Time-series models vary in predicting crowds at different locations and times
How Good Are Time-Series Foundation Models for Pedestrian Crowd Count Forecasting? A Cross-Dataset Comparative Study
Machine LearningArtificial Intelligence
Summary
Predicting how many people will be in a crowd helps manage busy places safely and efficiently. The authors tested several forecasting methods to see how well they predict pedestrian counts in different situations, like special events or regular city days. They found that simple methods work well when data is limited, while advanced models shine with more data and regular patterns. This shows that the best prediction method depends on the amount and type of data as well as the time period being forecasted.
What this means in practice
- •For urban planners: Improve crowd management strategies by selecting forecasting models best suited to sensor data volume and event types.
- •For transportation system operators: Optimize pedestrian traffic control during special events using simple forecasting models when limited data is available.
Authors
Theivaprakasham Hari, Ziteng Li, Yanan Xin, Winnie Daamen, Serge Hoogendoorn
Abstract
Pedestrian-count forecasting supports pedestrian-oriented Intelligent Transportation Systems (ITS), including crowd monitoring, pedestrian-traffic staffing and routing, and proactive risk mitigation during surges. Recent time-series foundation models (FMs) report strong zero-shot accuracy on heterogeneous forecasting benchmarks, but it remains unclear whether these gains transfer reliably to pedestrian sensing deployments. We benchmark seven univariate forecasting approaches spanning four paradigms: Seasonal Naive, gradient-boosted trees (LightGBM, CatBoost), deep learning models (N-HiTS, PatchTST), and two pretrained FMs (TimesFM, Chronos-2). Experiments cover two complementary regimes: (i) a five-day special event dataset SAIL2025 at 3-minute resolution with limited in-domain history; and (ii) Melbourne pedestrian sensors as a multi-year hourly dataset (2010--2017) with strong seasonality. We compare the MAE and RMSE results per sensor across datasets and multiple forecast horizons. Results show three consistent findings. First, with limited historical data, Seasonal Naive remains a strong baseline for long-horizon forecasting on high-volume sensors, while trained models can degrade when the next day differs substantially from prior days. Second, boosted trees can be competitive on lower-volume sensors but exhibit higher sensitivity on high-volume sensors under event-driven shift. Third, FMs excel in the seasonal and data-rich regime under long-context configuration. The findings highlight the importance of choosing pedestrian forecasting models based on both the underlying data conditions and the forecasting horizon.