Long-term forecasting: research map
429 papers on Long-term forecasting in Time series, from ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR (2016–2026), grouped into 2 clusters and 5 approaches. The busiest year so far is 2026.
What it is
Predicting a time series far into the future, such as electricity load, traffic or weather. Transformers, simple linear models and graph models compete here; a surprising finding was that simple models often match complex ones.
Terms you will see on this page:
- Forecast horizon
- how far ahead the model predicts.
- Multivariate forecasting
- predicting many related series together.
- Spatiotemporal
- data indexed by both location and time, as in traffic sensors.
Start reading with the most cited papers here:
- Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting (AAAI 2021 · 7,037 citations)
- Are Transformers Effective for Time Series Forecasting? (AAAI 2023 · 3,150 citations)
Within Time series, its share grew from 27.2% in 2023–24 to 31.0% in 2025–26 (103 → 244 papers at ICML, NeurIPS, ICLR, CVPR and AAAI, the venues with data for all four years).
Explore Long-term forecasting in the interactive map
Working on something in this topic? Describe your idea in scime atlas to see which approach it falls under, the closest papers by meaning and how crowded the spot has become.
What changed recently
Approaches whose share of the topic moved most between 2023–24 and 2025–26 (ICML, NeurIPS, ICLR, CVPR and AAAI):
- Attention for time series forecasting rising: 25% → 35% of the topic (26 → 85 papers)
- Lightweight long-term forecasters rising: 22% → 23% of the topic (23 → 57 papers)
- Transformers for long-term forecasting falling: 14% → 4% of the topic (14 → 9 papers)
Papers from 2025–2026 that are already cited most:
- WPMixer: Efficient Multi-Resolution Mixing for Long-Term Time Series Forecasting (AAAI 2025 · 77 citations)
- Unlocking the Power of Patch: Patch-Based MLP for Long-Term Time Series Forecasting (AAAI 2025 · 41 citations)
- RhythmMamba: Fast, Lightweight, and Accurate Remote Physiological Measurement (AAAI 2025 · 18 citations)
- Motif-aware Graph Neural Networks for Networked Time Series Imputation (AAAI 2025 · 5 citations)
Approaches and key papers
Multivariate and spatiotemporal forecasting · 306 papers
Keywords: graph, multivariate time, spatial
Approaches in this cluster:
- Attention for time series forecasting (127 papers)
Forecast long series with spectral, multi-scale and adaptive transformers and normalization. - Spatiotemporal forecasting models (87 papers)
Model time series with spatiotemporal graphs, convolutions, state-space and diffusion pretraining. - Graph-based multivariate forecasting (92 papers)
Forecast multivariate series by learning channel dependence with graph neural networks.
Most cited and most cited since 2024:
- Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting (AAAI 2021 · 7,037 citations)
- Enhancing the Locality and Breaking the Memory Bottleneck of Transformer on Time Series Forecasting (NeurIPS 2019 · 1,086 citations)
- iTransformer: Inverted Transformers Are Effective for Time Series Forecasting (ICLR 2024 · 385 citations)
- MSGNet: Learning Multi-Scale Inter-series Correlations for Multivariate Time Series Forecasting (AAAI 2024 · 234 citations)
Long-term forecasting models · 123 papers
Keywords: long term, term time, term forecasting
Approaches in this cluster:
- Lightweight long-term forecasters (98 papers)
Forecast long horizons with lean models exploiting periodicity, memory and mixing. - Transformers for long-term forecasting (25 papers)
Adapt transformers with patching, decomposition and sharpness-aware training for long-term forecasting.
Most cited and most cited since 2024:
- Are Transformers Effective for Time Series Forecasting? (AAAI 2023 · 3,150 citations)
- A Time Series is Worth 64 Words: Long-term Forecasting with Transformers (ICLR 2023 · 554 citations)
- TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting (ICLR 2024 · 93 citations)
- WPMixer: Efficient Multi-Resolution Mixing for Long-Term Time Series Forecasting (AAAI 2025 · 77 citations)
