SPTTE: A Spatiotemporal Probabilistic Framework for Travel Time Estimation

TL;DR

SPTTE is a spatiotemporal probabilistic framework that models the evolving joint distribution of multi-trip travel times by formulating the estimation task as a spatiotemporal stochastic process regression problem with fragmented observations and employs a prior-based heterogeneity smoothing strategy to correct unreliable learning caused by unevenly distributed trips.

Abstract

Accurate travel time estimation is essential for navigation and itinerary planning. While existing research employs probabilistic modeling to assess travel time uncertainty and account for correlations between multiple trips, modeling the temporal variability of multi-trip travel time distributions remains a significant challenge. Capturing the evolution of joint distributions requires large, well-organized datasets; however, real-world trip data are often temporally sparse and spatially unevenly distributed. To address this issue, we propose SPTTE, a spatiotemporal probabilistic framework that models the evolving joint distribution of multi-trip travel times by formulating the estimation task as a spatiotemporal stochastic process regression problem with fragmented observations. SPTTE incorporates an RNN-based temporal Gaussian process parameterization to regularize sparse observations and capture temporal dependencies. Additionally, it employs a prior-based heterogeneity smoothing strategy to correct unreliable learning caused by unevenly distributed trips, effectively modeling temporal variability under sparse and uneven data distributions. Evaluations on real-world datasets demonstrate that SPTTE outperforms state-of-the-art deterministic and probabilistic methods by over 10.13%. Ablation studies and visualizations further confirm the effectiveness of the model components.

Figures (9)

• Figure 1: Fragmented realization of the trip spatiotemporal stochastic process. The space is simplified to one dimension for illustration. Each curve represents a realization of the trip stochastic process, with the solid line indicating the observed fragment and the dashed line representing the unobserved fragment.
• Figure 2: Overall architecture of SPTTE.
• Figure 3: Process of Heterogeneity Smoothing. Introduce prior intervention based on heterogeneous trip coverage frequency to asymmetrically smooth link representations. Dashed lines represent smoothing weights, with red/blue indicating high/low weights.
• Figure 4: Estimation error by time period in a day. (a). MAPE in Chengdu dataset. (b). MAPE in Harbin dataset.
• Figure 5: Coverage frequency distribution and heterogeneous weight function.