Spatial-temporal Data Research Articles

Bayesian High-Rank Hankel Matrix Completion for Nonlinear Synchrophasor Data Recovery

Phasor measurement units (PMUs) provide high temporal-resolution synchrophasor measurements for power system monitoring and control. The frequent data quality issues, such as missing and bad data, prevent the incorporation of synchrophasor data in real-time operations. Most existing data-driven data recovery methods assume the power system dynamics can be approximated by a linear dynamical system, and the recovery performance degrades significantly when the power system is experiencing nonlinear dynamics during significant events. This paper proposes a data-driven Bayesian nonlinear synchrophasor data recovery method (Ba-NSDR) that can recover a consecutive time period of simultaneous data losses or errors across all channels, even when the underlying system is highly nonlinear. The idea is to lift the Hankel matrix of the spatial-temporal synchrophasor data to a higher dimension such that the lifted Hankel matrix is low-rank in that space and can be processed with the kernel trick. Our proposed Bayesian method then infers the probabilistic distributions of synchrophasor from the partial observations. Some distinctive features of Ba-NSDR include an uncertainty index to measure the accuracy of the recovery result and the robustness to parameter selections. Our method is verified on both synthetic and recorded event datasets.

MFDS-STGCN: Predicting the Behaviors of College Students With Fine-Grained Spatial-Temporal Activities Data

MFDS-STGCN: Predicting the Behaviors of College Students With Fine-Grained Spatial-Temporal Activities Data

Empirical Distribution of Glottal Edges (EDGE): A Statistical Assessment of Vocal Fold Kinematics Using High-Speed Videoendoscopy.

Although laryngeal high-speed videoendoscopy (HSV) is crucial for studying vocal fold vibrations, its translation to clinical practice has been hindered by the large volume of data it produces and the difficulty in interpreting current analysis methods. Although image processing techniques have been developed to map spatial-temporal data into two-dimensional representations, they alter the geometrical construction of the glottis and do not provide standard quantitative features, thus challenging clinical interpretation. In response, we propose a new visualization and analysis framework for assessing the dynamics of vocal folds based on the empirical distribution of the glottal edge using HSV. This procedure analyzes vocal fold oscillations by preserving the shape of the glottis and quantifying the asymmetry between right and left vocal fold displacements along the anterior-posterior axis. This method was evaluated on four groups of participants: ten with normal voices, ten with vocal fold nodules, ten with muscle tension dysphonia, and two with unilateral vocal fold paralysis. The proposed method produces distinct representations for normal and pathological vocal fold vibratory behaviors and derived features based on amplitude and phase asymmetry metrics that show statistically significant differences between normal and pathological groups. Comparative analysis with state-of-the-art techniques indicates that our proposed method can complement the assessment of vocal fold vibration and enhance the clinical translation of HSV.

Bicycle Sharing Demand Analysis and Operation Community Division based on Riding Spatial-Temporal Data -Taking Beijing as an Example

Under the background of vigorously building sustainable cities, shared bicycles have developed rapidly, and some operation and management problems have gradually become prominent. In recent years, the research based on spatial-temporal big data of shared bicycles mostly focuses on the analysis of hotspots in time and space, and the current management of shared bicycle is mostly divided into operating communities by administrative divisions. The construction of bicycle operation management is not carried out in combination with the characteristics of spatial-temporal data and actual needs. Based on the shared bicycle order data in Beijing, this paper analyzes the basic spatial and temporal characteristics of shared bicycles, and uses geographically weighted regression model to analyze the impact of various types of built environments on the demand distribution for shared bicycles. Finally using Fast Unfolding algorithm to identify communities with close demand for shared bicycles. The study finds that the overall utilization rate of bicycle resources is currently low, and various types of POIs have different ranges and degrees of impact on parking demand, as well as the division of bicycle communities does not fully coincide with administrative divisions. This study helps to understand the overall demand and impact characteristics of shared bicycles, provides a basis for the development of differentiated strategies for the delivery, management, and scheduling of shared bicycles, and helps urban managers to optimize the design of the shared bicycle system.

