Next Article in Journal
Analysis of Economic and Growth Synchronization Between China and the USA Using a Markov-Switching–VAR Model: A Trend and Cycle Approach
Previous Article in Journal
The Explainability of Machine Learning Algorithms for Victory Prediction in the Video Game Dota 2 
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Computer Sciences & Mathematics Forum
Volume 11
Issue 1
10.3390/cmsf2025011027
csmf-logo
Submit to this Journal
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Proceeding Paper

Simplicity vs. Complexity in Time Series Forecasting: A Comparative Study of iTransformer Variants †

by
Polycarp Shizawaliyi Yakoi
1,*,
Xiangfu Meng
1,
Danladi Suleman
2,
Adeleye Idowu
3,
Victor Adeyi Odeh
4 and
Chunlin Yu
1
1
School of Electronic and Information Engineering, Liaoning Technical University, Huludao 125000, China
2
School of Science, Technology and Engineering, University of the Sunshine Coast, UniSC Moreton Bay, Petrie, QLD 4502, Australia
3
Department of Management Information Systems, Cyprus International University, 99258 Lefkoşa, Turkey
4
School of Information and Communication Engineering, University of Electronic Science and Technology of China, 611731 Chengdu, China
*
Author to whom correspondence should be addressed.
†
Presented at the 11th International Conference on Time Series and Forecasting, Canaria, Spain, 16–18 July 2025.
Comput. Sci. Math. Forum 2025, 11(1), 27; https://doi.org/10.3390/cmsf2025011027
Published: 22 August 2025
(This article belongs to the Proceedings of The 11th International Conference on Time Series and Forecasting)
Browse Figures
Versions Notes

Abstract

This study re-examines the balance between architectural intricacy and generalization in Transformer models for long-term time series predictions. We perform a systematic comparison involving a lightweight baseline (iTransformer) and two enhanced versions: MiTransformer, which incorporates an external memory component for extending context, and DFiTransformer, which features dual-frequency decomposition along with Learnable Cross-Frequency Attention. All models undergo training using the same protocols across eight standard benchmarks and four forecasting periods. Findings indicate that both MiTransformer and DFiTransformer do not reliably surpass the baseline. In many instances, the increased complexity leads to greater variance and decreased accuracy, especially with unstable or inconsistent datasets. These results imply that architectural minimalism, when effectively refined, can match or surpass the effectiveness of more complex designs—challenging the prevailing trend toward increasingly intricate forecasting architectures.
Keywords: time series forecasting; transformer architectures; model complexity; iTransformer; memory-augmented models; frequency-aware transformers; long-term forecasting; deep learning for time series; inductive bias; forecasting model evaluation time series forecasting; transformer architectures; model complexity; iTransformer; memory-augmented models; frequency-aware transformers; long-term forecasting; deep learning for time series; inductive bias; forecasting model evaluation

Share and Cite

MDPI and ACS Style

Yakoi, P.S.; Meng, X.; Suleman, D.; Idowu, A.; Odeh, V.A.; Yu, C. Simplicity vs. Complexity in Time Series Forecasting: A Comparative Study of iTransformer Variants. Comput. Sci. Math. Forum 2025, 11, 27. https://doi.org/10.3390/cmsf2025011027

AMA Style

Yakoi PS, Meng X, Suleman D, Idowu A, Odeh VA, Yu C. Simplicity vs. Complexity in Time Series Forecasting: A Comparative Study of iTransformer Variants. Computer Sciences & Mathematics Forum. 2025; 11(1):27. https://doi.org/10.3390/cmsf2025011027

Chicago/Turabian Style

Yakoi, Polycarp Shizawaliyi, Xiangfu Meng, Danladi Suleman, Adeleye Idowu, Victor Adeyi Odeh, and Chunlin Yu. 2025. "Simplicity vs. Complexity in Time Series Forecasting: A Comparative Study of iTransformer Variants" Computer Sciences & Mathematics Forum 11, no. 1: 27. https://doi.org/10.3390/cmsf2025011027

APA Style

Yakoi, P. S., Meng, X., Suleman, D., Idowu, A., Odeh, V. A., & Yu, C. (2025). Simplicity vs. Complexity in Time Series Forecasting: A Comparative Study of iTransformer Variants. Computer Sciences & Mathematics Forum, 11(1), 27. https://doi.org/10.3390/cmsf2025011027

Article Metrics

Back to TopTop