Time series forecasting and spatiotemporal kriging are the two most important tasks in data analysis. Recent research on graph neural networks has made substantial progress time forecasting, while little attention been paid to problem---recovering signals for unsampled locations/sensors. Most existing scalable methods (e.g., matrix/tensor completion) transductive, thus full retraining is required when we have a new sensor interpolate. In this paper, develop an Inductive Graph Neural Network...

This paper studies the traffic state estimation (TSE) problem using sparse observations from mobile sensors. Most existing TSE methods either rely on well-defined physical flow models or require large amounts of simulation data as input to train learning algorithms. Different previous studies, in this we propose a purely data-driven and model-free solution. We consider spatiotemporal matrix completion/interpolation apply delay embedding transform original incomplete into fourth-order Hankel...

Recent studies have significantly improved the prediction accuracy of travel demand using graph neural networks. However, these largely ignored uncertainty that inevitably exists in prediction. To fill this gap, study proposes a framework probabilistic networks (Prob-GNN) to quantify spatiotemporal demand. This Prob-GNN is substantiated by deterministic and assumptions, empirically applied task predicting transit ridesharing Chicago. We found assumptions (e.g. distribution tail, support)...

Origin-Destination (O-D) travel demand prediction is a fundamental challenge in transportation. Recently, spatial-temporal deep learning models demonstrate the tremendous potential to enhance accuracy. However, few studies tackled uncertainty and sparsity issues fine-grained O-D matrices. This presents serious problem, because vast number of zeros deviate from Gaussian assumption underlying deterministic models. To address this issue, we design Spatial-Temporal Zero-Inflated Negative...

Effective traffic prediction is a cornerstone of intelligent transportation systems, enabling precise forecasts flow, speed, and congestion. While traditional spatio-temporal graph neural networks (ST-GNNs) have achieved notable success in short-term forecasting, their performance long-term predictions remains limited. This challenge arises from over-squashing problem, where bottlenecks limited receptive fields restrict information flow hinder the modeling global dependencies. To address...

Urban prediction tasks, such as forecasting traffic flow, temperature, and crime rates, are crucial for efficient urban planning management. However, existing Spatiotemporal Graph Neural Networks (ST-GNNs) often rely solely on accuracy, overlooking spatial demographic disparities in their predictions. This oversight can lead to imbalanced resource allocation exacerbate inequities areas. study introduces a Residual-Aware Attention (RAA) Block an equality-enhancing loss function address these...

As a typical problem of Spatiotemporal Resource Management, Time Series Supplier Allocation (TSSA) poses complex NP-hard challenge, aimed at refining future order dispatching strategies to satisfy the trade-off between demands and maximum supply. The Black-Litterman (BL) model, which comes from financial portfolio management, offers new perspective for TSSA by balancing expected returns against insufficient supply risks. However, BL model is not only constrained manually constructed matrices...

Understanding Origin-Destination (O-D) travel demand is crucial for transportation management. However, traditional spatial-temporal deep learning models grapple with addressing the sparse and long-tail characteristics in high-resolution O-D matrices quantifying prediction uncertainty. This dilemma arises from numerous zeros over-dispersed patterns within these matrices, which challenge Gaussian assumption inherent to deterministic models. To address challenges, we propose a novel approach:...

Short-term demand forecasting for on-demand ride-hailing services is a fundamental issue in intelligent transportation systems. However, previous research predominantly focused on improving prediction accuracy, ignoring fairness issues such as systematic underestimations of travel disadvantaged neighborhoods. This study investigates how to measure, evaluate, and enhance between privileged communities spatial-temporal services. We developed socially-aware neural network (SA-Net) that...

Traffic data serves as a fundamental component in both research and applications within intelligent transportation systems. However, real-world data, collected from loop detectors or similar sources, often contains missing values (MVs), which can adversely impact associated research. Instead of discarding this incomplete researchers have sought to recover these through numerical statistics, tensor decomposition, deep learning techniques. In paper, we propose an innovative approach for...

The rapid growth of the ride-hailing industry has revolutionized urban transportation worldwide. Despite its benefits, equity concerns arise as underserved communities face limited accessibility to affordable services. A key issue in this context is vehicle rebalancing problem, where idle vehicles are moved areas with anticipated demand. Without equitable approaches demand forecasting and strategies, these practices can further deepen existing inequities. In realm ride-hailing, three main...