Objects detection theory for evaluating the city environmental quality

IntroductionThe primary focus of this paper is to assess urban ecological environments by employing object detection on spatial-temporal data images within a city, in conjunction with other relevant information through data mining.MethodsFirstly, an improved YOLOv7 algorithm is applied to conduct object detection, particularly counting vehicles and pedestrians within the urban spatial-temporal data. Subsequently, the k-means superpixel segmentation algorithm is utilized to calculate vegetation coverage within the urban spatial-temporal data, allowing for the quantification of vegetation area. This approach involves the segmentation of vegetation areas based on color characteristics, providing the vegetation area’s measurements. Lastly, an ecological assessment of the current urban environment is conducted based on the gathered data on human and vehicle density, along with vegetation coverage.ResultsThe enhanced YOLOv7 algorithm employed in this study yields a one-percent improvement in mean AP (average precision) compared to the original YOLOv7 algorithm. Furthermore, the AP values for key categories of interest, namely, individuals and vehicles, have also improved in this ecological assessment.DiscussionSpecifically, the AP values for the ‘person’ and ‘pedestrian’ categories have increased by 13.9% and 9.3%, respectively, while ‘car’ and ‘van’ categories have seen AP improvements of 6.7% and 4.9%. The enhanced YOLOv7 algorithm contributes to more accurate data collection regarding individuals and vehicles in subsequent research. In the conclusion of this paper, we further validate the reliability of the urban environmental assessment results by employing the Recall-Precision curve.

Spatial-temporal traffic data imputation based on dynamic multi-level generative adversarial networks for urban governance

Complete and accurate traffic data lays an essential foundation for urban governance and intelligent transportation system. Data missing is an inevitable issue in road network analysis with the expansion of urban roads and increasing application of data monitors. Data imputation is regarded as a cost-effective approach for solving data missing problem. However, the spatial-temporal features of massive traffic data have not been comprehensively considered in the previous related studies, and non-uniform distribution of missing data often causes poor results. Therefore, we propose a dynamic multi-level generative adversarial networks (MLGAN) model for the imputation of large-scale spatial-temporal traffic data. The model first captures the spatial-temporal features of traffic data through composite features extractor based on the bidirectional gated recurrent unit (BGRU) and the graph convolutional network (GCN) structure. Then, different sampler structures with varied missing rate are adopted to analyze the feature data and present the reconstruction results. Enhanced by the dynamic stratification strategy, the model is less sensitive to the variation of missing rates. Finally, data imputation experiments are conducted to validate the proposed approach using the data collected from the PeMS dataset. The results show that the proposed model has superior data imputation performance on large-scale dataset compared with the traditional models.

Regional route guidance with realistic compliance patterns: Application of deep reinforcement learning and MPC

Solving link-based route guidance problems for large-scale networks is computationally challenging and faces practical issues, such as spatial–temporal data coverage. Thus, regional route guidance has emerged as a promising strategy, which utilizes regional approaches (i.e., network-level macroscopic traffic models) to capture traffic dynamics. Existing regional route guidance models have mainly focused on macroscopic flows and aggregated splitting rates, employing uniformly sampled vehicles as controllable targets. These models, however, overlook the inherent nature of individual drivers’ compliance. It may deteriorate the guidance performance as existing route guidance models cannot effectively generate customized route plans for compliant vehicles. This paper aims to introduce a regional route guidance framework with the utilization of different solution approaches, i.e. model-based and data-driven, considering the compliance pattern. In particular, MPC-based and deep reinforcement learning-based schemes are proposed for the information service provider to send customized route plans to compliant individuals. Within each solution approach, different route guidance strategies are developed for specific purposes and evaluated through numerical experiments under low- and high-congestion scenarios. Besides, the trade-off between total travel time and average trip length is studied, wherein the balance reward in a deep reinforcement learning-based approach enables controllable agents to reduce trip costs while compromising partial system utility. The findings indicate the effectiveness of the proposed strategies in alleviating network congestion. Additionally, multi-agent approaches outperform single-agent ones, highlighting the benefits of cooperative decision-making in traffic management.

