Demand prediction for bike-sharing systems: A spatial and semantic modeling approach for enhanced accuracy and operational efficiency

Background:A wide range of epidemiological modeling approaches have been applied to the SARS-CoV-2 pandemic, which presents an opportunity to assess common approaches applied to specific research questions. Spatial models interrogate how heterogeneities and host movement dynamics influence local and regional patterns of disease, issues that were of great interest for understanding and controlling SARS-CoV-2.Objective:Here we present a systematic review of spatial epidemiological modeling approaches of SARS-CoV-2. We describe common themes and highlight unique strategies, providing a foundation for researchers to devise spatial models most appropriate for future pathogens and epidemics. Our review also categorizes the research questions that were addressed with spatial models, highlights parameter estimation techniques, and describes the cyber infrastructure used for model development.Methods:We conducted a systematic review using Web of Science and a standardized set of keywords, followed by thorough examination of abstracts and full texts to determine which studies met our inclusion criteria. To guide our description and comparisons of models, we developed a Geography, Population, Movement (GPM) framework that conceptualizes the interactions between three distinct subcomponents of any spatial model. The geographic model represents the physical arena in which the model is implemented, the intra-population model describes the transmission and disease processes that occur within distinct spatial units of the geography, and the movement model describes the algorithms that dictate how hosts move among spatial units within the geography.Results:The search identified a total of 193 articles, of which 109 were included in our review. The most abundant intra-population modeling methods were agent-based (47.7%) and compartmental modeling (29.4%) approaches. Movement models ranged in complexity, with the most complex models implementing commuter movement among many points of interest in the geographic arena, which were sometimes parameterized by fine-scale mobility data. Geographic models ranged from describing microcosms, such as single classrooms, all the way up to multi-country models. Of the 63.3% of models studies that specified the programming language used, we detected ten different languages, with Matlab and Python being the most frequent, although only 30.6% of studies provided open-access code for their models. We also described eight specialized software systems that were used to construct agent-based or compartment models of COVID-19.Conclusions:Our review identified and characterized a variety of spatial modeling strategies and software that were usefully employed to address many relevant epidemiological questions for COVID-19. Future research is needed to quantitatively assess which modeling approaches are most appropriate in specific situations, to answer specific questions, or to apply to certain disease systems. Moreover, future cyberinfrastructure could help to modularize and standardize modeling approaches, which would increase transparency and reproducibility, and which would facilitate a detailed examination of which model attributes relate to model performance in a variety of contexts.

Customer preferences are found to evolve over time and correlate with geographical locations. Studying the spatiotemporal heterogeneity of customer preferences is crucial to engineering design as it provides a dynamic perspective for understanding the trend of customer preferences. However, existing choice models for demand modeling do not take the spatiotemporal heterogeneity of customer preferences into consideration. Learning-based spatiotemporal data modeling methods usually require large-scale datasets for model training, which are not applicable to small aggregated data, such as the sale records of a product in several regions and years. To fill this research gap, we propose a spatial panel modeling approach to investigate the spatiotemporal heterogeneity of customer preferences. Product and regional attributes varying in time are included as model inputs to support demand forecasting in engineering design. With case studies using the dataset of small SUVs and compact sedans in China's automotive market, we demonstrate that the spatial panel modeling approach outperforms other statistical spatiotemporal data models and non-parametric regression methods in goodness of fit and prediction accuracy. We also illustrate a potential design application of the proposed approach in a portfolio optimization of two vehicles from the same producer. While the spatial panel modeling approach exists in econometrics, applying this approach to support engineering decisions by considering spatiotemporal heterogeneity and introducing engineering attributes in demand forecasting is the contribution of this work. Our paper is focused on presenting the approach rather than the results per se.

Lessons learned for spatial modelling of ecosystem services in support of ecosystem accounting

High-precision global ionospheric modeling is important for radio communication, navigation, or studies on space weather. Traditional spatial ionospheric modeling approaches include spherical harmonics and trigonometric B-splines. The Ionospheric Associated Analysis Centers (IAAC) of the International GNSS Service (IGS) use these methods to model vertical total electron content (VTEC) globally, and generate Global Ionospheric Maps (GIMs). Due to the limitations of spatial modeling approaches, conventional GIMs cannot comprehensively describe the spatial feature of the ionosphere. With the capability of capturing complex and non-linear relationships of diverse data, machine learning (ML) has been increasingly applied to ionospheric modeling. Currently, most of the existing ML-related studies focused on temporal prediction of ionospheric states and rarely considered the aspect of the spatial modeling of VTEC. Although some studies predicted global ionosphere maps using machine learning, they used conventional GIMs as inputs, implying that the precision of the ML-based spatial modeling could be limited by traditional methods and quantity of input GNSS observations utilized to generate GIMs.The goal of this study is the spatial interpolation of VTEC using ML methods for the generation of ML-based GIMs. We first determine VTEC using carrier-to-code levelling through Kalman filter and based on geometry-free multi-GNSS observations from GNSS stations of the IGS network. The derived satellite-specific VTEC time series are then used to train the ML models. Several algorithms, such as extreme gradient boosting and random forest, are applied and their performance is evaluated. Moreover, VTEC from satellite altimetry is used as an additional means to assess the quality of the generated ML models. Finally, we compare the acquired ML-based GIMs with conventional GIMs to investigate the advantage of using the proposed approach for global VTEC modeling.

