Time Series Foundation Models as Strong Baselines in Transportation Forecasting: A Large-Scale Benchmark Analysis
2026-02-27 • Machine Learning
Machine Learning
AI summaryⓘ
The authors tested a general-purpose time-series model called Chronos-2 to see how well it predicts various transportation data without any special training for each task. They used ten real-world datasets, including traffic flow, bike sharing, and electric vehicle charging data. Even without customizing the model, Chronos-2 matched or beat many specialized methods, especially when making predictions further into the future. The authors also showed that Chronos-2 can estimate how uncertain its predictions are, which is useful for planning. Overall, they suggest that such foundation models can be reliable starting points for transportation forecasting.
time-series forecastingfoundation modelszero-shot learningtransportation datadeep learningprobabilistic forecastingprediction intervalstraffic volumeurban mobilityelectric vehicle charging
Authors
Javier Pulido, Filipe Rodrigues
Abstract
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.