Pith. sign in

REVIEW 3 major objections 6 minor 51 references

How Much Can Time-related Features Enhance Time Series Forecasting?

T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that explicit time-stamp encoding, added as a lightweight branch, cuts a linear forecaster's average MSE by about 23% on Electricity and Traffic and lets a 100k-parameter model match much larger networks across seven…

desk verdict TimeSter's core claim holds—timestamp features genuinely help—but the headline 23% gain is the best of several feature sets chosen on test labels, so treat it as optimistic until a validation protocol is added. read the letter →

arxiv 2412.01557 v1 pith:O57UAMKS submitted 2024-12-02 cs.LG stat.ML

Chaolv Zeng , Yuan Tian , Guanjie Zheng , Yunjun Gao This is my paper · ORCID
classification cs.LGstat.ML
keywords timeseriesforecastingtime-relatedfeaturesstampencodinglong-termlinearmodelsseasonalitytemporalfeatureselectionmultivariate
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that the calendar information attached to each data point—hour of day, day of week, month, and season—is itself a powerful forecasting signal that most modern models leave underused. To make that signal explicit, it introduces TimeSter, a small encoder that turns historical timestamps into pseudo-observations and projects them to the future, and TimeLinear, which adds TimeSter's forecast to a plain linear backbone. On Electricity and Traffic this addition reduces average MSE by about 23% relative to the linear model alone, and across seven datasets TimeLinear matches or beats larger Transformer- and convolution-based models with far fewer parameters. The claim matters because it suggests a cheap, architecture-agnostic way to improve long-term forecasts.

What carries the argument

The load-bearing object is the TimeSter encoder-decoder pair coupled with a weighted late fusion. TimeSter encodes historical timestamps $U$ through two nonlinear layers, a 1D convolution, and LayerNorm into pseudo-observations $X_U$, then applies a single linear layer $Y_U = W X_U + b$ to produce a future forecast; the final prediction is $Y' = \beta Y_B + (1-\beta) Y_U$ with the backbone forecast $Y_B$, plus simplified reversible instance normalization. This machinery isolates stationary calendar cycles, which repeat every day, week, month, or season, from the residual non-cyclical part the backbone must learn, which is why a linear backbone can suddenly compete with much larger models.

What would settle it

Hold out a later year whose seasonal pattern is shifted relative to the training years, for example a dataset spanning a policy change or an unusually warm year, and train TimeLinear on the earlier years: if the MSE gain over the plain linear backbone disappears or becomes negative, the calendar-stationarity assumption is the load-bearing part.

Watch

Extended reading notes

Core claim

The paper's central discovery is that time stamps are not just auxiliary metadata but a first-class predictive signal: given only the hour, weekday, month, and season of a future point, a single linear layer can already forecast strongly periodic series competitively, and adding that timestamp forecast to a backbone's observation-based forecast yields large and consistent gains. TimeSter models the conditional distribution of observations given timestamps and decodes it with one linear map, so its contribution is added to the backbone's as a fixed weighted sum. The paper further finds that the benefit is largest for channel-independent backbones and for datasets with clear daily, weekly, or seasonal autocorrelation, and that the optimal timestamp features shift with dataset and prediction horizon.

Load-bearing premise

The timestamp branch assumes the calendar pattern learned from the training years repeats unchanged in the forecast period, because the year feature is deliberately omitted and a fixed linear map is used; a multi-year trend or a shift in seasonal timing would break that contribution.

