Although numerous models have been proposed to predict the intensity of human activities in urban areas, two major issues hamper the performance of existing models: (1) fail to incorporate appropriate prior knowledge instrumental for improving accuracy and interpretability; (2) fail to integrate probabilistic and deterministic predictions to achieve complementary strengths, namely uncertainty quantification and high predictive accuracy. To address these challenges, we proposed a prior-enhanced dual-mode spatiotemporal graph neural network (PED-STGNN) to support both probabilistic and deterministic predictions. Specifically, we introduced a hypergraph node-to-vector (hypernode2vec) method to capture the multivariate functional similarity prior derived from complex and multivariate relations between urban regions. This functional similarity characterizes urban systems more precisely than existing methods relying on first-order pairwise relations. It improves accuracy and interpretability while enabling spatial modeling of higher-order multivariate relations beyond first-order pairwise relations. We also designed a plug-and-play probabilistic prediction module that enables switches between probabilistic and deterministic modes. Experiments based on the human activity intensity in Fuzhou, China, demonstrated the advantages in accuracy, interpretability and multi-scenario applicability.

Accurate and efficient air quality prediction is crucial for public health protection and environmental sustainability. While numerous grid-based and graph-based prediction models have been developed, they encounter challenges in large-scale scenarios: (1) Grid-based models, though computationally efficient, have limited prediction accuracy in large-scale sparse scenarios; (2) Graph-based models, despite higher prediction accuracy, suffer from significant computational inefficiencies when dealing with a large number of sensors, i.e. graph nodes. To address these issues, we propose a Lightweight Ensemble Predictor (LiEnPred) for efficient air quality prediction in large-scale sparse scenarios. First, we present a data structure transformation algorithm that converts sparse monitoring sensors from graph structures to compact grid structures, preserving the connections between graph nodes. Next, we present a lightweight parameter-shared spatio-temporal dilation convolution network that efficiently captures spatio-temporal dependencies in air quality data without significantly increasing computation time or parameter scale. In our experiments, we collected air quality data from over 2000 sensors across China over the past three years and evaluated LiEnPred’s prediction performance in large-scale scenarios using PM2.5 and NO2 concentration data. The experimental results demonstrate that the proposed LiEnPred model matches or exceeds the predictive accuracy of eight baselines with faster time efficiency and fewer model parameters.

Modeling population-level human mobility has been attracting multidisciplinary research attention due to its profound implications for sustainable urban development. However, previous studies have often neglected the explicit consideration of spatial heterogeneity of travel demand, which limits their abilities to accurately estimate mobility flows. In this study, we introduce a prior-guided, data-driven human mobility model that integrates the position of origins and destinations, spatial travel patterns, and physical models as priors to capture spatial heterogeneity of human mobility. Specifically, we introduce the concept of ‘relative attractiveness’ to emulate the underlying driving force for the formation of spatial heterogeneity in human mobility. To learn the embeddings of ‘relative attractiveness’, we propose a suite of methods that integrate prior knowledge and graph neural networks, mainly including a relative position encoding module to encode the position of different origin-destination (OD) pairs relative to the entire geographical space and a message-passing method inspired by the classical physical models to simulate the mechanisms of mobility flow generation. Finally, a gradient boosting regression tree is trained to generate the mobility flow based on the learned embeddings. Extensive experiments on two real-world datasets have showed our model outperforms state-of-the-art data-driven mobility models in terms of accuracy and generalization.

Ensemble learning synthesizes the advantages of different models and has been widely applied in the field of spatial prediction. However, the nonlinear constraints of spatial heterogeneity on the model ensemble process make it difficult to adaptively determine the ensemble weights, greatly limiting the predictive ability of the ensemble learning model. This paper therefore proposes a novel geographical spatial heterogeneous ensemble learning method (GSH-EL). Firstly, the geographically weighted regression model, geographically optimal similarity model, and random forest model are used as three base learners to express local spatial heterogeneity, global feature correlation, and nonlinear relationship of geographic elements, respectively. Then, a spatially weighted ensemble neural network module (SWENN) of GSH-EL is proposed to express spatial heterogeneity by exploring the complex nonlinear relationship between the spatial proximity and ensemble weights. Finally, the outputs of the three base learners are combined with the spatial heterogeneous ensemble weights from SWENN to obtain the spatial prediction results. The proposed method is validated on the PM2.5 air quality and landslide dataset in China, both of which obtain more accurate prediction results than the existing ensemble learning strategies. The results confirm the need to accurately express spatial heterogeneity in the model ensemble process.