Sidney Holt and Ray Beverton are primarily recognized for developing the basis of demographic stock assessment modelling, but their enduring legacy continues to influence and guide advancements in many fields of fisheries science. Although largely forgotten, their contributions to spatial modelling laid the foundation for a variety of applications in aquatic and terrestrial populations. Spatial modelling approaches are rapidly evolving beyond even the visionary scope of Beverton and Holt due to advancements in understanding of spatial population structure, collection of spatially explicit data, and statistical parameter estimation. A review of Beverton and Holt’s original movement models demonstrates that understanding the origins and basic underlying assumptions can help ensure that current models are consistent with fundamental principles. Additionally, recent simulation studies show that conforming to or revising spatial model assumptions is essential for accurate estimation. As fisheries science transitions to more complex spatial stock assessment models, understanding their conceptual development and the lessons learned by our predecessors is essential for proper model specification and application.

Improved farm management of soil organic carbon (SOC) is critical if national governments and agricultural businesses are to achieve net-zero targets. There are opportunities for farmers to secure financial benefits from carbon trading, but field measurements to establish SOC baselines for each part of a farm can be prohibitively expensive. Hence there is a potential role for spatial modelling approaches that have the resolution, accuracy, and estimates to uncertainty to estimate the carbon levels currently stored in the soil. This study uses three spatial modelling approaches to estimate SOC stocks, which are compared with measured data to a 10 cm depth and then used to determine carbon payments. The three approaches used either fine- (100 m × 100 m) or field-scale input soil data to produce either fine- or field-scale outputs across nine geographically dispersed farms. Each spatial model accurately predicted SOC stocks (range: 26.7–44.8 t ha−1) for the five case study farms where the measured SOC was lowest (range: 31.6–48.3 t ha−1). However, across the four case study farms with the highest measured SOC (range: 56.5–67.5 t ha−1), both models underestimated the SOC with the coarse input model predicting lower values (range: 39.8–48.2 t ha−1) than those using fine inputs (range: 43.5–59.2 t ha−1). Hence the use of the spatial models to establish a baseline, from which to derive payments for additional carbon sequestration, favoured farms with already high SOC levels, with that benefit greatest with the use of the coarse input data. Developing a national approach for SOC sequestration payments to farmers is possible but the economic impacts on individual businesses will depend on the approach and the accounting method.

Background and Objectives: Continuous cover forestry is of increasing importance, but operational forest growth models are still lacking. The debate is especially open if more complex spatial approaches would provide a worthwhile increase in accuracy. Our objective was to compare a nonspatial versus a spatial approach for individual Norway spruce tree growth models under single-tree selection cutting. Materials and Methods: We calibrated nonlinear mixed models using data from a long-term experiment in Finland (20 stands with 3538 individual trees for 10,238 growth measurements). We compared the use of nonspatial versus spatial predictors to describe the competitive pressure and its release after cutting. The models were compared in terms of Akaike Information Criteria (AIC), root mean square error (RMSE), and mean absolute bias (MAB), both with the training data and after cross-validation with a leave-one-out method at stand level. Results: Even though the spatial model had a lower AIC than the nonspatial model, RMSE and MAB of the two models were similar. Both models tended to underpredict growth for the highest observed values when the tree-level random effects were not used. After cross-validation, the aggregated predictions at stand level well represented the observations in both models. For most of the predictors, the use of values based on trees’ height rather than trees’ diameter improved the fit. After single-tree selection cutting, trees had a growth boost both in the first and second five-year period after cutting, however, with different predicted intensity in the two models. Conclusions: Under the research framework here considered, the spatial modeling approach was not more accurate than the nonspatial one. Regarding the single-tree selection cutting, an intervention regime spaced no more than 15 years apart seems necessary to sustain the individual tree growth. However, the model’s fixed effect parts were not able to capture the high growth of the few fastest-growing trees, and a proper estimation of site potential is needed for uneven-aged stands.

The ecology of ditches : a modeling perspective

This study aims to analyze the distribution of land value in the Administrative City of East Jakarta through spatial modeling. With the increasing concentration of economic activities in East Jakarta, the demand for land also continues to increase, which has an impact on the increase in land values in the area. This study uses market data and Tax Object Selling Value (NJOP) from 1,522 land sample points, with variables that affect land value, such as land area, building area, and distance to the village and sub-district administrative centers. The spatial analysis approach with the Geoda application is used to evaluate spatial dependencies through the Classic, Spatial Lag, and Spatial Error models. The results of the study show that the variables of building area and distance to the village have a significant influence on market prices and NJOP in East Jakarta. The spatial dependency test carried out by the Lagrange Multiplier (LM) Lag and LM Error methods indicated the existence of spatial autocorrelation in the dependent variables. In addition, the results of the Moran's I and LISA indices identified areas with high (hot spots) and low (cold spots) value patterns, which describe the spatial distribution of land values in the region. This research is expected to contribute to spatial planning and property tax policy. By understanding the factors that affect land value, the government and property developers can make more effective decisions in the management of urban areas of East Jakarta.

