Probabilistic Time Series Forecasting

Follow this topic Ask about this topic

Probabilistic time series forecasting studies not only what may happen next, but how uncertain that prediction is, often by producing quantiles, intervals, or complete future scenarios. Approaches range from Bayesian and statistical models, quantile regression, and conformal calibration to ensembles, diffusion and flow-based generators, and pretrained time-series foundation models. These methods are evaluated for accuracy, calibration, sharpness, and computational cost across settings such as energy, finance, health, and traffic, with increasing attention to multivariate dependence, irregular observations, and distribution shifts. A central tension is that better point accuracy does not guarantee reliable uncertainty: performance can vary across horizons and regimes, and foundation models do not consistently outperform simpler task-specific methods.

Latest papers

162 papers23 papers in the last four weeks, down 15% on the four weeks before. 0.1% of all new papers.

Time Series Foundation Models as Strong Baselines in Transportation Forecasting: A Large-Scale Benchmark Analysis
Feb 27, 2026Javier Yanes-Pulido, Filipe RodriguesTime-Series Foundation Models Zero-Shot Time Series Forecasting
Technical University of Denmark, Kgs. Lyngby, Denmark
Switch-Hurdle: A MoE Encoder with AR Hurdle Decoder for Intermittent Demand Forecasting
Feb 26, 2026Fabian Muşat, Simona CăbuzTime Series Forecasting Probabilistic Time Series Forecasting
eMAG, Romania
Probabilistic Wind Power Forecasting with Tree-Based Machine Learning and Weather Ensembles
Feb 13, 2026Max Bruninx, Diederik van Binsbergen, Timothy Verstraeten +2Gradient-Boosted Decision Trees Ensemble Learning
Vrije Universiteit Brussel, Brussels, Belgium · Norwegian University of Science and Technology, Trondheim, Norway
Is Flow Matching Just Trajectory Replay for Sequential Data?
Feb 9, 2026Soon Hoe Lim, Shizheng Lin, Michael W. Mahoney +1Flow Matching Time Series Generation
Department of Mathematics, KTH Royal Institute of Technology · Nordita, KTH Royal Institute of Technology and Stockholm University · Department of Statistics, University of California at Berkeley +2
DiTS: Multimodal Diffusion Transformers Are Time Series Forecasters
Feb 6, 2026Haoran Zhang, Haixuan Liu, Yong Liu +4Exogenous Time Series Forecasting Time Series Forecasting
School of Software, BNRist, Tsinghua University
Probabilistic NDVI Forecasting from Sparse Satellite Time Series and Weather Covariates
Feb 4, 2026Irene Iele, Giulia Romoli, Daniele Molino +4Irregular Time-Series Modeling Multivariate Time Series Forecasting
Unit of Artificial Intelligence and Computer Systems, Universit`a Campus Bio-Medico di Roma, Italy · Department of Diagnostics and Intervention, Biomedical Engineering and Radiation Physics, Ume˚a University, Sweden · Department of Naval, Electrical, Electronics and Telecommunications Engineering, University of Genoa, Italy
EMFusion: Uncertainty-Aware Conditional Diffusion Model for Multivariate Narrow-band Exposure Forecasting
Dec 17, 2025Zijiang Yan, Yixiang Huang, Jianhua Pei +2Multivariate Time Series Forecasting Conditional Diffusion Models
department of Electrical Engineering and Computer Science, York University, Toronto, Canada · School of Electrical and Electronic Engineering, Huazhong University of Science and Technology, Wuhan, China · China Southern Grid Electric Power Dispatch Control Center, Guangzhou, China +3
Auditable Context-Aware HFMD Forecasting with Structured LLM Agents
Nov 28, 2025Joongwon Chae, Runming Wang, Chen Xiong +5Epidemic Forecasting Explainable Time Series Forecasting
Institute of Biomedicine and Health Engineering, Tsinghua Shenzhen International Graduate School (SIGS), Tsinghua University, Shenzhen, China · The First Hospital of Hebei Medical University, Shijiazhuang, China · The Fifth Affiliated Hospital of Wenzhou Medical University, Lishui 323000, China +1
DistDF: Time-Series Forecasting Needs Joint-Distribution Wasserstein Alignment
Oct 28, 2025Hao Wang, Licheng Pan, Yuan Lu +7Time Series Forecasting Optimal Transport
Xiaohongshu Inc. · College of Control Science and Technology, Zhejiang University · College of Computer Science and Technology, Zhejiang University +4
Double-Diffusion: Balancing Speed, Accuracy, and Uncertainty in Probabilistic Forecasting for Urban Sensor Networks
Jun 29, 2025Hanlin Dong, Arian Prabowo, Hao Xue +4Spatiotemporal Forecasting Diffusion Model Inference Acceleration
University of New South Wales Sydney, Australia · University of Maryland College Park, United States · The Hong Kong University of Science and Technology (Guangzhou) Guangzhou, China
Temporal Variational Implicit Neural Representations
Jun 2, 2025Batuhan Koyuncu, Rachael DeVries, Ole Winther +1Irregular Time-Series Modeling Multivariate Time Series Forecasting
Department of Computer Science, Saarland University · Department of Biology, University of Copenhagen · Novo Nordisk A/S +1
Bellman Conformal Inference: Calibrating Prediction Intervals For Time Series
Feb 7, 2024Zitong Yang, Yu Ming Huang, Emmanuel Candès +1Conformal Prediction Probabilistic Time Series Forecasting
Department of Statistics, Stanford University · Stanford Graduate School of Bussiness

Feb 27, 2026J/K move · Enter open · S save

Time Series Foundation Models as Strong Baselines in Transportation Forecasting: A Large-Scale Benchmark Analysis

Javier Yanes-Pulido, Filipe Rodrigues

Technical University of Denmark, Kgs. Lyngby, Denmark

Time-Series Foundation Models Zero-Shot Time Series Forecasting Traffic Forecasting Forecasting Benchmarks Probabilistic Time Series Forecasting

Open paper AskPDF arXiv

Accurate forecasting of transportation dynamics is essential for urban mobility and infrastructure planning. Although recent work has achieved strong performance with deep learning models, these methods typically require dataset-specific training, architecture design and hyper-parameter tuning. This paper evaluates whether general-purpose time-series foundation models can serve as forecasters for transportation tasks by benchmarking the zero-shot performance of the state-of-the-art model, Chronos-2, across ten real-world datasets covering highway traffic volume and flow, urban traffic speed, bike-sharing demand, and electric vehicle charging station data. Under a consistent evaluation protocol, we find that, even without any task-specific fine-tuning, Chronos-2 delivers state-of-the-art or competitive accuracy across most datasets, frequently outperforming classical statistical baselines and specialized deep learning architectures, particularly at longer horizons. Beyond point forecasting, we evaluate its native probabilistic outputs using prediction-interval coverage and sharpness, demonstrating that Chronos-2 also provides useful uncertainty quantification without dataset-specific training. In general, this study supports the adoption of time-series foundation models as a key baseline for transportation forecasting research.