Long-Term Time Series Forecasting

Overview

Long-term time series forecasting uses past observations to predict many steps into the future, supporting planning when decisions must be made well ahead of time. Methods range from trend and seasonal decomposition, linear models, and patch-based mixers to Transformers, state-space models, and graph networks for linked locations or variables. Some approaches add physical dynamics, external signals, or retrieval of similar historical cases, while probabilistic models represent uncertainty as well as a single forecast. A central challenge is keeping forecasts stable under distribution shifts and accumulated errors; strong pointwise accuracy can still come at the cost of realistic uncertainty, structural fidelity, or efficiency.

Latest papers

CardsList

124 papers8 papers in the last four weeks, down 67% on the four weeks before. 0.1% of all new papers.

  1. Time-o1: Time-Series Forecasting Needs Transformed Label Alignment

    May 23, 2025Hao Wang, Licheng Pan, Zhichao Chen +5Long-Term Time Series ForecastingTime Series Forecasting

  2. Human Supervisor Workload Prediction: Lag Horizon Selection

    May 21, 2025Mark-Robin Giolando, Julie A. AdamsMultivariate Time Series ForecastingLong-Term Time Series Forecasting

  3. BEAT: Balanced Frequency Adaptive Tuning for Long-Term Time-Series Forecasting

    Jan 31, 2025Zhixuan Li, Naipeng Chen, Seonghwa Choi +2Long-Term Time Series ForecastingFrequency-Domain Time Series Modeling

  4. Spatio-Temporal Partial Sensing Forecast for Long-term Traffic

    Aug 2, 2024Zibo Liu, Zhe Jiang, Zelin Xu +5Long-Term Time Series ForecastingSpatiotemporal Forecasting