Many important ecological questions require accounting for spatial variation in demographic rates (e.g., survival) and population variables (e.g., abundance per unit area). However, ecologists have few spatial modelling approaches that (i) fit directly to spatially referenced data, (ii) represent population dynamics explicitly and mechanistically, and (iii) estimate parameters using rigorous statistical methods. We therefore demonstrate a new and computationally efficient approach to spatial modelling that uses random fields in place of the random variables typically used in spatially aggregated models. We adapt this approach to delay-difference dynamics to estimate the impact of fishing and natural mortality, recruitment, and individual growth on spatial population dynamics for a fish population. In particular, we develop this approach to estimate spatial variation in average production of juvenile fishes (termed recruitment), as well as annual variation in the spatial distribution of recruitment. We first use a simulation experiment to demonstrate that the spatial delay-difference model can, in some cases, explain over 50% of spatial variance in recruitment. We also apply the spatial delay-difference model to data for rex sole (Glyptocephalus zachirus) in the Gulf of Alaska and show that average recruitment (across all years) is greatest near Kodiak Island but that some years show greatest recruitment in Southeast Alaska or the western Gulf of Alaska. Using model developments and software advances presented here, we argue that future research can develop models to approximate adult movement, incorporate spatial covariates to explain annual variation in recruitment, and evaluate management procedures that use spatially explicit estimates of population abundance.

Spatial resource allocation modeling for marine protected areas design: The case of Kaomei coastal wetland

This research aims to examine the significant factors influencing economic growth in the East Nusa Tenggara Province. The estimation is carried out through a spatial regression model approach. The research variables were selected based on the economic growth model proposed by Mankiw-Romer-Weil, namely Gross Regional Domestic Product at constant prices (Y), Labor ( X1), Electricity Consumption (X2 ), Regional Original Income (X3 ), Capital Expenditure ( X4), and Schooling Duration (X5 ). The research data consists of secondary data from the year 2022 in 22 regencies/cities obtained from the Central Statistics Agency (BPS) of East Nusa Tenggara Province. Modeling with the Ordinary Least Squares (OLS) regression method resulted in three significant independent variables at α=5%: local revenue, capital expenditure, and average years of schooling. Based on diagnostic tests, there was spatial dependence in lag and error, leading to the use of Spatial Autoregressive Model (SAR) and Spatial Error Model (SEM) for spatial regression. From these spatial models, the significant independent variables influencing economic growth in the 22 regencies/cities in East Nusa Tenggara were local revenue, capital expenditure, and average years of schooling. The queen contiguity weighting was used. Based on R2 and AIC criteria, the best spatial regression model was the Spatial Error Model (SEM) with the highest R2 of 0.837610 and the smallest AIC value of 393.03. For further research, it is recommended to consider local factors that may influence economic growth, such as the sustainability of the agricultural or tourism sector, which may vary in each region.

Browsing target extraction and spatiotemporal preference mining from the complex virtual trajectories

This research aims to examine the significant factors influencing Infant Mortality Rate (IMR) in the East Nusa Tenggara Province. Estimation is carried out using a spatial regression model approach. The variables under investigation are the Infant Mortality Rate (Y), the percentage of Low Birth Weight (X1), the percentage of infants receiving breastfeeding (X2), and the percentage of deliveries assisted by medical personnel (X3). The research data consist of secondary data from the year 2022 in 22 regencies/cities obtained from the Central Statistics Agency (BPS) of the East Nusa Tenggara Province. Modeling with Ordinary Least Squares (OLS) regression produces one significant independent variable at α=5%, namely the percentage of deliveries assisted by medical personnel. Based on diagnostic tests, spatial dependence occurs at lag, indicating that the appropriate spatial regression model is the Spatial Autoregressive Model (SAR). However, a Spatial Error Model (SEM) is still used as a comparison. From these two spatial models, it is found that the significant independent variable affecting the IMR in the 22 regencies/cities in East Nusa Tenggara is the percentage of deliveries assisted by medical personnel. The weight used is queen contiguity. Based on R2 and AIC criteria, the best spatial regression model is the Spatial Autoregressive Model (SAR) because it has the highest R2 of 0.778282 and the smallest AIC of 132.518. For further research, it is recommended to consider local factors that may influence IMR, such as access to clean water, sanitation, educational level, electrification ratio, which may vary in each region.

Modeling the dynamics of landscape structure in Asia’s emerging desakota regions: a case study in Shenzhen