Deep Learning for Time Series Forecasting: A Survey
Xiangjie Kong, Zhenghao Chen, Weiyao Liu, Kaili Ning, Lechao Zhang, Syauqie Muhammad Marier, Yichen Liu, Yuhao Chen, Feng Xia
TL;DR
This survey addresses the challenge of organizing the rapidly evolving field of deep learning for time series forecasting (DTSF) by proposing a dynamic architecture taxonomy and detailing feature extraction methodologies. It categorizes DTSF models into explicit-structure and implicit-structure families, covering encoder-decoder, Transformer, GAN-based, integrated, and cascade designs, and discusses their respective strengths and limitations. It also emphasizes series composition and feature-enhancement techniques, including dimension decomposition, time-frequency conversion, pre-training, and patch-based segmentation, while compiling datasets across energy, healthcare, traffic, meteorology, and economics. Finally, the paper highlights pressing challenges—privacy, interpretability, temporal continuity, parallel computing, and scaling—along with promising directions such as representation learning, causal inference, diffusion models, ensemble weighting, and cross-disciplinary approaches.
Abstract
Time series forecasting (TSF) has long been a crucial task in both industry and daily life. Most classical statistical models may have certain limitations when applied to practical scenarios in fields such as energy, healthcare, traffic, meteorology, and economics, especially when high accuracy is required. With the continuous development of deep learning, numerous new models have emerged in the field of time series forecasting in recent years. However, existing surveys have not provided a unified summary of the wide range of model architectures in this field, nor have they given detailed summaries of works in feature extraction and datasets. To address this gap, in this review, we comprehensively study the previous works and summarize the general paradigms of Deep Time Series Forecasting (DTSF) in terms of model architectures. Besides, we take an innovative approach by focusing on the composition of time series and systematically explain important feature extraction methods. Additionally, we provide an overall compilation of datasets from various domains in existing works. Finally, we systematically emphasize the significant challenges faced and future research directions in this field.