Journal article · 2026 · Open access

Transformers for Multivariate Time Series Forecasting: Comprehensive Analysis, Challenges, Research Opportunities, and Future Prospects

M K Saravana, M. S. Roopa, J S Arunalatha, K R Venugopal

IEEE Access, 14, 11424–11457 · Published 1 January 2026 · DOI 10.1109/access.2026.3654408

Open full textCite this paper

In brief

“Transformers for Multivariate Time Series Forecasting: Comprehensive Analysis, Challenges, Research Opportunities, and Future Prospects” is a 2026 journal article by M K Saravana and 3 co-authors, published in IEEE Access. OpenAlex counts 58 citations as of 17 November 2026, which puts it in the top 1% of works published in 2026 by citations. It is open access under a CC BY licence: the full text is free to read at IEEE Access.

Citations
58
Most cited in
2026 (56)
Field-weighted impact
102.15
References
150
Authors
4

Abstract

Multivariate time series (MTS) forecasting appears in many applied settings, ranging from finance and healthcare to energy systems, climate analysis, and IoT-based platforms. In these fields, practitioners rely on forecasts not just for reporting, but for making everyday operational and planning decisions. Over the years, a variety of sequence-modeling methods such as RNNs, LSTMs, CNNs, and TCNs have been used for this purpose. In practice, however, these models often have difficulty dealing with long temporal dependencies, interactions across multiple variables, and the computational demands that arise when datasets grow large. As a result, their performance and scalability can be inconsistent when they are deployed in real applications. In recent years, Transformer-based models have gained attention as an alternative because their self-attention mechanism and parallel processing capabilities allow more flexible handling of long and complex sequences. This survey therefore reviews the main Transformer-based approaches developed for MTS forecasting, covering early architectures, efficient attention variants, decomposition-oriented designs, patch-based techniques, hybrid frameworks, and large-scale pretrained models. In addition, we organize the existing literature into a clear taxonomy that includes commonly used datasets and evaluation metrics, and we discuss the relative strengths and limitations of these methods. Additionally, we discuss key challenges such as computational complexity, interpretability, data sparsity, and generalization, and outline future research directions, including spatiotemporal modeling, self-supervised learning, physics-informed designs, explainable architectures, and federated learning. The survey consolidates existing knowledge and provides guidance for researchers and practitioners to select, adapt, and innovate Transformer-based models for effective MTS forecasting.

Cite this paper

Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.

APA 7

Saravana, M. K., Roopa, M. S., Arunalatha, J. S., & Venugopal, K. R. (2026). Transformers for Multivariate Time Series Forecasting: Comprehensive Analysis, Challenges, Research Opportunities, and Future Prospects. IEEE Access, 14, 11424–11457. https://doi.org/10.1109/access.2026.3654408

MLA 9

Saravana, M K, et al. “Transformers for Multivariate Time Series Forecasting: Comprehensive Analysis, Challenges, Research Opportunities, and Future Prospects.” IEEE Access, vol. 14, 2026, pp. 11424–11457. https://doi.org/10.1109/access.2026.3654408.

Chicago

Saravana, M K, M. S. Roopa, J S Arunalatha, and K R Venugopal. 2026. “Transformers for Multivariate Time Series Forecasting: Comprehensive Analysis, Challenges, Research Opportunities, and Future Prospects.” IEEE Access 14: 11424–11457. https://doi.org/10.1109/access.2026.3654408.

Harvard

Saravana, M. K. et al. (2026) ‘Transformers for Multivariate Time Series Forecasting: Comprehensive Analysis, Challenges, Research Opportunities, and Future Prospects’, IEEE Access, 14, pp. 11424–11457. doi:10.1109/access.2026.3654408.

IEEE

M. K. Saravana, M. S. Roopa, J. S. Arunalatha, and K. R. Venugopal, “Transformers for Multivariate Time Series Forecasting: Comprehensive Analysis, Challenges, Research Opportunities, and Future Prospects,” IEEE Access, vol. 14, pp. 11424–11457, 2026, doi: 10.1109/access.2026.3654408.

BibTeX

@article{saravana2026transformers,
  title     = {{Transformers for Multivariate Time Series Forecasting: Comprehensive Analysis, Challenges, Research Opportunities, and Future Prospects}},
  author    = {Saravana, M K and Roopa, M. S. and Arunalatha, J S and Venugopal, K R},
  journal   = {IEEE Access},
  year      = {2026},
  volume    = {14},
  pages     = {11424--11457},
  doi       = {10.1109/access.2026.3654408},
  url       = {https://doi.org/10.1109/access.2026.3654408}
}

Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)

Topic and keywords

Physical Sciences › Computer Science › Signal Processing › Time Series Analysis and Forecasting

  • multivariate time series forecasting
  • transformer models
  • self-attention
  • long-range temporal dependencies
  • efficient attention mechanisms
  • pre-trained models
  • computational complexity
  • interpretability

Work with this paper

Most-cited papers in Computer Science

All highly cited Computer Science papers

Questions about this paper

How do I cite this paper in APA style?

Saravana, M. K., Roopa, M. S., Arunalatha, J. S., & Venugopal, K. R. (2026). Transformers for Multivariate Time Series Forecasting: Comprehensive Analysis, Challenges, Research Opportunities, and Future Prospects. IEEE Access, 14, 11424–11457. https://doi.org/10.1109/access.2026.3654408

How many times has this paper been cited?

58 times, according to OpenAlex on 17 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.

Is this paper free to read?

Yes. It is open access at IEEE Access: sign in and open the full text from this page.

Who wrote this paper?

M K Saravana, M. S. Roopa, J S Arunalatha, K R Venugopal.

What is the DOI of this paper?

10.1109/access.2026.3654408, permanently at https://doi.org/10.1109/access.2026.3654408. Sign in and open it from this page.

Bibliographic data from OpenAlex (CC0), last checked 17 November 2026. ResearchGuru does not host this paper; once you are signed in, the links open it at the publisher or the repository that does.