Editorial extensions

If this is right

  • Adding TimeSter to a plain linear projector cuts average MSE by about 23% on Electricity and Traffic and improves results across all seven benchmark datasets.
  • TimeSter helps across architectures, with larger gains for channel-independent backbones such as RLinear, FITS, and PatchTST than for channel-dependent ones such as ModernTCN, TimesNet, and iTransformer.
  • On strongly periodic data, timestamps alone can outperform observation-only linear forecasting for a 720-step horizon, implying that calendar cycles dominate the predictable signal in such series.
  • The optimal timestamp feature set changes with dataset and forecast length: hour suffices for daily cycles, while month and season become useful for 336- and 720-step horizons.
  • The benefit of explicit timestamps shrinks as the look-back window grows, because longer history already contains the cycles the timestamp branch encodes.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper omits the year feature deliberately, so the method implicitly assumes calendar effects are stationary across years; a natural extension, which the paper does not test, is a year-aware or debiased timestamp embedding.
  • A testable boundary: on series with weak or drifting periodicity, such as financial returns or sensor noise, the timestamp branch should add little, so measuring the gain against autocorrelation strength would delimit where TimeSter helps.
  • The fixed mixing weight $\beta$ could be made time-varying, letting the model down-weight calendar priors during regime shifts, an option the paper leaves unexplored.
  • Because TimeSter is a small separate branch, it could double as a lightweight domain-calibration module: re-fitting only the timestamp embedding on new data might adapt a pretrained backbone to a new seasonality without retraining the whole model.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. This paper studies whether explicit timestamp features (hour of day, day of week, month, season, minute) can improve long-term multivariate time series forecasting. It introduces TimeSter, a module that encodes historical timestamp features with two nonlinear layers, a Conv1d, and a linear layer, then predicts via a single linear projector; the TimeSter output is combined with a backbone forecaster through a fixed weighting β. The main model TimeLinear uses a linear backbone and simplified RevIN. The authors report that TimeSter reduces the average MSE of RLinear by about 23% on Electricity and Traffic, improves several other backbones (FITS, PatchTST, ModernTCN, TimesNet, iTransformer), and that TimeLinear is competitive with or better than more complex state-of-the-art models on seven benchmark datasets with 100k–1M parameters. The paper includes ablations of feature subsets, module components, decoder variants, longer look-back windows, robustness over three seeds, and an ACF-based explanation for feature selection.

Significance. If the headline result is reproducible under a pre-specified protocol, this is a useful contribution to the LTSF empirical literature. The proposed module is simple, has a clear intuition, and is evaluated across multiple architectures; the paper ships code, reports three-seed robustness for the main model, and provides an ACF analysis that makes the feature choices falsifiable. The main caveat is that the headline 23% gain currently depends on choosing the best time-feature subset on the test set, which inflates the expected improvement of a fixed model. With a validation-based selection protocol or a fixed feature set, the quantitative claim would be much stronger.

major comments (3)
  1. [§5.2.1, §5.3.5, Tables 5, 8, 9] No validation protocol is defined for the per-dataset/per-horizon choice of time-related features. The selections in Table 9 coincide with the best-performing rows on the test set (Table 5 for T=720, with the Min_H addition for ETTm1 from Table 8): for Electricity at T=720, H_D_S gives 0.198 while H alone gives 0.249, and for Traffic, H_D gives 0.512 while H gives 0.646. Because the same test labels are used both to select features and to report the 23% averaged improvement, the headline number is the best-tested configuration, not an unbiased estimate for a pre-specified model. Please fix the feature set per dataset (or per horizon) before accessing test labels, or use a hold-out validation split for selection, and report the resulting estimates.
  2. [§4.1, Eq. (2), §5.2.1] TimeSter encodes only historical timestamps U, while the future timestamp features P defined in Section 2 are not used. The fixed linear map in Eq. (2) therefore assumes the calendar patterns (hour, weekday, month, season) in the training window are stationary across the forecast span, and the deliberate omission of a year feature means inter-annual trends or regime shifts cannot be represented. The Weather discussion in §5.2.1 acknowledges the one-year-span problem for seasonality, but the paper should state this as a general limitation of the timestamp branch and scope the generalization claims accordingly; a constructive test would be to evaluate on a train/test year split with a year shift.
  3. [Abstract, §5.1.1, Table 2] The phrase 'consistent state-of-the-art performance across seven real-world datasets' overstates the results. In Table 2, TimeLinear is behind MSD-Mixer on ETTm1 (0.385 vs 0.361) and ETTm2 (0.273 vs 0.271), behind MSD-Mixer on Weather (0.251 vs 0.233), and behind SOFTS on Traffic (0.480 vs 0.409). The more modest abstract wording 'on par with or exceed' is supported, but 'consistent SOTA' is not, and should be revised or substantiated with a statistical comparison.
minor comments (6)
  1. [Tables 2–4, Table 10] The main comparisons report only mean values; Table 10 gives standard deviations for TimeLinear but not for baselines. Add error bars or standard-deviation columns for the key comparisons, or state that the differences are within seed noise.
  2. [Tables 3 and 4] The repeated 'A vg' headings should be 'Avg'.
  3. [§5.2.3] 'our mode' should read 'our model'.
  4. [§5.3.5] The text refers to 'the optimal time-related features' without specifying the selection criterion; please state the performance metric and the split used to define optimality.
  5. [Figure 1(a)] The figure referenced in the introduction does not define the baselines it compares; please add a caption or legend sufficient to interpret the comparison.
  6. [Table 1] Timespans such as '2016.7 - 2018.6' should be given as explicit dates or ranges for clarity.

