Pub Date : 2025-04-28DOI: 10.1016/j.spasta.2025.100903
Gianfranco Piras , Mauricio Sarrias
Spatial autoregressive binary models are well established in spatial statistics and econometric literature. Recently, different estimation methods have been proposed that account for logistic as well as probit regressions. In spatial models the choice of the spatial weighting matrix is crucial to reflect the amount of correlation in the data. This article proposes a simple -test procedure for spatial autoregressive binary model. Since the -test is a non-nested test, it can be used, among other things, to test the specification of the spatial weighting matrix. The -test is based on augmenting the null model with the predictor from the alternative model(s). After defining these predictors, we develop the theory and derive the steps for the -test. We also evaluate the finite sample properties in the context of a Monte Carlo experiment. An empirical application on firms’ decisions to reopen in the aftermath of Hurricane Katrina for New Orleans is also presented.
{"title":"A J-test for spatial autoregressive binary models","authors":"Gianfranco Piras , Mauricio Sarrias","doi":"10.1016/j.spasta.2025.100903","DOIUrl":"10.1016/j.spasta.2025.100903","url":null,"abstract":"<div><div>Spatial autoregressive binary models are well established in spatial statistics and econometric literature. Recently, different estimation methods have been proposed that account for logistic as well as probit regressions. In spatial models the choice of the spatial weighting matrix is crucial to reflect the amount of correlation in the data. This article proposes a simple <span><math><mi>J</mi></math></span>-test procedure for spatial autoregressive binary model. Since the <span><math><mi>J</mi></math></span>-test is a non-nested test, it can be used, among other things, to test the specification of the spatial weighting matrix. The <span><math><mi>J</mi></math></span>-test is based on augmenting the null model with the predictor from the alternative model(s). After defining these predictors, we develop the theory and derive the steps for the <span><math><mi>J</mi></math></span>-test. We also evaluate the finite sample properties in the context of a Monte Carlo experiment. An empirical application on firms’ decisions to reopen in the aftermath of Hurricane Katrina for New Orleans is also presented.</div></div>","PeriodicalId":48771,"journal":{"name":"Spatial Statistics","volume":"67 ","pages":"Article 100903"},"PeriodicalIF":2.1,"publicationDate":"2025-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143894998","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-25DOI: 10.1016/j.spasta.2025.100900
Arthur Machado, Francisco José A. Cysneiros, Abraão D.C. Nascimento
Solving remote sensing (RS) problems is crucial for society when it comes to environmental and climate dynamics, to name just a few examples. An efficient RS source is the use of synthetic aperture radar (SAR) to describe natural and man-made phenomena through imagery. Our approach is to understand the data behind SAR images as outcomes of random variables, and then use statistics to solve RS problems. In this paper, we consider the input of a SAR image as a random variable in regular space and describe the nature of SAR intensity (a strictly positive and asymmetric feature that is affected by speckle noise and prevents direct interpretation) using a new proposal for a log-symmetric (LOGSYM) regression model in two dimensions, the 2-D LOGSYM autoregressive moving-average (2-D LOGSYMARMA) model. Besides a discussion on the physical relationship between the proposed model and SAR intensity (mentioning that it can extend a commonly used lognormal law), we derive some mathematical properties of 2-D LOGSYMARMA: matrix-based score function and Fisher information. We discuss in detail the conditional maximum likelihood (CML) estimation for the 2-D LOGSYMARMA parameters. We conduct a Monte Carlo study to quantify the performance of the resulting estimates and to verify that the asymptotic behavior expected from CML estimators is achieved. Finally, we perform an application to real SAR data, where our proposal is applied to different types of regions – ocean, forest, and urban areas – utilizing the versatility of the log-symmetric family. Results of both artificial and real experiments show that our model is an important tool for the extraction and classification of spatial information in SAR images.
