Geographically Weighted Regression Research Articles

Dengue hemorrhagic fever prediction in coastal area using geographically weighted regression

Dengue hemorrhagic fever (DHF) is still a health problem globally, including in Indonesia. Geographical and climatic conditions in coastal areas are different from other areas, which may impact differences in environmental risk factors for dengue. This study aims to create a prediction model for the incidence of DHF in coastal areas. The research was conducted in Bantul Regency, Indonesia, involving data from 2015-2019. Dengue incidence data were collected from the health office. Climatic data were from climatology station. Data on altitude and shoreline distance were obtained by geographic information system (GIS) processing. Population density and wide settlement area are obtained from the Bureau of Statistics. The geographically weighted regression (GWR) analysis was carried out using GWR4. The results showed that GWR with a weighting of Fixed Bi-Square Kernel obtained an R2 value of 0.7768, better than the global model (R2 0.5254). It indicates that DHF (Y) in Bantul Regency is 77.68% determined by population density (X1), altitude (X2), settlement area (X3), shoreline distance (X4) and rainfall (X5) and the remaining 22.32% are influenced by other variables which are not investigated. Geographically, the predictor variables explain the DHF incidence with a strong category in the central region, and weak in coastal area.

Construction of A Geographically Weighted Nonparametric Regression Model Fit Test

One of the approach of Geographically Weighted Regression (GWR) models is the Geographically Weighted Nonparametric Regression (GWNR) has more parameters than the GWR model. Models with more parameters usually have better match values, which is an advantage, while models with fewer parameters have the advantage of being easier to use and interpret. However, a model with more parameters should be used if it is proven to be significantly superior. Therefore, the purpose of this study was to develop a hypothesis test of goodness of fit test for GWNR model. The goodness of fit test was performed for the real data. We found that the GWNR model was more suitable than the mixed nonparametric regression model. Some highlights of the proposed method are:• A new model for GWR to overcome the unknown regression function by using mixed estimator spline truncated and fourier series at nonparametric regression• Goodness of fit for GWNR to testing the model fit between the mixed nonparametric regression model and GWNR• Applied goodness of fit test to poverty data in Sulawesi Island and infant mortality in East Java

Generalized Geographically Weighted Regression Model within a Modularized Bayesian Framework.

Geographically weighted regression (GWR) models handle geographical dependence through a spatially varying coefficient model and have been widely used in applied science, but its general Bayesian extension is unclear because it involves a weighted log-likelihood which does not imply a probability distribution on data. We present a Bayesian GWR model and show that its essence is dealing with partial misspecification of the model. Current modularized Bayesian inference models accommodate partial misspecification from a single component of the model. We extend these models to handle partial misspecification in more than one component of the model, as required for our Bayesian GWR model. Information from the various spatial locations is manipulated via a geographically weighted kernel and the optimal manipulation is chosen according to a Kullback-Leibler (KL) divergence. We justify the model via an information risk minimization approach and show the consistency of the proposed estimator in terms of a geographically weighted KL divergence.

Spatial Heterogeneity Analysis of Factors on Commuting Carbon Emissions: Evidence from the Shenzhen Metropolitan Area in China

As an essential part of daily life, commuting produces considerable carbon emissions and is currently receiving increased amounts of attention. Comprehensive explorations of carbon emissions and the spatial distribution of their effects based on previous studies are lacking. First, we adopt stepwise regression and geographically weighted regression (GWR) to explore the diverse impacts of carbon emissions on the different layers of metropolitan areas, employing factors from the perspectives of socioeconomics, transportation services, and road networks. Our findings show that optimizing the road network structure could be an effective approach to reducing carbon emissions from commuting, especially in the periphery of metropolitan areas. In addition, the mixed use of land contributes to reducing carbon emissions from commuting, especially in the central areas. Thus, the coverage of public transport should be improved, especially in peripheral regions. Policymakers should monitor the spatial heterogeneity of variables and develop suitable policies to adapt to the conditions of the different layers of metropolitan areas.