As a sustainable transportation alternative, cycling has been developed by many large cities as one essential measure for addressing the "last mile problem" in urban areas with high mobility demand. Developing safe and friendly lane network become an urgent task governments, especially Chinese encountering rapid increase use of "dockless" shared bikes, such Mobike Ofo. The emergence kind app-driven dockless bike sharing systems results fast growth demand, leads to gap between growing demand...

Spatiotemporal kriging is an important application in spatiotemporal data analysis, aiming to recover/interpolate signals for unsampled/unobserved locations based on observed signals. The principle challenge how effectively model and leverage the dependencies within data. Recently, graph neural networks (GNNs) have shown great promise tasks. However, standard GNNs often require a carefully designed adjacency matrix specific aggregation functions, which are inflexible general...

Predicting traffic incident risks at granular spatiotemporal levels is challenging. The datasets predominantly feature zero values, indicating no incidents, with sporadic high-risk values for severe incidents. Notably, a majority of current models, especially deep learning methods, focus solely on estimating risk overlooking the uncertainties arising from inherently unpredictable nature To tackle this challenge, we introduce Spatiotemporal Zero-Inflated Tweedie Graph Neural Networks...

Recent studies have demonstrated the great success of graph convolutional networks in short-term traffic forecasting (e.g., 15–30 min ahead) tasks by capturing dependencies road network structure. Based on these models, long-term can be achieved two approaches: (1) recursively generating a one-step-ahead prediction and (2) adapting models to sequence-to-sequence (seq2seq) learning. However, practice, approaches often show poor performance tasks. The recursive approach suffers from error...

Mode share analysis is important to various real-world transportation tasks as it helps researchers understand the travel behaviors and choices of passengers. A typical example prediction communities' mode by accounting for their sociodemographics like age, income, etc., modes' attributes (e.g. cost time). However, there exist only limited efforts in integrating structure urban built environment, e.g., road networks, into models capture impacts environment. This task usually requires manual...

Time series forecasting and spatiotemporal kriging are the two most important tasks in data analysis. Recent research on graph neural networks has made substantial progress time forecasting, while little attention been paid to problem -- recovering signals for unsampled locations/sensors. Most existing scalable methods (e.g., matrix/tensor completion) transductive, thus full retraining is required when we have a new sensor interpolate. In this paper, develop an Inductive Graph Neural Network...

Recent studies have significantly improved the prediction accuracy of travel demand using graph neural networks. However, these largely ignored uncertainty that inevitably exists in prediction. To fill this gap, study proposes a framework probabilistic networks (Prob-GNN) to quantify spatiotemporal demand. This Prob-GNN is substantiated by deterministic and assumptions, empirically applied task predicting transit ridesharing Chicago. We found assumptions (e.g. distribution tail, support)...

This paper studies the traffic state estimation (TSE) problem using sparse observations from mobile sensors. Most existing TSE methods either rely on well-defined physical flow models or require large amounts of simulation data as input to train machine learning models. Different previous studies, we propose a purely data-driven and model-free solution in this paper. We consider spatiotemporal matrix completion/interpolation problem, apply delay embedding transform original incomplete into...

We introduce the BL model and Perspective Matrix to optimize supplier selection order allocation, focusing on both temporal spatial dynamics. Our development of a Supplier Relationship Network, using Spatio-Temporal Graph Neural enhances understanding complex interdependencies. Additionally, we address credibility issues in zero-order scenarios with Masked Ranking Mechanism, improving ranking efficiency. demonstrates superior results two datasets compared traditional models. evaluations...

Transportation mode share analysis is important to various real-world transportation tasks as it helps researchers understand the travel behaviors and choices of passengers. A typical example prediction communities' by accounting for their sociodemographics like age, income, etc., modes' attributes (e.g. cost time). However, there exist only limited efforts in integrating structure urban built environment, e.g., road networks, into models capture impacts environment. This task usually...

Quantifying uncertainty is crucial for robust and reliable predictions. However, existing spatiotemporal deep learning mostly focuses on deterministic prediction, overlooking the inherent in such prediction. Particularly, highly-granular datasets are often sparse, posing extra challenges prediction quantification. To address these issues, this paper introduces a novel post-hoc Sparsity-awar Uncertainty Calibration (SAUC) framework, which calibrates both zero non-zero values. develop SAUC, we...