{"title":"A new regular grid-based spatial process on the log-symmetric model for speckled clutter","authors":"Arthur Machado, Francisco José A. Cysneiros, Abraão D.C. Nascimento","doi":"10.1016/j.spasta.2025.100900","DOIUrl":"10.1016/j.spasta.2025.100900","url":null,"abstract":"<div><div>Solving remote sensing (RS) problems is crucial for society when it comes to environmental and climate dynamics, to name just a few examples. An efficient RS source is the use of synthetic aperture radar (SAR) to describe natural and man-made phenomena through imagery. Our approach is to understand the data behind SAR images as outcomes of random variables, and then use statistics to solve RS problems. In this paper, we consider the input of a SAR image as a random variable in regular space and describe the nature of SAR intensity (a strictly positive and asymmetric feature that is affected by speckle noise and prevents direct interpretation) using a new proposal for a log-symmetric (LOGSYM) regression model in two dimensions, the 2-D LOGSYM autoregressive moving-average (2-D LOGSYMARMA) model. Besides a discussion on the physical relationship between the proposed model and SAR intensity (mentioning that it can extend a commonly used lognormal law), we derive some mathematical properties of 2-D LOGSYMARMA: matrix-based score function and Fisher information. We discuss in detail the conditional maximum likelihood (CML) estimation for the 2-D LOGSYMARMA parameters. We conduct a Monte Carlo study to quantify the performance of the resulting estimates and to verify that the asymptotic behavior expected from CML estimators is achieved. Finally, we perform an application to real SAR data, where our proposal is applied to different types of regions – ocean, forest, and urban areas – utilizing the versatility of the log-symmetric family. Results of both artificial and real experiments show that our model is an important tool for the extraction and classification of spatial information in SAR images.</div></div>","PeriodicalId":48771,"journal":{"name":"Spatial Statistics","volume":"67 ","pages":"Article 100900"},"PeriodicalIF":2.1,"publicationDate":"2025-04-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143882136","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-25DOI: 10.1016/j.spasta.2025.100902
Xinyi Lu , Andee Kaplan , Yoichiro Kanno , George Valentine , Jacob M. Rash , Mevin Hooten
Spatial stream networks (SSN) models characterize correlated ecological processes in dendritic ecosystems. Conventional SSN models rely on pre-processed stream networks and point-to-point hydrologic distances. However, this data processing may be labor-intensive and time-consuming over large spatial domains. Therefore, we propose to infer the functional connectivity of stream networks stochastically. Our physically-guided model utilizes the knowledge that water flows from high elevation to low elevation, and flow rate typically increases when two tributaries merge. We also leverage the hierarchical branching architecture of dendritic networks to alleviate computing and reduce uncertainty. Spatial autoregressive models composed of inferred SSNs propagate stochasticity between network connectivity and dynamic ecological processes in a Bayesian framework. We show in simulated examples that our mechanistic model facilitated learning about the functional network and enhanced predictive performance. We also demonstrate our approach in a large-scale case study using native brook trout (Salvelinus fontinalis) count data. A population model based on our stochastic SSN outperformed that with a conventional SSN in predicting abundance and expedited the analysis by circumventing data processing.