Most mobility modeling methods are designed to solve specific tasks, leading to questions regarding their deficiency in generalizability. Inspired by the bloom of foundation models, we proposed a Trajectory Generation framework based on the Diffusion Model (TrajGDM) to capture the universal mobility pattern in a trajectory dataset by learning the trajectory generation process. The process is modeled as a step-by-step uncertainty-reducing process, in which a deep learning network with a novel training method is proposed to learn from the process. We compared the proposed trajectory generation method with six baselines on two public trajectory datasets. The results showed that the similarity between the generated and real trajectory movements measured by the Jensen-Shannon Divergence improved significantly on both datasets. Moreover, we applied zero-shot inferences on two basic trajectory tasks: trajectory prediction and trajectory reconstruction. The accuracy improved by a maximum of 25.6% on two tasks. The universal mobility pattern that is suitable for solving multiple trajectory tasks is verified, inferring the strong generalizability of our model. Finally, the study provides insights into artificial intelligence’s understanding of human mobility by exploring the way the model maps the trajectory in the latent space into reality.

Explainable spatio-temporal prediction gains attraction in the development of geospatial artificial intelligence. The neural ordinal differential equation (NODE) emerges as a new solution for explainable spatio-temporal prediction. However, challenges still need to be solved in most existing NODE-based prediction models, such as difficulty modeling spatial data and mining long-term temporal dependencies in data. In this study, we propose a spatio-temporal attentional NODE (STA-ODE) to address the two challenges above. First, we define a spatio-temporal ordinary differential equation to predict a value at each time iteratively by a novel spatio-temporal derivative network. Second, we develop an attention mechanism to fuse multiple prediction values for capturing long-term temporal dependencies in data. To train the STA-ODE model, we design a loss function that aligns the prediction results in spatial dimension with prediction results in temporal dimension to calibrate the parameters of the model. The proposed model was validated with three real-world spatio-temporal datasets (traffic flow dataset, PM2.5 monitoring dataset, and temperature monitoring dataset). Experimental results showed that STA-ODE outperformed seven existing baselines regarding prediction accuracy. In addition, we used visualization to demonstrate the sound interpretability and prediction accuracy of the STA-ODE model.

Accurate traffic flow prediction on the urban road network is an indispensable function of Intelligent Transportation Systems (ITS), which is of great significance for urban traffic planning. However, the current traffic flow prediction methods still face many challenges, such as missing values and dynamic spatial relationships in traffic flow. In this study, a dynamic temporal graph neural network considering missing values (D-TGNM) is proposed for traffic flow prediction. First, inspired by the Bidirectional Encoder Representations from Transformers (BERT), we extend the classic BERT model, called Traffic BERT, to learn the dynamic spatial associations on the road structure. Second, we propose a temporal graph neural network considering missing values (TGNM) to mine traffic flow patterns in missing data scenarios for traffic flow prediction. Finally, the proposed D-TGNM model can be obtained by integrating the dynamic spatial associations learned by Traffic BERT into the TGNM model. To train the D-TGNM model, we design a novel loss function, which considers the missing values problem and prediction problem in traffic flow, to optimize the proposed model. The proposed model was validated on an actual traffic dataset collected in Wuhan, China. Experimental results showed that D-TGNM achieved good prediction results under four missing data scenarios (15% random missing, 15% block missing, 30% random missing, and 30% block missing), and outperformed ten existing state-of-the-art baselines.

Accurate estimation of missing traffic data is one of the essential components in intelligent transportation systems (ITS). The non-Euclidean data structure and complex missing traffic flow patterns make it challenging to capture nonlinear spatiotemporal correlations of missing traffic flow, which are critical for the imputation of missing traffic data. In this study, we propose a novel multi-view bidirectional spatiotemporal graph network called Multi-BiSTGN to impute urban traffic data with complex missing patterns. First, three spatiotemporal graph sequences are constructed to comprehensively describe traffic conditions from different temporal correlation views, i.e. temporal closeness view, daily periodicity view, and weekly periodicity view. Then, three bidirectional spatiotemporal graph networks are fused by a parametric-matrix-based method to obtain the final imputation results. To train the Multi-BiSTGN model, a novel loss function that considers the interactions between three temporal correlation views is designed to optimize the parameters of the Multi-BiSTGN model. The proposed model was validated on real-world traffic datasets collected in Wuhan, China. Experimental results showed that Multi-BiSTGN outperformed ten existing baselines under different missing types (random missing, block missing, and mixed missing) and missing rates.