Estimating Forest Aboveground Biomass Using a Combination of Geographical Random Forest and Empirical Bayesian Kriging Models

Accurately estimating forest aboveground biomass (AGB) is imperative for comprehending carbon cycling, calculating carbon budgets, and formulating sustainable forest management plans. Currently, random forest (RF) and other machine learning models are widely used to estimate forest AGB, as they can effectively handle nonlinear relationships. However, by constructing a global model using all the samples collected from a study area, these models fail to account for the spatial heterogeneity in the AGB and cannot correct the prediction biases, thereby constraining the estimation accuracy. To overcome these limitations, we proposed a novel approach termed geographical random forest and empirical Bayesian kriging (GRFEBK). This hybrid model combines the localized modeling capability of geographical random forest (GRF) with the bias correction strength of empirical Bayesian kriging (EBK). GRF adapts RF to account for the spatial heterogeneity of the AGB, while EBK utilizes the spatial autocorrelation of residuals to correct the prediction deviations. This study was conducted in Hainan Island, utilizing spectral bands, vegetation indices, tasseled cap components derived from Landsat-8 imagery, backscattering coefficients from ALOS-2 synthetic aperture radar, topographic features, and the forest canopy height as the explanatory variables. A total of 195 forest aboveground biomass (AGB) samples were collected for modeling and assessing the predictive accuracy. The results demonstrate that, among the tested models, including GRFEBK, RF, support vector machine (SVM), k-nearest neighbor (KNN), geographically weighted regression (GWR), GRF, and EBK, GRFEBK attains the highest R2 (0.78) and the lowest RMSE (36.04 Mg/ha) and RRMSE (22.87%), significantly outperforming the conventional models and using GRF or EBK alone. These results demonstrate that by accounting for local non-stationarity in AGB and correcting prediction biases, GRFEBK achieves significantly higher accuracy than conventional RF and other models. While the results are promising, the computational cost of GRFEBK and its performance under varying geographical conditions warrant further investigation at larger scales to assess its broader applicability. Nevertheless, GRFEBK provides an innovative and more reliable approach for accurate forest AGB estimation with great potential to support global forest resource monitoring.

Sedimentary contamination study in the Great Chao Phraya River watershed, Thailand, using geographically weighted regression

ABSTRACT This research investigated the spatial variations of heavy metals and nutrients in the sediment of the Great Chao Phraya River network of Thailand using geographically weighted regression (GWR). The performance of GWR in explaining the spatiality of selected contaminants using single and multiple predictors was compared and discussed, and a suitable approach was recommended for sediment study. Extensive explanatory variables, including land use type and topographic, socio-economic, and meteorological indicators were utilized in the geospatial regressions of river sediment contaminants at the sub-catchment scale. In the case of simple regression, GWR outperformed the ordinary least square (OLS) regressions with global R 2 (range 0.34–0.72 for GWR and 0–0.24 for OLS, respectively), over 2-fold of which was derived by OLS. Similar observations were obtained in multiple regressions, where GWR offered global R 2 values between 0.27 and 0.6, higher than those that were provided by OLS (i.e. 0.12–0.44). Furthermore, multiple GWR model presented higher local explanatory power and performance with significantly increased local R 2 (up to 51%) compared with simple GWRs (p ≤ 0.05). It is recommended to combine the applications of simple and multiple GWR models to understand the effects of selected explanatory predictors on sediment contaminations from different perspectives.

Geospatial epidemiology of coronary artery disease treated with percutaneous coronary intervention in Crete, Greece

Coronary artery disease (CAD) constitutes a leading cause of morbidity and mortality worldwide. Percutaneous coronary intervention (PCI) is indicated in a significant proportion of CAD patients, either to improve prognosis or to relieve symptoms not responding to optimal medical therapy. Thus the annual number of patients undergoing PCI in a given geographical area could serve as a surrogate marker of the total CAD burden there. The aim of this study was to analyze the potential, spatial patterns of PCItreated CAD patients in Crete. We evaluated data from all patients subjected to PCI at the island's sole reference centre for cardiac catheterization within a 4-year study period (2013-2016). The analysis focused on regional variations of yearly PCI rates, as well as on the effect of several clinical parameters on the severity of the coronary artery stenosis treated with PCI across Crete. A spatial database within the ArcGIS environment was created and an analysis carried out based on global and local regression using ordinary least squares (OLS) and geographically weighted regression (GWR), respectively. The results revealed significant inter-municipality variation in PCI rates and thus potentially CAD burden, while the degree and direction of correlation between key clinical factors to coronary stenosis severity demonstrated specific geographical patterns. These preliminary results could set the basis for future research, with the ultimate aim to facilitate efficient healthcare strategies planning.