{"title":"Stochastic spatial stream networks for scalable inferences of riverscape processes","authors":"Xinyi Lu , Andee Kaplan , Yoichiro Kanno , George Valentine , Jacob M. Rash , Mevin Hooten","doi":"10.1016/j.spasta.2025.100902","DOIUrl":"10.1016/j.spasta.2025.100902","url":null,"abstract":"<div><div>Spatial stream networks (SSN) models characterize correlated ecological processes in dendritic ecosystems. Conventional SSN models rely on pre-processed stream networks and point-to-point hydrologic distances. However, this data processing may be labor-intensive and time-consuming over large spatial domains. Therefore, we propose to infer the functional connectivity of stream networks stochastically. Our physically-guided model utilizes the knowledge that water flows from high elevation to low elevation, and flow rate typically increases when two tributaries merge. We also leverage the hierarchical branching architecture of dendritic networks to alleviate computing and reduce uncertainty. Spatial autoregressive models composed of inferred SSNs propagate stochasticity between network connectivity and dynamic ecological processes in a Bayesian framework. We show in simulated examples that our mechanistic model facilitated learning about the functional network and enhanced predictive performance. We also demonstrate our approach in a large-scale case study using native brook trout (<em>Salvelinus fontinalis</em>) count data. A population model based on our stochastic SSN outperformed that with a conventional SSN in predicting abundance and expedited the analysis by circumventing data processing.</div></div>","PeriodicalId":48771,"journal":{"name":"Spatial Statistics","volume":"67 ","pages":"Article 100902"},"PeriodicalIF":2.1,"publicationDate":"2025-04-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143898418","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-24DOI: 10.1016/j.spasta.2025.100904
Nicolas Coloma, William Kleiber
Statistical modeling and interpolation of space–time processes has gained increasing relevance over the last few years. However, real world data often exhibit characteristics that challenge conventional methods such as nonstationarity and temporal misalignment. For example, high frequency solar irradiance data are typically observed at fine temporal scales, but at sparse spatial sampling, so space–time interpolation is necessary to support solar energy studies. The nonstationarity and phase misalignment of such data challenges extant approaches. We propose random elastic space–time (REST) prediction, a novel method that addresses temporally-varying phase misalignment by combining elastic alignment and conventional kriging techniques. Moreover, uncertainty in both amplitude and phase alignment can be readily quantified in a conditional simulation framework, whereas conventional space–time methods only address amplitude uncertainty. We illustrate our approach on a challenging solar irradiance dataset, where our method demonstrates superior predictive distributions compared to existing geostatistical and functional data analytic techniques.
{"title":"Random elastic space–time (REST) prediction","authors":"Nicolas Coloma, William Kleiber","doi":"10.1016/j.spasta.2025.100904","DOIUrl":"10.1016/j.spasta.2025.100904","url":null,"abstract":"<div><div>Statistical modeling and interpolation of space–time processes has gained increasing relevance over the last few years. However, real world data often exhibit characteristics that challenge conventional methods such as nonstationarity and temporal misalignment. For example, high frequency solar irradiance data are typically observed at fine temporal scales, but at sparse spatial sampling, so space–time interpolation is necessary to support solar energy studies. The nonstationarity and phase misalignment of such data challenges extant approaches. We propose random elastic space–time (REST) prediction, a novel method that addresses temporally-varying phase misalignment by combining elastic alignment and conventional kriging techniques. Moreover, uncertainty in both amplitude and phase alignment can be readily quantified in a conditional simulation framework, whereas conventional space–time methods only address amplitude uncertainty. We illustrate our approach on a challenging solar irradiance dataset, where our method demonstrates superior predictive distributions compared to existing geostatistical and functional data analytic techniques.</div></div>","PeriodicalId":48771,"journal":{"name":"Spatial Statistics","volume":"67 ","pages":"Article 100904"},"PeriodicalIF":2.1,"publicationDate":"2025-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143886951","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-23DOI: 10.1016/j.spasta.2025.100901
Duncan Lee
Disease mapping models estimate the spatio-temporal variation in population-level disease risks or rates across a set of areal units for time periods, aiming to identify temporal trends and spatial hotspots. Highly parameterised Bayesian hierarchical models with over random effects are commonly used to estimate this spatio-temporal variation, which are assigned autoregressive and conditional autoregressive prior distributions. These models work well when there are tens of thousands of data points, but are likely to be computationally burdensome when this rises to hundreds of thousands or above. This paper proposes a computationally efficient alternative, which can fit a range of spatio-temporal disease trends almost as well as existing highly parameterised models but only takes around 5% to 40% of the time to implement. It achieves this by modelling the average spatial and temporal trends in the data with autoregressive type random effects, which are augmented by an observation-driven process using functions of earlier data as additional covariates in the model. The efficacy of this methodology is tested by simulation, before being applied to the motivating study that estimates the spatio-temporal trends in asthma, cancer, coronary heart and chronic obstructive pulmonary disease prevalences for small areas over years in England.