Backpack satellite transmitters reduce survival but not nesting propensity or success of greater sage-grouse.

Telemetry technology is ubiquitous for studying the behavior and demography of wildlife, including the use of traditional very high frequency (VHF) radio telemetry and more recent methods that record animal locations using global positioning systems (GPS). Satellite-based GPS telemetry allows researchers to collect high spatial-temporal resolution data remotely but may also come with additional costs. For example, recent studies from the southern Great Basin suggested GPS transmitters attached via backpacks may reduce the survival of greater sage-grouse (Centrocercus urophasianus) relative to VHF transmitters attached via collars that have been in use for decades. While some evidence suggests GPS backpacks reduce survival, no studies have examined the effects of GPS backpacks on breeding behavior and success. Therefore, we compared survival, breeding behavior, and nest success of sage-grouse hens marked with both VHF collars and GPS backpack transmitter over a 7-year period in central Idaho, USA. GPS backpacks reduced spring-summer survival of sage-grouse hens relative to hens with VHF collars, where daily mortality probability was 68%-82% higher from March 1 to August 1. Yet satellite GPS backpacks did not consistently affect nest success or the likelihood or timing of nest initiation relative to VHF collars. Daily nest survival varied annually and with timing of nest initiation and nest age, but marginal effects of transmitter type were statistically insignificant and interactions between transmitter type and study year produced no meaningful patterns. Our results corroborate recent studies for the effect of satellite GPS backpacks on sage-grouse survival, but also suggest that these transmitters do not appear to affect components of fecundity. Our results therefore add important context to recent debate surrounding the effects of GPS backpacks on sage-grouse, and the relative strengths and weaknesses of different transmitter types for understanding behavior and population dynamics.

Detecting, Analyzing, and Predicting Land Use/Land Cover (LULC) Changes in Arid Regions Using Landsat Images, CA-Markov Hybrid Model, and GIS Techniques

Understanding the change dynamics of land use and land cover (LULC) is critical for efficient ecological management modification and sustainable land-use planning. This work aimed to identify, simulate, and predict historical and future LULC changes in the Sohag Governorate, Egypt, as an arid region. In the present study, the detection of historical LULC change dynamics for time series 1984–2002, 2002–2013, and 2013–2022 was performed, as well as CA-Markov hybrid model was employed to project the future LULC trends for 2030, 2040, and 2050. Four Landsat images acquired by different sensors were used as spatial–temporal data sources for the study region, including TM for 1984, ETM+ for 2002, and OLI for 2013 and 2022. Furthermore, a supervised classification technique was implemented in the image classification process. All remote sensing data was processed and modeled using IDRISI 7.02 software. Four main LULC categories were recognized in the study region: urban areas, cultivated lands, desert lands, and water bodies. The precision of LULC categorization analysis was high, with Kappa coefficients above 0.7 and overall accuracy above 87.5% for all classifications. The results obtained from estimating LULC change in the period from 1984 to 2022 indicated that built-up areas expanded to cover 12.5% of the study area in 2022 instead of 5.5% in 1984. This urban sprawl occurred at the cost of reducing old farmlands in old towns and villages and building new settlements on bare lands. Furthermore, cultivated lands increased from 45.5% of the total area in 1984 to 60.7% in 2022 due to ongoing soil reclamation projects in desert areas outside the Nile Valley. Moreover, between 1984 and 2022, desert lands lost around half of their area, while water bodies gained a very slight increase. According to the simulation and projection of the future LULC trends for 2030, 2040, and 2050, similar trends to historical LULC changes were detected. These trends are represented by decreasing desert lands and increasing urban and cultivated newly reclaimed areas. Concerning CA-Markov model validation, Kappa indices ranged across actual and simulated maps from 0.84 to 0.93, suggesting that this model was reasonably excellent at projecting future LULC trends. Therefore, using the CA-Markov hybrid model as a prediction and modeling approach for future LULC trends provides a good vision for monitoring and reducing the negative impacts of LULC changes, supporting land use policy-makers, and developing land management.