Circularity Check

1 steps flagged · score 6.0 of 10

Test-set feature selection inflates the headline 23% improvement, though the core forecasting equations are not otherwise circular.

  1. fitted input called prediction [Section 5.2.1 (Table 5), Section 5.3.4 (Table 8), and Section 5.3.5 (Table 9); used in Tables 2-4]
    "Due to the differences in granularity and temporal characteristics of datasets and the prediction time span, the optimal time-related features are different for different datasets and prediction lengths. In Table 9, we illustrate the optimal time-related features of each dataset and prediction length under TimeLinear."

    Table 9's per-dataset/per-horizon feature sets coincide with the test-best rows of the ablations in Tables 5 and 8: at T=720, Electricity's selected H_D_S gives 0.198 (vs 0.249 with H alone), Traffic's selected H_D gives 0.512 (vs 0.646 with H), ETTh2's H_M_S gives 0.377, and ETTh1's H gives 0.464. These exact values are then reported in Tables 3/4 and averaged into the headline 23% MSE reduction on Electricity and Traffic. Because no validation split or selection rule is described, the feature choice is fitted to the test labels; the claimed improvement is the best over tested configurations rather than an unbiased estimate for a fixed, pre-specified model. The reported gain is therefore partially forced by the selection procedure.

full rationale

The model equations themselves are not circular: Eq. (2) is a learned linear decoder on encoded timestamp features, Eq. (4) is a linear backbone on observations, and Eq. (5) is a convex combination; none of these is defined in terms of the target labels or restates a fitted value as a prediction. The ACF analysis in Section 5.3.3 is post hoc and is not used to derive the equations, and there is no load-bearing chain of self-citations or imported uniqueness theorem. The one substantial circularity-like step is the selection of 'optimal' time-related features per dataset and horizon: the set is chosen from test-set ablations and then the resulting test MSE is presented as the model's improvement. This makes the headline improvement a selection-maximized quantity rather than a prediction from a fixed protocol. Since this selection directly affects the central quantitative claim (the 23% average reduction on Electricity and Traffic), the score is elevated to 6, while recognizing that the underlying architecture and most other comparisons retain independent empirical content.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central empirical claim rests on several tuned choices: the blend weight beta, the per-dataset feature set, the convolution kernel size, and the hidden layer size. The domain assumptions are stationarity of calendar effects and the sufficiency of a linear mapping from historical timestamp encodings.

free parameters (4)
  • beta (trade-off coefficient) = not given in text
    Weights TimeSter and backbone outputs in Eq. (5); sensitivity is shown in Figure 9, but the exact values used per dataset are not reported.
  • time-related feature subset = H, H_D, H_D_M, H_M_S, etc. per dataset and horizon (Table 9)
    The feature combination is chosen separately for each dataset and prediction length; the selection protocol (validation versus test) is not specified.
  • convolution kernel size (ksize) = not reported
    A hyperparameter of the TimeSter encoder; sensitivity is analyzed in Figure 9, but the chosen values are not listed.
  • hidden layer reduction rate = not reported
    Controls the first hidden layer size of TimeSter; chosen per dataset, exact values are not given.
assumptions (4)
  • domain assumption Observations at a time point are approximately determined by a combination of calendar timestamps plus noise.
    Used in Section 1 and Eq. (1), where X_U is modeled as q_theta(X_U|U); the entire TimeSter branch depends on this conditional structure.
  • domain assumption Calendar effects are stationary across years, so the omitted year feature is not needed.
    Section 2 defines timestamps without year, and Section 6 acknowledges that future years are unknown; inter-annual trends are therefore not modeled.
  • domain assumption A linear decoder W can map the L historical time-related observations to the T future values.
    Eq. (2) uses Y_U = W X_U + b without consuming the future timestamp features P; W must extrapolate calendar phases beyond the input window.
  • domain assumption The final denormalization in Eq. (6) is valid for the sum of two differently sourced predictions.
    The backbone output comes from normalized observations while the TimeSter output is produced from timestamps; the model learns scales during training, but the additive form is assumed stable.