{"title":"Computationally efficient spatio-temporal disease mapping for big data","authors":"Duncan Lee","doi":"10.1016/j.spasta.2025.100901","DOIUrl":"10.1016/j.spasta.2025.100901","url":null,"abstract":"<div><div>Disease mapping models estimate the spatio-temporal variation in population-level disease risks or rates across a set of <span><math><mi>K</mi></math></span> areal units for <span><math><mi>N</mi></math></span> time periods, aiming to identify temporal trends and spatial hotspots. Highly parameterised Bayesian hierarchical models with over <span><math><mrow><mi>K</mi><mi>N</mi></mrow></math></span> random effects are commonly used to estimate this spatio-temporal variation, which are assigned autoregressive and conditional autoregressive prior distributions. These models work well when there are tens of thousands of data points, but are likely to be computationally burdensome when this rises to hundreds of thousands or above. This paper proposes a computationally efficient alternative, which can fit a range of spatio-temporal disease trends almost as well as existing highly parameterised models but only takes around 5% to 40% of the time to implement. It achieves this by modelling the average spatial and temporal trends in the data with autoregressive type random effects, which are augmented by an observation-driven process using functions of earlier data as additional covariates in the model. The efficacy of this methodology is tested by simulation, before being applied to the motivating study that estimates the spatio-temporal trends in asthma, cancer, coronary heart and chronic obstructive pulmonary disease prevalences for <span><math><mrow><mi>K</mi><mo>=</mo><mn>32</mn><mo>,</mo><mn>751</mn></mrow></math></span> small areas over <span><math><mrow><mi>N</mi><mo>=</mo><mn>13</mn></mrow></math></span> years in England.</div></div>","PeriodicalId":48771,"journal":{"name":"Spatial Statistics","volume":"67 ","pages":"Article 100901"},"PeriodicalIF":2.1,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143873638","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-15DOI: 10.1016/j.spasta.2025.100893
Sjoerd Hermes , Joost van Heerwaarden , Pariya Behrouzi
Within the statistical literature, a significant gap exists in methods capable of modelling asymmetric multivariate spatial effects that elucidate the relationships underlying complex spatial phenomena. For such a phenomenon, observations at any location are expected to arise from a combination of within- and between-location effects, where the latter exhibit asymmetry. This asymmetry is represented by heterogeneous spatial effects between locations pertaining to two different categories, that is, a feature inherent to each location in the data, such that based on the feature label, asymmetric spatial relations are postulated between neighbouring locations with different labels. Our novel approach synergises the principles of multivariate spatial autoregressive models and the Gaussian graphical model. This synergy enables us to effectively address the gap by accommodating asymmetric spatial relations, overcoming the usual constraints in spatial analyses. However, the resulting flexibility comes at a cost: the spatial effects are not identifiable without either prior knowledge of the underlying phenomenon or additional parameter restrictions. Using a Bayesian-estimation framework, the model performance is assessed in a simulation study. We apply the model on intercropping data, where spatial effects between different crops are unlikely to be symmetric, in order to illustrate the usage of the proposed methodology. An R package containing the proposed methodology can be found on https://CRAN.R-project.org/package=SAGM.