Micro-scale urbanization-based risk factors for dengue epidemics.

Dengue is one of the world's most rapidly spreading mosquito-borne viral diseases. As it is found mostly in urban and semi-urban areas, urbanization and associated human activities that affect the environment and larval habitats could become risk factors (e.g., lane width, conditions of street ditches) for the spread of dengue. However, there are currently no systematic studies of micro-scale urbanization-based risk factors for the spread of dengue epidemics. We describe the study area, two micro-scale environmental risk factors associated with urbanization, and meteorological data. Since the observations involve spatial and temporal correlations, we also use some statistical methods for the analysis of spatial and spatial-temporal data for the relationship between urbanization and dengue. In this study, we analyzed data from Kaohsiung, a densely populated city in southern Taiwan, and found a positive correlation between environmental risk factors associated with urbanization (ditches positive for mosquito larvae and closely packed streets termed "dengue lanes") and clustering effects in dengue cases. The statistical analysis also revealed that the occurrence of positive ditches was significantly associated with that of dengue lanes in the study area. The relationship between climate variables and positive ditches was also analyzed in this paper, indicating a relationship between dengue and both rainfall and temperature, with temperature having a greater effect. Overall, this work is immediately relevant and applicable for policymakers in government, who will need to reduce these favorable habitats for vector-born disease spreaders and implement regulations for new urban constructions to thus reduce dengue spread in future outbreaks.

Comparison and analysis of various deep learning models for super-resolution reconstruction of turbulent flows

Single image super-resolution (SR) has become a promising research topic, with many deep learning-based models invented to reconstruct high-fidelity high-resolution (HR) images from low-resolution (LR) images. Motivated by a large amount of turbulent flow field data collected by experimental measurements and numerical simulation, researchers begin investigating the application of these data-driven deep learning models to conduct SR reconstruction of LR flow field data. Due to the limitations of experimental equipment and computing power, sometimes researchers can only obtain LR data. However, deep learning models can quickly reconstruct HR spatial-temporal turbulent data from LR data so that researchers can easily conduct further qualitative and quantitative analyses. This article reviews the development of flow field data SR reconstruction models and the problems encountered from the two aspects of network structure and loss function definition. Finally, we propose the research direction of applying the conditional generative adversarial network (cGAN) to turbulent flow SR reconstruction since seldom study has been conducted in this field.

Co-Movement Pattern Mining from Videos

Co-movement pattern mining from GPS trajectories has been an intriguing subject in spatial-temporal data mining. In this paper, we extend this research line by migrating the data source from GPS sensors to surveillance cameras, and presenting the first investigation into co-movement pattern mining from videos. We formulate the new problem, re-define the spatial-temporal proximity constraints from cameras deployed in a road network, and theoretically prove its hardness. Due to the lack of readily applicable solutions, we adapt existing techniques and propose two competitive baselines using Apriori-based enumerator and CMC algorithm, respectively. As the principal technical contributions, we introduce a novel index called temporal-cluster suffix tree (TCS-tree), which performs two-level temporal clustering within each camera and constructs a suffix tree from the resulting clusters. Moreover, we present a sequence-ahead pruning framework based on TCS-tree, which enables the concurrent utilization of all pattern constraints to filter candidate paths. Finally, to reduce verification cost on the candidate paths, we propose a sliding-window based co-movement pattern enumeration strategy and a hashing-based dominance eliminator, both of which are effective in avoiding redundant operations. We conduct extensive experiments for scalability and effectiveness analysis. Our results validate the efficiency of the proposed index and mining algorithm, which runs remarkably faster than the two baseline methods. Additionally, we construct a video database with 1169 cameras and perform an end-to-end pipeline analysis to study the performance gap between GPS-driven and video-driven methods. Our results demonstrate that the derived patterns from the video-driven approach are similar to those derived from groundtruth trajectories, providing evidence of its effectiveness.

