Do simpler statistical methods perform better in multivariate long sequence time-series forecasting?
Li, Hao, Shao, Jie, Liao, Kewen and Tang, Mingjian. (2022). Do simpler statistical methods perform better in multivariate long sequence time-series forecasting? CIKM '22: The 31st ACM International Conference on Information and Knowledge Management. Atlanta, Georgia, United States of America 17 - 21 Oct 2022 Association for Computing Machinery (ACM). pp. 4168-4172 https://doi.org/10.1145/3511808.3557585
|Authors||Li, Hao, Shao, Jie, Liao, Kewen and Tang, Mingjian|
Long sequence time-series forecasting has become a central problem in multivariate time-series analysis due to its difficulty of consistently maintaining low prediction errors. Recent research has concentrated on developing large deep learning frameworks such as Informer and SCINet with remarkable results. However, these complex approaches were not benchmarked with simpler statistical methods and hence this part of the puzzle is missing for multivariate long sequence time-series forecasting (MLSTF). We investigate two simple statistical methods for MLSTF and provide analysis to indicate that linear regression owns a lower upper bound of error than deep learning methods and SNaive can act as an effective nonparametric method with unpredictable trends. Evaluations across six real-world datasets demonstrate that linear regression and SNaive are able to achieve state-of-the-art performance for MLSTF.
|Keywords||time-series forecasting; deep learning; statistical methods|
|Publisher||Association for Computing Machinery (ACM)|
|Digital Object Identifier (DOI)||https://doi.org/10.1145/3511808.3557585|
All rights reserved
File Access Level
|Book title||CIKM '22 : Proceedings of the 31st ACM International Conference on Information & Knowledge Management|
|Web address (URL) of conference proceedings||https://doi.org/10.1145/3511808|
|Online||17 Oct 2022|
|Publication process dates|
|Deposited||23 Jun 2023|
0views this month
0downloads this month