on Spatial Smoothing and Sequential General Variational Mode Decomposition

TL;DR

The paper tackles non-stationary, nonlinear time series prediction in market sales by proposing a hybrid SS-LSTM model that fuses Spatial Smoothing-derived dynamic features with Sequential General Variational Mode Decomposition to capture global market trends and seasonal frequencies. Dynamic features are modeled by multiple LSTM networks dedicated to trend, frequency, and residual components, and forecasts are reconstructed to the original series via diagonal averaging. Empirical results on both small- and large-amplitude seasonal datasets show that SS-LSTM generally outperforms traditional decomposition-based hybrids and plain LSTM, offering improved accuracy in calmer market conditions and robustness to complex dynamics, with limitations in highly volatile regimes. The approach provides a principled way to extract interpretable dynamic features and utilize them in deep learning models for improved time series forecasting in finance and economics.

Abstract

In order to solve the problems such as difficult to extract effective features and low accuracy of sales volume prediction caused by complex relationships such as market sales volume in time series prediction, we proposed a time series prediction method of market sales volume based on Sequential General VMD and spatial smoothing Long short-term memory neural network (SS-LSTM) combination model. Firstly, the spatial smoothing algorithm is used to decompose and calculate the sample data of related industry sectors affected by the linkage effect of market sectors, extracting modal features containing information via Sequential General VMD on overall market and specific price trends; Then, according to the background of different Market data sets, LSTM network is used to model and predict the price of fundamental data and modal characteristics. The experimental results of data prediction with seasonal and periodic trends show that this method can achieve higher price prediction accuracy and more accurate accuracy in specific market contexts compared to traditional prediction methods Describe the changes in market sales volume.

Figures (14)

• Figure 1: Antenna array receives arrival signals and its coherent signals.
• Figure 2: Schematic of data processing in Spatial Smoothing algorithm and Dynamic Features extraction.
• Figure 3: The LSTM cell
• Figure 4: After the dynamic feature extraction is completed, it is divided into three different types according to the variation form of the dynamic feature series via SGVMD and the workflow diagram of the multi-parameter LSTM method.
• Figure 5: Construction of snapshots from the data trajectory matrix.