{"title":"A spatial autoregressive graphical model","authors":"Sjoerd Hermes , Joost van Heerwaarden , Pariya Behrouzi","doi":"10.1016/j.spasta.2025.100893","DOIUrl":"10.1016/j.spasta.2025.100893","url":null,"abstract":"<div><div>Within the statistical literature, a significant gap exists in methods capable of modelling asymmetric multivariate spatial effects that elucidate the relationships underlying complex spatial phenomena. For such a phenomenon, observations at any location are expected to arise from a combination of within- and between-location effects, where the latter exhibit asymmetry. This asymmetry is represented by heterogeneous spatial effects between locations pertaining to two different categories, that is, a feature inherent to each location in the data, such that based on the feature label, asymmetric spatial relations are postulated between neighbouring locations with different labels. Our novel approach synergises the principles of multivariate spatial autoregressive models and the Gaussian graphical model. This synergy enables us to effectively address the gap by accommodating asymmetric spatial relations, overcoming the usual constraints in spatial analyses. However, the resulting flexibility comes at a cost: the spatial effects are not identifiable without either prior knowledge of the underlying phenomenon or additional parameter restrictions. Using a Bayesian-estimation framework, the model performance is assessed in a simulation study. We apply the model on intercropping data, where spatial effects between different crops are unlikely to be symmetric, in order to illustrate the usage of the proposed methodology. An R package containing the proposed methodology can be found on <span><span>https://CRAN.R-project.org/package=SAGM</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":48771,"journal":{"name":"Spatial Statistics","volume":"67 ","pages":"Article 100893"},"PeriodicalIF":2.1,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143839305","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-05DOI: 10.1016/j.spasta.2025.100898
Jakub J. Pypkowski, Adam M. Sykulski, James S. Martin
Several hypothesis testing methods have been proposed to validate the assumption of isotropy in spatial point patterns. A majority of these methods are characterised by an unknown distribution of the test statistic under the null hypothesis of isotropy. Parametric approaches to approximating the distribution involve simulation of patterns from a user-specified isotropic model. Alternatively, nonparametric replicates of the test statistic under isotropy can be used to waive the need for specifying a model. In this paper, we first present a general framework which allows for the integration of a selected nonparametric replication method into isotropy testing. We then conduct a large simulation study comprising application-like scenarios to assess the performance of tests with different parametric and nonparametric replication methods. In particular, we explore distortions in test size and power caused by model misspecification, and demonstrate the advantages of nonparametric replication in such scenarios.
{"title":"Isotropy testing in spatial point patterns: nonparametric versus parametric replication under misspecification","authors":"Jakub J. Pypkowski, Adam M. Sykulski, James S. Martin","doi":"10.1016/j.spasta.2025.100898","DOIUrl":"10.1016/j.spasta.2025.100898","url":null,"abstract":"<div><div>Several hypothesis testing methods have been proposed to validate the assumption of isotropy in spatial point patterns. A majority of these methods are characterised by an unknown distribution of the test statistic under the null hypothesis of isotropy. Parametric approaches to approximating the distribution involve simulation of patterns from a user-specified isotropic model. Alternatively, nonparametric replicates of the test statistic under isotropy can be used to waive the need for specifying a model. In this paper, we first present a general framework which allows for the integration of a selected nonparametric replication method into isotropy testing. We then conduct a large simulation study comprising application-like scenarios to assess the performance of tests with different parametric and nonparametric replication methods. In particular, we explore distortions in test size and power caused by model misspecification, and demonstrate the advantages of nonparametric replication in such scenarios.</div></div>","PeriodicalId":48771,"journal":{"name":"Spatial Statistics","volume":"67 ","pages":"Article 100898"},"PeriodicalIF":2.1,"publicationDate":"2025-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143799115","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-05DOI: 10.1016/j.spasta.2025.100896
Pilar García-Soidán , Tomás R. Cotos-Yáñez
The estimation of the spatial quantiles provides information on the thresholds of a spatial variable. This methodology is particularly appealing for its application to data of pollutants, so as to assess their level of risk. A spatial quantile can be approximated through different mechanisms, proposed in the statistics literature, although these approaches suffer from several drawbacks, regarding their lack of optimality or the fact of not leading to direct approximations. Thus, the current work introduces alternative procedures, which try to overcome the aforementioned issues by employing order statistics, similarly as done for independent data. With this aim, the available observations are appropriately transformed to yield a sample of the process at each target site, so that the data obtained are then ordered and used to derive the spatial quantile at the corresponding location. The new methodology can be directly applied to data from processes that are either stationary or that deviate from this condition for a non-constant trend and, additionally, it can be even extended to heteroscedastic data. Simulation studies under different scenarios have been accomplished, whose results show the adequate performance of the proposed estimators. A further step of this research is the application of the new approaches to data of nitrogen dioxide concentrations, to exemplify the potential of the quantile estimates to check the thresholds of a pollutant at a specific moment, as well as their evolution over time.