Studying long term relationship between carbon Emissions, Soil, and climate Change: Insights from a global Earth modeling Framework

The persistent increase in greenhouse gas (GHG) emissions, notably carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O), since the mid-20th century has been a key driver of significant climate alterations. This study investigates the complex feedback mechanisms that both influence and are influenced by global climate dynamics, soil processes, and GHG emissions. Our statistical approach incorporates correlation measures, highlighting the limitations of such analyses, namely their inability to confirm causality, sensitivity to outliers, and the exclusive capture of linear relationships. Geographically Weighted Regression (GWR) models reveal spatial variations in the relationship between environmental factors and GHGs, while Path Analysis aids in delineating direct and indirect influences among variables. The research pinpoints significant spatial heterogeneity in the impacts of economic and environmental factors on GHGs, underscoring the necessity of localized strategies for climate change mitigation and sustainable land management. This study also identifies potential threats to agricultural productivity due to soil degradation, which hinder climate adaptation efforts. Our findings advocate for a concerted global response to reduce GHG emissions and address the challenges posed by the interplay of climate change, soil dynamics, and GHG emissions.

Seasonal changes in ecosystem health and their spatial relationship with landscape structure in china's Loess plateau

Environmental managers have sought to optimize landscape structure to maintain regional ecosystem health (EH). However, we still lack a full understanding of the effects of changing landscape structure on EH, especially in terms of shorter time scales (e.g., monthly or seasonal scales) and spatial heterogeneity. To address this gap, this study used the vigor-organization-resilience-services (VORS) model to assess the annual and seasonal variation of EH in the Loess Plateau (LP) during 2000–2015, and then applied the geographically weighted regression (GWR) model to investigate the spatial effects of different aspects (i.e., landscape composition and configuration) of the landscape structure change on EH over time. Our results showed that the EH level in the LP from 2000 to 2015 displayed temporal non-stationarity at annual and seasonal scales, and a spatial pattern of decreasing from southeast to northwest. The effects of landscape structure on EH had obvious spatial and temporal heterogeneity, depending on the local landscape composition and configuration. In general, landscape configuration exerted a greater impact on EH than landscape composition apart from Shannon’s diversity index and percentage of forestland area, but the intensity and direction of the impact displayed higher temporally instability and therefore should receive more attention. Furthermore, regulation of the dominant patch size, landscape connectivity and forest spatial spacing is the optimal choice for regional ecosystem management. The results suggest that EH responses to changes in landscape structure in the LP vary by time and place, and that the integrating temporal variation and spatial characteristics can help decision-makers make precise spatial arrangements and temporal adjustments to keep the negative ecological impacts within reasonable limits.

The spatial variation and drivers of soil δ13C in an alpine-gorge region on the eastern Qinghai-Tibetan Plateau