how reviews work

0 comments
Cite this review

Pith. "Pith review of How Much Can Time-related Features Enhance Time Series Forecasting?." pith.science (2026). https://pith.science/paper/O57UAMKS

@misc{pith2026241201557,
  author       = {Pith},
  title        = {Pith review of: How Much Can Time-related Features Enhance Time Series Forecasting?},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O57UAMKS}},
  note         = {Machine review of arXiv:2412.01557}
}
read the original abstract

Recent advancements in long-term time series forecasting (LTSF) have primarily focused on capturing cross-time and cross-variate (channel) dependencies within historical data. However, a critical aspect often overlooked by many existing methods is the explicit incorporation of \textbf{time-related features} (e.g., season, month, day of the week, hour, minute), which are essential components of time series data. The absence of this explicit time-related encoding limits the ability of current models to capture cyclical or seasonal trends and long-term dependencies, especially with limited historical input. To address this gap, we introduce a simple yet highly efficient module designed to encode time-related features, Time Stamp Forecaster (TimeSter), thereby enhancing the backbone's forecasting performance. By integrating TimeSter with a linear backbone, our model, TimeLinear, significantly improves the performance of a single linear projector, reducing MSE by an average of 23\% on benchmark datasets such as Electricity and Traffic. Notably, TimeLinear achieves these gains while maintaining exceptional computational efficiency, delivering results that are on par with or exceed state-of-the-art models, despite using a fraction of the parameters.

Figures

Figures reproduced from arXiv: 2412.01557 by the authors.

Figure 1
Figure 1. (a) Comparison of forecasting using only time [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overview of the proposed method. (a) The TimeSter [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Distribution of training data across datasets at noon [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Accuracy, parameter count, and training time of [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Performance promotion with longer historical win [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Data distribution of the Traffic dataset at noon [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: ACF of Electricity and Traffic datasets. The variable [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: ACF of ETTm1, ETTm2, and Weather. Granularity [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Hyperparameter sensitivity of trade-off coefficient [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: Prediction results of three datasets. 𝐿 is 96 and 𝑇 is 336. From left to right, the result of TimeLinear, the result of TimeSter, and the result of BonSter, respectively. 5.3.7 Hyperparameter Sensitivity. In addition to the basic learning rate, batch size, etc., vario…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 39 canonical work pages

  1. [1]

    Maddix, Hao Wang, Michael W

    Abdul Fatir Ansari, Lorenzo Stella, Caner Turkmen, Xiyuan Zhang, Pedro Mercado, Huibin Shen, Oleksandr Shchur, Syama Sundar Rangapuram, Sebas- tian Pineda Arango, Shubham Kapoor, Jasper Zschiegner, Danielle C. Maddix, Hao Wang, Michael W. Mahoney, Kari Torkkola, Andrew Gordon Wilson, Michael Bohlke-Schneider, and Yuyang Wang. 2024. Chronos: Learning the L...

    arXiv 2024

  2. [2]

    Parikshit Bansal, Prathamesh Deshpande, and Sunita Sarawagi. 2021. Missing Value Imputation on Multidimensional Time Series. Proceedings of the VLDB Endowment 14, 11 (2021), 2533–2545

    work page 2021

  3. [3]

    Joos-Hendrik Böse, Valentin Flunkert, Jan Gasthaus, Tim Januschowski, Dustin Lange, David Salinas, Sebastian Schelter, Matthias Seeger, and Yuyang Wang

    work page

  4. [4]