PM2.5 Estimation in the Czech Republic using Extremely Randomized Trees: A Comprehensive Data Analysis

The accuracy of artificial intelligence techniques in estimating air quality is contingent upon a multitude of influencing factors. Unlike our previous study that examined PM2.5 over whole Europe using unbalanced spatial-temporal data, the focus of this study was on estimating PM2.5 specifically over the Czech Republic using more balanced dataset to train and evaluate the model. Moreover, the spatial autocorrelation between the ground-based station was taken into consideration while building the model. The feature importance while developing the Extra Trees model revealed that spatial autocorrelation had greater significance in comparison to commonly used inputs such as elevation and NDVI. We found that R2 of the 10-CV for the new model was 16% higher than the previous one. R2 reached 0.85 when predicting unseen data in new locations. The developed spatiotemporal model was employed to generate comprehensive daily maps covering the entire study area throughout the 2018–2020 years. The temporal analysis showed that the levels of PM2.5 exceeded recommended limits of 20 µg/m3 during the year 2018 in many regions. The eastern part of the country suffered from the highest concentrations especially over Zlín and Moravian-Silesian Regions where in the 2018 winter, the values reached risky average concentrations of 30 µg/m3 and 35 µg/m3 respectively. Air quality improved during the next two years in all regions reaching promising levels in 2020 where almost all regions had average concentrations less than 20 µg/m3. The generated dataset will be available for other future air quality studies.

Evaluating Road Hazard Maintenance Efficiency Using Citizen Science Data to Improve Road Safety

Accidents caused by road hazards can be prevented through regular inspections by road management agencies. To this end, traffic agencies allocate substantial budgets and workforces to maintain the performance of roads. Additionally, traffic agencies require comprehensive data such as the classifications and sizes of road hazards. However, collecting spatial–temporal data on various road hazards is challenging, and evaluating it comprehensively with respect to work efficiency and budget allocation is difficult due to stakeholder interests across agencies. This study proposes a process of evaluating operational efficiency in terms of maintaining roads and preventing hazards by analyzing citizen scientist-based data. First, we collected data from drivers through a mobile application and applied text mining techniques to classify each complaint into several types of road hazard maintenance. Second, we developed an indicator to measure operational efficiency using the processed data and evaluated each regional agency per each type of maintenance. The results of this study provide evidence that specific types of road hazards occur prominently under specific agencies. In addition, the time required to provide maintenance for identical road hazards can vary among agencies. These results suggest that the maintenance budget for the entire national highway may need to be distributed differently based upon regional characteristics.

Multi-point deformation monitoring model of concrete arch dam based on MVMD and 3D-CNN

It is difficult for single measuring point model to synchronously characterize the spatial correlation of deformation sequence of concrete dam and the similarity of displacements of different measuring points. Considering the limitation of conventional model in dealing with spatial-temporal data structure, In this study, a deep learning technique for multi-point and multi-output deformation prediction of concrete dams is proposed, which is based on Multivariate Variational Mode Decomposition (MVMD) and three-dimensional convolutional neural network (3D-CNN) combined model. Firstly, the multi-point sequence is decomposed into several relatively stable sub-sequences with different frequency scales by MVMD, which reduces the complexity of the multi-point displacement sequence. Then, In order to enhance the model 's ability to fit the time and spatial features of the sequence, the 3D-CNN model is used to predict the decomposed components and reconstruct the displacement data of the measuring points. This study focuses on the principle, process and calculation steps of the proposed model, and verifies the proposed model through the deformation data of three typical measuring points. The test results show that the prediction model proposed in this paper is an effective method for dam deformation prediction.