{"title":"Nonparametric approaches for direct approximation of the spatial quantiles","authors":"Pilar García-Soidán , Tomás R. Cotos-Yáñez","doi":"10.1016/j.spasta.2025.100896","DOIUrl":"10.1016/j.spasta.2025.100896","url":null,"abstract":"<div><div>The estimation of the spatial quantiles provides information on the thresholds of a spatial variable. This methodology is particularly appealing for its application to data of pollutants, so as to assess their level of risk. A spatial quantile can be approximated through different mechanisms, proposed in the statistics literature, although these approaches suffer from several drawbacks, regarding their lack of optimality or the fact of not leading to direct approximations. Thus, the current work introduces alternative procedures, which try to overcome the aforementioned issues by employing order statistics, similarly as done for independent data. With this aim, the available observations are appropriately transformed to yield a sample of the process at each target site, so that the data obtained are then ordered and used to derive the spatial quantile at the corresponding location. The new methodology can be directly applied to data from processes that are either stationary or that deviate from this condition for a non-constant trend and, additionally, it can be even extended to heteroscedastic data. Simulation studies under different scenarios have been accomplished, whose results show the adequate performance of the proposed estimators. A further step of this research is the application of the new approaches to data of nitrogen dioxide concentrations, to exemplify the potential of the quantile estimates to check the thresholds of a pollutant at a specific moment, as well as their evolution over time.</div></div>","PeriodicalId":48771,"journal":{"name":"Spatial Statistics","volume":"67 ","pages":"Article 100896"},"PeriodicalIF":2.1,"publicationDate":"2025-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143783992","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Unlike a geographically weighted regression (GWR), a similarity and geographically weighted regression (SGWR) calculates weights using attribute and geographic similarities, thereby effectively improving the accuracy of the model. Nevertheless, SGWR does not set an attribute similarity bandwidth. This leads to its inability to measure the scale of variation of spatial processes in feature space. In addition, owing to the solving method used, SGWR can get stuck in local optima. To address these issues, this study proposed an improved similarity and geographically weighted regression model (SGWR-GD) that adds bandwidth to the attribute similarity kernel function. This parameter gives SGWR-GD the ability to measure the scale of change of spatial processes in the attribute dimension and thus enhances the flexibility of modelling. When solving the model, SGWR-GD first calculated the gradient of the model's Modified Akaike Information Criterion (AICc) with respect to the two bandwidths and the impact ratio. Subsequently, the optimal global solution of the model was obtained based on a gradient descent algorithm with box constraints. SGWR-GD and SGWR were applied to five different datasets and the accuracies of their fitting results were compared. SGWR-GD significantly improved the accuracy of the model compared to SGWR. In addition, the SGWR-GD stably determined the global optimal solution for each parameter. Simultaneously, the distribution of local residuals was also more stable.
