TelecomTS: A Multi-Modal Observability Dataset for Time Series and Language Analysis

Austin Feng, Andreas Varvarigos, Ioannis Panitsas, Daniela Fernandez, Jinbiao Wei, Yuwei Guo, Jialin Chen, Ali Maatouk, Leandros Tassiulas, Zhitao Ying
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:30439-30469, 2026.

Abstract

Modern enterprises generate vast streams of time series metrics when monitoring complex systems, known as observability data. Unlike conventional time series from domains such as climate, observability data are zero-inflated, highly stochastic, and exhibit minimal temporal structure. Despite their importance, observability datasets remain underrepresented in public benchmarks due to proprietary restrictions and privacy concerns. Existing datasets are often anonymized and normalized, removing scale information and limiting their use for tasks such as anomaly detection, root cause analysis, and multi-modal reasoning. To address this gap, we introduce TelecomTS, a large-scale observability dataset derived from a 5G telecommunications network. TelecomTS features heterogeneous, de-anonymized covariates with explicit absolute scale information and provides a diverse suite of downstream tasks, including anomaly detection, root cause analysis, and multi-modal question-answering. Benchmarking state-of-the-art time series, language, reasoning, and multi-modal foundation models reveals that existing approaches struggle with the abrupt, noisy, and high-variance dynamics characteristic of observability data. Our experiments further underscore the importance of preserving covariates’ absolute scale, emphasizing the need for foundation time series models that natively leverage scale information for practical real-world observability applications. The code is available at: https://github.com/Ali-maatouk/TelecomTS.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-feng26s, title = {{T}elecom{TS}: A Multi-Modal Observability Dataset for Time Series and Language Analysis}, author = {Feng, Austin and Varvarigos, Andreas and Panitsas, Ioannis and Fernandez, Daniela and Wei, Jinbiao and Guo, Yuwei and Chen, Jialin and Maatouk, Ali and Tassiulas, Leandros and Ying, Zhitao}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {30439--30469}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/feng26s/feng26s.pdf}, url = {https://proceedings.mlr.press/v306/feng26s.html}, abstract = {Modern enterprises generate vast streams of time series metrics when monitoring complex systems, known as observability data. Unlike conventional time series from domains such as climate, observability data are zero-inflated, highly stochastic, and exhibit minimal temporal structure. Despite their importance, observability datasets remain underrepresented in public benchmarks due to proprietary restrictions and privacy concerns. Existing datasets are often anonymized and normalized, removing scale information and limiting their use for tasks such as anomaly detection, root cause analysis, and multi-modal reasoning. To address this gap, we introduce TelecomTS, a large-scale observability dataset derived from a 5G telecommunications network. TelecomTS features heterogeneous, de-anonymized covariates with explicit absolute scale information and provides a diverse suite of downstream tasks, including anomaly detection, root cause analysis, and multi-modal question-answering. Benchmarking state-of-the-art time series, language, reasoning, and multi-modal foundation models reveals that existing approaches struggle with the abrupt, noisy, and high-variance dynamics characteristic of observability data. Our experiments further underscore the importance of preserving covariates’ absolute scale, emphasizing the need for foundation time series models that natively leverage scale information for practical real-world observability applications. The code is available at: https://github.com/Ali-maatouk/TelecomTS.} }
Endnote
%0 Conference Paper %T TelecomTS: A Multi-Modal Observability Dataset for Time Series and Language Analysis %A Austin Feng %A Andreas Varvarigos %A Ioannis Panitsas %A Daniela Fernandez %A Jinbiao Wei %A Yuwei Guo %A Jialin Chen %A Ali Maatouk %A Leandros Tassiulas %A Zhitao Ying %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-feng26s %I PMLR %P 30439--30469 %U https://proceedings.mlr.press/v306/feng26s.html %V 306 %X Modern enterprises generate vast streams of time series metrics when monitoring complex systems, known as observability data. Unlike conventional time series from domains such as climate, observability data are zero-inflated, highly stochastic, and exhibit minimal temporal structure. Despite their importance, observability datasets remain underrepresented in public benchmarks due to proprietary restrictions and privacy concerns. Existing datasets are often anonymized and normalized, removing scale information and limiting their use for tasks such as anomaly detection, root cause analysis, and multi-modal reasoning. To address this gap, we introduce TelecomTS, a large-scale observability dataset derived from a 5G telecommunications network. TelecomTS features heterogeneous, de-anonymized covariates with explicit absolute scale information and provides a diverse suite of downstream tasks, including anomaly detection, root cause analysis, and multi-modal question-answering. Benchmarking state-of-the-art time series, language, reasoning, and multi-modal foundation models reveals that existing approaches struggle with the abrupt, noisy, and high-variance dynamics characteristic of observability data. Our experiments further underscore the importance of preserving covariates’ absolute scale, emphasizing the need for foundation time series models that natively leverage scale information for practical real-world observability applications. The code is available at: https://github.com/Ali-maatouk/TelecomTS.
APA
Feng, A., Varvarigos, A., Panitsas, I., Fernandez, D., Wei, J., Guo, Y., Chen, J., Maatouk, A., Tassiulas, L. & Ying, Z.. (2026). TelecomTS: A Multi-Modal Observability Dataset for Time Series and Language Analysis. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:30439-30469 Available from https://proceedings.mlr.press/v306/feng26s.html.

Related Material