STRP-DBSCAN: A Parallel DBSCAN Algorithm Based on Spatial-Temporal Random Partitioning for Clustering Trajectory Data

Trajectory clustering algorithms analyze the movement trajectory of the target objects to mine the potential movement trend, regularity, and behavioral patterns of the object. Therefore, the trajectory clustering algorithm has a wide range of applications in the fields of traffic flow analysis, logistics and transportation management, and crime analysis. Existing algorithms do not make good use of the temporal attributes of trajectory data, resulting in a long clustering time and low clustering accuracy of spatial-temporal trajectory data. Meanwhile, the density-based clustering algorithms represented by DBSCAN are very sensitive to the clustering parameters. The radius value Eps and the minimal points number MinPts within Eps radius, defined by the user, have a significant impact on the clustering results, and tuning these parameters is difficult. In this paper, we present STRP-DBSCAN, a parallel DBSCAN algorithm based on spatial-temporal random partitioning for clustering trajectory data. It adopts spatial-temporal random partitioning to distribute balanced computation among different computing nodes and reduce the communication overhead of the parallel clustering algorithm, thus improving the execution efficiency of the DBSCAN algorithm. We also present the PER-SAC algorithm, which uses deep reinforcement learning to combine the prioritized experience replay (PER) and the soft actor-critic (SAC) algorithm for autotuning the optimal parameters of DBSCAN. The experimental results show that STRP-DBSCAN effectively reduces the clustering time of spatial-temporal trajectory data by up to 96.2% and 31.2% compared to parallel DBSCAN and the state-of-the-art RP-DBSCAN. The PER-SAC algorithm also outperforms the state-of-the-art DBSCAN parameter tuning algorithms and improves the clustering accuracy by up to 8.8%. At the same time, the proposed algorithm obtains a higher stability of clustering accuracy.

Graph construction on complex spatiotemporal data for enhancing graph neural network-based approaches

Graph neural networks (GNNs) haven proven to be an indispensable approach in modeling complex data, in particular spatial temporal data, e.g., relating to sensor data given as time series with according spatial information. Although GNNs provide powerful modeling capabilities on such kind of data, they require adequate input data in terms of both signal and the underlying graph structures. However, typically the according graphs are not automatically available or even predefined, such that typically an ad hoc graph representation needs to be constructed. However, often the construction of the underlying graph structure is given insufficient attention. Therefore, this paper performs an in-depth analysis of several methods for constructing graphs from a set of sensors attributed with spatial information, i.e., geographical coordinates, or using their respective attached signal data. We apply a diverse set of standard methods for estimating groups and similarities between graph nodes as location-based as well as signal-driven approaches on multiple benchmark datasets for evaluation and assessment. Here, for both areas, we specifically include distance-based, clustering-based, as well as correlation-based approaches for estimating the relationships between nodes for subsequent graph construction. In addition, we consider two different GNN approaches, i.e., regression and forecasting in order to enable a broader experimental assessment. Typically, no predefined graph is given, such that (ad hoc) graph creation is necessary. Here, our results indicate the criticality of factoring in the crucial step of graph construction into GNN-based research on spatial temporal data. Overall, in our experimentation no single approach for graph construction emerged as a clear winner. However, in our analysis we are able to provide specific indications based on the obtained results, for a specific class of methods. Collectively, the findings highlight the need for researchers to carefully consider graph construction when employing GNNs in the analysis of spatial temporal data.

Forecasting of oil production driven by reservoir spatial–temporal data based on normalized mutual information and Seq2Seq-LSTM

Traditional machine learning methods are difficult to accurately forecast oil production when development measures change. A method of oil reservoir production prediction based on normalized mutual information and a long short-term memory-based sequence-to-sequence model (Seq2Seq-LSTM) was proposed to forecast reservoir production considering the influence of liquid production and well spacing density. First, the marine sandstone reservoirs in the Y basin were taken as the research object to establish the sample database. Then, the feature selection was carried out according to the normalized mutual information, and liquid production, production time, equivalent well spacing density, fluidity and original formation pressure were determined as input features. Finally, a Seq2Seq-LSTM model was established to forecast reservoir production by learning the interaction from multiple samples and multiple sequences, and mining the relationship between oil production and features. The research showed that the model has a high accuracy of production prediction and can forecast the change of production when the liquid production and well spacing density change, which can provide scientific recommendations to help the oilfield develop and adjust efficiently.