    George EP Box and Gwilym M Jenkins. 1968. Some Recent Advances in Fore- casting and Control. Journal of the Royal Statistical Society. Series C (Applied Statistics) 17, 2 (1968), 91–109

    work page 1968

  5. [5]

    Shengchao Chen, Guodong Long, Tao Shen, Jing Jiang, and Chengqi Zhang

    work page

  6. [6]

    Yue Cui, Kai Zheng, Dingshan Cui, Jiandong Xie, Liwei Deng, Feiteng Huang, and Xiaofang Zhou. 2021. METRO: A Generic Graph Neural Network Framework for Multivariate Time Series Forecasting. Proceedings of the VLDB Endowment 15, 2 (2021), 224–236

    work page 2021

  7. [7]

    Abhimanyu Das, Weihao Kong, Andrew Leach, Shaan Mathur, Rajat Sen, and Rose Yu. 2024. Long-term Forecasting with TiDE: Time-series Dense Encoder. arXiv:2304.08424 [stat.ML]

    arXiv 2024

  8. [8]

    Ziquan Fang, Lu Pan, Lu Chen, Yuntao Du, and Yunjun Gao. 2021. MDTP: A Multi-source Deep Trafic Prediction Framework over Spatio-Temporal Trajectory Data. Proceedings of the VLDB Endowment 14, 8 (2021), 1289–1297

    work page 2021

Show all 51 references
  1. [9]

    Mononito Goswami, Konrad Szafer, Arjun Choudhry, Yifu Cai, Shuo Li, and Artur Dubrawski. 2024. MOMENT: A Family of Open Time-series Foundation Models. In International Conference on Machine Learning

    2024

  2. [10]

    Lu Han, Xu-Yang Chen, Han-Jia Ye, and De-Chuan Zhan. 2024. SOFTS: Efficient Multivariate Time Series Forecasting with Series-Core Fusion. In Advances in Neural Information Processing Systems

    2024

  3. [11]

    Lu Han, Han-Jia Ye, and De-Chuan Zhan. 2024. The Capacity and Robustness Trade-off: Revisiting the Channel Independent Strategy for Multivariate Time Series Forecasting. IEEE Transactions on Knowledge and Data Engineering 36, 11 (2024), 7129–7142

    2024

  4. [12]

    Ming Jin, Shiyu Wang, Lintao Ma, Zhixuan Chu, James Y Zhang, Xiaoming Shi, Pin-Yu Chen, Yuxuan Liang, Yuan-Fang Li, Shirui Pan, et al . 2024. Time- LLM: Time Series Forecasting by Reprogramming Large Language Models. In International Conference on Learning Representations

    2024

  5. [13]

    Mourad Khayati, Alberto Lerner, Zakhar Tymchenko, and Philippe Cudré- Mauroux. 2020. Mind the Gap: An Experimental Evaluation of Imputation of Missing Values Techniques in Time Series. Proceedings of the VLDB Endowment 13, 5 (2020), 768–782

    2020

  6. [14]

    Taesung Kim, Jinhee Kim, Yunwon Tae, Cheonbok Park, Jang-Ho Choi, and Jaegul Choo. 2021. Reversible Instance Normalization for Accurate Time-Series Forecasting against Distribution Shift. In International Conference on Learning Representations

    2021

  7. [15]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. 2017. Adam: A Method for Stochastic Opti- mization. arXiv:1412.6980 [cs.LG]

    2017 arXiv

  8. [16]

    Diederik P Kingma and Max Welling. 2014. Auto-Encoding Variational Bayes. In International Conference on Learning Representations

    2014

  9. [17]

    Guokun Lai, Wei-Cheng Chang, Yiming Yang, and Hanxiao Liu. 2018. Model- ing Long- and Short-Term Temporal Patterns with Deep Neural Networks. In International ACM SIGIR Conference on Research and Development in Information Retrieval

    2018

  10. [18]

    Zhe Li, Shiyi Qi, Yiduo Li, and Zenglin Xu. 2023. Revisiting Long-term Time Series Forecasting: An Investigation on Linear Mapping. arXiv:2305.10721 [cs.LG]

    2023 arXiv

  11. [19]