Soil stable carbon isotope (soil δ13C) can reflect soil carbon metabolic processes and record environmental and vegetation change information, so mapping the spatial distribution of soil δ13C (isoscapes of soil δ13C) can help us better understand the spatial variability of ecosystem carbon cycles. However, efficient approaches for obtaining the isoscapes of soil δ13C remain challenging, especially in mountainous areas with complex terrain. A total of 150 soil samples were collected and their soil δ13C was measured to investigate the spatial variation of soil δ13C in an alpine-gorge region on the eastern Qinghai-Tibetan Plateau which is characterized by complex terrain. Four prediction methods included ordinary kriging interpolation, multiple linear regression, random forest regression, and geographically weighted regression (GWR) were selected and compared to find the best prediction model for mapping the isoscapes of soil δ13C. The pathways of vegetation, topography, soil, and spatial factors influencing the spatial variability of soil δ13C were explored using variance partitioning analysis and structural equation model. The soil δ13C was significantly different among vegetation types, and ranged from −27.155‰ to −9.647‰. The spatial heterogeneity of soil δ13C showed moderate variation, and was dominated by spatial structural factors. The GWR model had higher prediction accuracy in the modeling soil δ13C in comparison to other models. Soil carbon to nitrogen ratio and normalized difference vegetation index were the main factors determining the spatial variability of soil δ13C, and other topographic and soil factors indirectly regulated the spatial distribution of soil δ13C by influencing these two factors. Our results provided a useful tool (GWR model) for mapping the isoscapes of soil δ13C and explored the factors controlling the spatial variability of soil δ13C in the alpine-gorge region on the eastern Qinghai-Tibetan Plateau, which are important for an in-depth understanding of soil carbon cycling in mountain ecosystems.

Enhancing geospatial prediction models with feature engineering from road networks: a graph-driven approach

Traditional geospatial predictive models for property valuation have naturally relied on coordinates as well as ‘hedonic’ (internal and external) features. In particular, location-centric methods such as Geographically Weighted Regression (GWR) and Kriging have focused on intrinsic target characteristics together with distances between individual targets. However, especially in the context of heavily urbanized areas, these approaches might overlook crucial aspects arising from the underlying topological structure that presents itself in such areas. Concretely, in this work, we focus on the structure arising from the road network connecting properties. We introduce a novel though straightforward technique for feature engineering based on graphs constructed on a road network. We then extract relevant features from these and utilize those as inputs for predictive models and assess their performance benefits when used together with a variety of both well-known geospatial models and state-of-art machine learning models. To this end, we present an exhaustive experiment using four different real-life data sets across various regions and exhibiting sizes outperforming many comparative works in the field. Our findings reveal that our feature engineering approach offers significant improvements in predictive performance. Finally, we apply Shapley values as an interpretability technique to confirm the reliability and effectiveness of our approach.

Evolution Model and Driving Mechanism of Urban Logistics Land: Evidence from the Yangtze River Delta

Logistics land is the spatial carrier for the development of logistics enterprises. Its evolution mode and driving mechanism determine the level of high-quality development of the logistics industry, and serve as an important basis for urban planning and territorial spatial planning. This study introduced a Boston consulting group (BCG) matrix and geographically weighted regression (GWR) spatial econometric models to carry out empirical research on the Yangtze River Delta (YRD), in an effort to provide scientific information for evidence-based decision-making by governments and enterprises. The scale and ratio of logistics land (LLS and LLR) in the YRD showed significant spatial heterogeneity and autocorrelation, cities with large logistics land use converging from clusters to belts from 2000 to 2020, and agglomerations with high logistics land ratio (LLR) migrating from inland to coastal areas. Diversified models of logistics land evolution also emerged, such as high scale–high speed cities, low scale–low speed cities, high scale–low speed cities, and low scale–high speed cities. In addition, the driving mechanism of LLS and LLR was very complex, with a great difference in the intensity, nature and spatial effects of the influence of different factors. The inspiration from empirical case studies is urgent to revise the planning norms and clarify the LLS and LLR control standards for logistics land use. Meanwhile, the synergistic development target of the logistics industry in the new era is changing from the manufacturing industry to the commerce and trade industry; the establishment of planning zoning and the designing of differentiated management policies significantly improve the planning applicability.