Evaluation of Urban Green Space Supply and Demand Based on Mobile Signal Data: Taking the Central Area of Shenyang City as an Example

The degree of coordination between the supply and demand for urban green spaces serves as a vital metric for evaluating urban ecological development and the well-being of residents. An essential principle in assessing this coordination is the precise quantification of both the demand and supply of green spaces, as well as the differential representation of their spatiotemporal structures. This study utilizes the entropy weight method (EWM) and principal component analysis (PCA) to comprehensively measure supply indicators for green space quantity and quality in the central urban area of Shenyang, China. To establish reliable and quantifiable demand indicators, mobile signaling spatial-temporal data are corrected by incorporating static population cross-sectional data. The Gaussian two-step floating catchment area method (Ga2SFCA) is employed to calculate the accessibility of green spaces in each community with ArcGIS 10.2 software, while the Gini coefficient is utilized to assess the equity of green space distribution within the study area. This study employs location entropy to determine the levels of supply and demand for green spaces in each subdistrict. Furthermore, the priority of community-scale green space regulation is accurately determined by balancing vulnerable areas of green space supply and replenishing green space resources for the ageing population. The findings suggest a Gini coefficient of 0.58 for the supply and demand of green spaces in Shenyang’s central metropolitan region, indicating a relatively low level of equalization in overall green space allocation. Based on location entropy, the classification of supply and demand at the street level yields the following outcomes: balanced areas comprise 21.98%, imbalanced areas account for 26.37%, and highly imbalanced regions represent 51.65%. After eliminating the balanced regions, the distribution of the elderly population is factored in, highlighting the spatial distribution and proportions of communities with distinct regulatory priorities: Level 1 (S1) constitutes 7.4%, Level 2 (S2) accounts for 60.9%, and Level 3 (S3) represents 31.7%. Notably, the communities in the S1 category exhibit spatial distribution characteristics of aggregation within the inner ring and the northern parts of the third ring. This precise identification of areas requiring urgent regulation and the spatial distribution of typical communities can provide reliable suggestions for prioritizing green space planning in an age-friendly city.

Robust spatial temporal imputation based on spatio-temporal generative adversarial nets

Spatial temporal imputation is an important part of intelligent transportation system, and spatial temporal data missing also largely affects the performance of its downstream tasks. Spatial temporal data are affected by spatial, temporal, and unexpected conditions, and has a complex change pattern. Existing researches mostly focus on time series and often ignore the modeling of spatial dependency. Moreover, the imputation performance is poor under high missing rate or block missing scenarios, needing to enhance the model robustness. Therefore, we proposed a spatio-temporal generative adversarial imputation network (named STGAIN) to handle with the complex spatial temporal dependency and the diversity of missing conditions on spatio-temporal data. It includes a spatio-temporal generator and a spatio-temporal discriminator, which uses a graph convolutional neural network as a spatial aggregator and a one-dimensional deformable convolutional neural network as a temporal extractor. The generator observes the unmissing part data and the spatial relationships to impute the missing parts. The discriminator then takes the completed data as input and attempts to determine which components were actually observed and which were imputed. The spatio-temporal modeling ability was improved in the competition with each other. In order to make the generator in GAN distinguish the observed part from the missing part, a fusion loss function (including reconstruction loss and BCE loss) was designed for the generator training. Finally, experiments were conducted on two real highway datasets for different missing rates. And we test the model robustness under different size of dataset and noise interference. The results show that STGAIN significantly outperforms the existing methods.