    Shengsheng Lin, Weiwei Lin, Xinyi Hu, Wentai Wu, Ruichao Mo, and Haocheng Zhong. 2024. CycleNet: Enhancing Time Series Forecasting through Modeling Periodic Patterns. In Advances in Neural Information Processing Systems

    2024

  12. [20]

    Yong Liu, Tengge Hu, Haoran Zhang, Haixu Wu, Shiyu Wang, Lintao Ma, and Mingsheng Long. 2024. iTransformer: Inverted Transformers Are Effective for Time Series Forecasting. In International Conference on Learning Representations

    2024

  13. [21]

    Yong Liu, Guo Qin, Xiangdong Huang, Jianmin Wang, and Mingsheng Long

  14. [22]

    Donghao Luo and Xue Wang. 2024. ModernTCN: A Modern Pure Convolu- tion Structure for General Time Series Analysis. In International Conference on Learning Representations

    2024

  15. [23]

    Yisheng Lv, Yanjie Duan, Wenwen Kang, Zhengxi Li, and Fei-Yue Wang. 2015. Traffic Flow Prediction With Big Data: A Deep Learning Approach. IEEE Trans- actions on Intelligent Transportation Systems 16, 2 (2015), 865–873

    2015

  16. [24]

    In Advances in Neural Information Processing Systems

    AutoTimes: Autoregressive Time Series Forecasters via Large Language Models. In Advances in Neural Information Processing Systems

  17. [25]

    Sai Shankar Narasimhan, Shubhankar Agarwal, Oguzhan Akcin, Sujay Sang- havi, and Sandeep Chinchali. 2024. Time Weaver: A Conditional Time Series Generation Model. In International Conference on Machine Learning

    2024

  18. [26]

    Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam

    Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. 2023. A Time Series is Worth 64 Words: Long-term Forecasting with Transformers. In International Conference on Learning Representations

    2023

  19. [27]

    Henrik Madsen. 2007. Time Series Analysis. Chapman and Hall/CRC

    2007

  20. [28]

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. 2019. PyTorch: An Imperative Style, High-Performance Deep Learning Library. In Advances in Neural Information Processing Systems

    2019

  21. [29]

    Luis Manuel Pereira, Addisson Salazar, and Luis Vergara. 2023. A Comparative Analysis of Early and Late Fusion for the Multimodal Two-Class Problem. IEEE Access 11 (2023), 84283–84300

    2023

  22. [30]

    Tsay, Themis Palpanas, and Michael J

    John Paparrizos, Yuhao Kang, Paul Boniol, Ruey S. Tsay, Themis Palpanas, and Michael J. Franklin. 2022. TSB-UAD: An End-to-End Benchmark Suite for Uni- variate Time-Series Anomaly Detection. Proceedings of the VLDB Endowment 15, 8 (2022), 1697–1711

    2022

  23. [31]

    Sebastian Schmidl, Phillip Wenig, and Thorsten Papenbrock. 2022. Anomaly Detection in Time Series: A Comprehensive Evaluation. Proceedings of the VLDB Endowment 15, 9 (2022), 1779–1797

    2022

  24. [32]

    Zhuangwei Shi. 2024. MambaStock: Selective state space model for stock predic- tion. arXiv:2402.18959 [cs.CE]

    2024 arXiv

  25. [33]

    Jensen, Zhenli Sheng, and Bin Yang

    Xiangfei Qiu, Jilin Hu, Lekui Zhou, Xingjian Wu, Junyang Du, Buang Zhang, Chenjuan Guo, Aoying Zhou, Christian S. Jensen, Zhenli Sheng, and Bin Yang

  26. [34]

    Proceedings of the VLDB Endowment 17, 9 (2024), 2363– 2377

    TFB: Towards Comprehensive and Fair Benchmarking of Time Series Forecasting Methods. Proceedings of the VLDB Endowment 17, 9 (2024), 2363– 2377

    2024

  27. [35]

    Huiqiang Wang, Jian Peng, Feihu Huang, Jince Wang, Junhui Chen, and Yifei Xiao. 2023. MICN: Multi-scale Local and Global Context Modeling for Long-term Series Forecasting. In International Conference on Learning Representations