与地理加权回归（GWR）不同，相似度和地理加权回归（SGWR）利用属性和地理相似度计算权重，从而有效地提高了模型的准确性。但是，SGWR不设置属性相似带宽。这导致其无法测量特征空间中空间过程的变化尺度。此外，由于采用的求解方法，SGWR可能陷入局部最优。为了解决这些问题，本研究提出了一种改进的相似度和地理加权回归模型（SGWR-GD），该模型在属性相似度核函数中增加了带宽。该参数使SGWR-GD能够在属性维度上度量空间过程的变化尺度，从而增强建模的灵活性。在求解模型时，SGWR-GD首先计算模型的修正赤池信息准则（Modified Akaike Information Criterion, AICc）相对于两个带宽和冲击比的梯度。随后，基于带框约束的梯度下降算法，得到了模型的全局最优解。将SGWR- gd和SGWR应用于5个不同的数据集，并比较了它们的拟合结果的精度。与SGWR相比，SGWR- gd显著提高了模型的精度。此外，SGWR-GD稳定地确定了各参数的全局最优解。同时，局部残差的分布也更加稳定。
{"title":"Similarity and geographically weighted regression considering spatial scales of features space","authors":"Shifeng Yu , Xiaoyu Hu , Yehua Sheng , Chenmeng Zhao","doi":"10.1016/j.spasta.2025.100897","DOIUrl":"10.1016/j.spasta.2025.100897","url":null,"abstract":"<div><div>Unlike a geographically weighted regression (GWR), a similarity and geographically weighted regression (SGWR) calculates weights using attribute and geographic similarities, thereby effectively improving the accuracy of the model. Nevertheless, SGWR does not set an attribute similarity bandwidth. This leads to its inability to measure the scale of variation of spatial processes in feature space. In addition, owing to the solving method used, SGWR can get stuck in local optima. To address these issues, this study proposed an improved similarity and geographically weighted regression model (SGWR-GD) that adds bandwidth to the attribute similarity kernel function. This parameter gives SGWR-GD the ability to measure the scale of change of spatial processes in the attribute dimension and thus enhances the flexibility of modelling. When solving the model, SGWR-GD first calculated the gradient of the model's Modified Akaike Information Criterion (AICc) with respect to the two bandwidths and the impact ratio. Subsequently, the optimal global solution of the model was obtained based on a gradient descent algorithm with box constraints. SGWR-GD and SGWR were applied to five different datasets and the accuracies of their fitting results were compared. SGWR-GD significantly improved the accuracy of the model compared to SGWR. In addition, the SGWR-GD stably determined the global optimal solution for each parameter. Simultaneously, the distribution of local residuals was also more stable.</div></div>","PeriodicalId":48771,"journal":{"name":"Spatial Statistics","volume":"67 ","pages":"Article 100897"},"PeriodicalIF":2.1,"publicationDate":"2025-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143760349","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-28DOI: 10.1016/j.spasta.2025.100895
Frederic Schoenberg
Current methods for evaluating earthquake forecasts, such as the -test, -test, or log-likelihood score, typically do not disproportionately reward a model for more accurately forecasting the largest events, or disproportionately punish a model for less accurately forecasting the largest events. However, since the largest earthquakes are by far the most destructive and therefore of most interest to practitioners, in many circumstances, a weighted likelihood score may be more useful. Here, we propose various weighted measures, weighting each earthquake by some function of its magnitude, such as potency-weighted log-likelihood, and consider their properties. The proposed methods are applied to a catalog of earthquakes in the Western United States.
{"title":"Magnitude-weighted goodness-of-fit scores for earthquake forecasting","authors":"Frederic Schoenberg","doi":"10.1016/j.spasta.2025.100895","DOIUrl":"10.1016/j.spasta.2025.100895","url":null,"abstract":"<div><div>Current methods for evaluating earthquake forecasts, such as the <span><math><mi>N</mi></math></span>-test, <span><math><mi>L</mi></math></span>-test, or log-likelihood score, typically do not disproportionately reward a model for more accurately forecasting the largest events, or disproportionately punish a model for less accurately forecasting the largest events. However, since the largest earthquakes are by far the most destructive and therefore of most interest to practitioners, in many circumstances, a weighted likelihood score may be more useful. Here, we propose various weighted measures, weighting each earthquake by some function of its magnitude, such as potency-weighted log-likelihood, and consider their properties. The proposed methods are applied to a catalog of earthquakes in the Western United States.</div></div>","PeriodicalId":48771,"journal":{"name":"Spatial Statistics","volume":"67 ","pages":"Article 100895"},"PeriodicalIF":2.1,"publicationDate":"2025-03-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143746591","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号