    2023

  28. [36]

    Shiyu Wang, Haixu Wu, Xiaoming Shi, Tengge Hu, Huakun Luo, Lintao Ma, James Y Zhang, and JUN ZHOU. 2024. TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting. In International Conference on Learning Representations

    2024

  29. [37]

    Mun, Matthew Lim, Jonah Yamato, Nathan Huh, and Cyrus Shahabi

    Luan Tran, Min Y. Mun, Matthew Lim, Jonah Yamato, Nathan Huh, and Cyrus Shahabi. 2020. DeepTRANS: A Deep Learning System for Public Bus Travel Time Estimation using Traffic Forecasting. Proceedings of the VLDB Endowment 13, 12 (2020), 2957–2960

    2020

  30. [38]

    Chengsen Wang, Qi Qi, Jingyu Wang, Haifeng Sun, Zirui Zhuang, Jinming Wu, and Jianxin Liao. 2024. Rethinking the Power of Timestamps for Robust Time Series Forecasting: A Global-Local Fusion Perspective. In Advances in Neural Information Processing Systems

    2024

  31. [39]

    Haixu Wu, Tengge Hu, Yong Liu, Hang Zhou, Jianmin Wang, and Mingsheng Long. 2023. TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis. In International Conference on Learning Representations

    2023

  32. [40]

    Haixu Wu, Jiehui Xu, Jianmin Wang, and Mingsheng Long. 2021. Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Fore- casting. In Advances in Neural Information Processing Systems

    2021

  33. [41]

    Xinlei Wang, Maike Feng, Jing Qiu, Jinjin Gu, and Junhua Zhao. 2024. From News to Forecast: Integrating Event Analysis in LLM-Based Time Series Forecasting with Reflection. In Advances in Neural Information Processing Systems

    2024

  34. [42]

    Phillip Wenig, Sebastian Schmidl, and Thorsten Papenbrock. 2022. TimeEval: A Benchmarking Toolkit for Time Series Anomaly Detection Algorithms. Proceed- ings of the VLDB Endowment 15, 12 (2022), 3678–3681

    2022

  35. [43]

    Ailing Zeng, Muxi Chen, Lei Zhang, and Qiang Xu. 2023. Are Transformers Effective for Time Series Forecasting?. In Proceedings of the AAAI Conference on Artificial Intelligence

    2023

  36. [44]

    Yuchen Zhang, Mingsheng Long, Kaiyuan Chen, Lanxiang Xing, Ronghua Jin, Michael I Jordan, and Jianmin Wang. 2023. Skilful Nowcasting of Extreme Precipitation with NowcastNet. Nature 619, 7970 (2023), 526–532

    2023

  37. [45]

    Yumo Xu and Shay B Cohen. 2018. Stock Movement Prediction from Tweets and Historical Prices. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

    2018

  38. [46]

    Zhijian Xu, Ailing Zeng, and Qiang Xu. 2024. FITS: Modeling Time Series with 10𝑘 Parameters. In International Conference on Learning Representations

    2024

  39. [47]

    Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang. 2021. Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting. In Proceedings of the AAAI Conference on Artificial Intelligence

    2021

  40. [49]

    Yunhao Zhang and Junchi Yan. 2023. Crossformer: Transformer Utilizing Cross- Dimension Dependency for Multivariate Time Series Forecasting. InInternational Conference on Learning Representations

    2023

  41. [50]

    Gary Chan

    Shuhan Zhong, Sizhe Song, Weipeng Zhuo, Guanyao Li, Yang Liu, and S.-H. Gary Chan. 2024. A Multi-Scale Decomposition MLP-Mixer for Time Series Analysis. Proceedings of the VLDB Endowment 17, 7 (2024), 1723–1736

    2024

  42. [2017]

    Proceedings of the VLDB Endowment 10, 12 (2017), 1694–1705

    Probabilistic Demand Forecasting at Scale. Proceedings of the VLDB Endowment 10, 12 (2017), 1694–1705

    2017

  43. [2024]

    arXiv:2305.14244 [cs.LG]

    Federated Prompt Learning for Weather Foundation Models on Devices. arXiv:2305.14244 [cs.LG]

    arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.