Spatial downscaling of SMAP soil moisture estimation using multiscale geographically weighted regression during SMAPVEX16

Passive microwave remote sensing missions such as Soil Moisture Active Passive (SMAP) are widely used for global soil moisture (SM) estimation. But the low spatial resolution of passive microwave SM products limits their applications at regional and local scales (normally 1–10 km). Using high-resolution auxiliary data to downscale passive microwave SM products is the primary way to obtain high-resolution SM data. However, the existing SM downscaling methods do not entirely account for the scale differences in the impact of various variables on the distribution of SM. To solve this problem, this paper introduced MGWR (Multiscale Geographically Weighted Regression) as a novel way to analyze the scale differences of normalized difference vegetation index (NDVI) and land surface temperature (LST) on the spatial pattern of SM. Accordingly, a new spatial downscaling algorithm for SMAP SM products is proposed and compared with the geographically weighted regression (GWR) algorithm and moving window algorithm (MW). The algorithm was applied to data taken during the SMAP Validation Experiment 2016 (SMAPVEX16), and the downscaled SM evaluated with in situ SM and aircraft observations. These results show that: 1) the SM downscaling conversion function, based on the MGWR, effectively reveals how different surface parameters relate to SM, with LST influencing SM globally and NDVI affecting it locally, 2) compared with GWR and MW, the 1 km SM obtained by the MGWR-based downscaling method showed better spatial agreement with airborne SM in the semiarid region with the ubRMSE decreasing by 0.023 m3/m3 and 0.026 m3/m3 for GWR and MW respectively, 3) when evaluating the 1 km SM with in situ SM in the semiarid region, the SM obtained through MGWR exhibited lower ubRMSE values as compared to both GWR and MW. Specifically, the median ubRMSE values for MGWR, GWR, and MW were 0.043 m3/m3, 0.053 m3/m3, and 0.094 m3/m3, respectively. In conclusion, this study demonstrates that MGWR can effectively improve the accuracy of downscaled SM in semiarid regions.

Spatial heterogeneity of the effects of river network patterns on water quality in highly urbanized city

River water quality deterioration is a serious problem in urban water environments. River network patterns affect water quality by influencing the flow, mixing, and other processes of water bodies. However, the effects of urban river network patterns on water quality remain poorly understood, thereby hindering the urban planning and management decision-making process. In this study, the geographically weighted regression (GWR) model was used to explore the spatial heterogeneity of the relationship between river network pattern and water quality. The results showed that the river network has a complex structure, high connectivity, and relatively even distribution and morphology. Important river structure indicators affecting water quality included the water surface ratio (Wp) and multifractal features (∆α, ∆f) while important river connectivity indicators included circuitry (α) and network connectivity (γ). River structure has a more complex effect on water quality than connectivity. This study recommends that the Wp should be increased in agricultural areas and appropriately reduced in urban built-up areas, and the number of river segments and nodes should be controlled within a rational configuration. Our study provides key insights for evaluating and optimizing the river network patterns to improve water quality of urban rivers. In the future, the land use intensity, hydrological processes, and human activities should be coupled with the river network pattern to deepen our understanding of urban river environment.

The relationship between elevation-dependent warming and long-term trend of surface air temperature lapse rate

The intensity and direction of elevation-dependent warming (EDW) could vary in different parts of the mountain, while less attention is paid to the local EDW effect. The Tibetan Plateau (TP) and the T'ai-hang & Lvliang Mountains in Shanxi Province of China (TLMsx) were taken as cases to examine the relationship between EDW and the long-term trend of surface air temperature (SAT) lapse rate (SATLR). For the observed data, the intensity of regional EDW (IEDW) was estimated with linear regression model, while the SATLR was estimated with geographically weighted regression (GWR) model. The long-term trend of SATLR was estimated applying the modified Theil-Sen method, and the regional average SATLR trend (kSATLR¯) was calculated and analyzed. Both observed data and simulation experiment revealed that there was not significant difference between IEDW and −kSATLR¯, with a little difference caused by the noise in the SAT trends. As the SATLR trend measured the difference in the surface air warming rate within 1-km vertical distance, i.e. the intensity of EDW in a small spatial scale, in the context of climate warming, it could be used to estimate the intensity of local EDW. The SATLR trend thus provided a new perspective to study the EDW effect in the future, and facilitated the quantitative analysis on the mechanisms of EDW.

A Global-Local collaborative approach to quantifying spatial non-stationarity in three-dimensional mineral prospectivity modeling

The three-dimensional (3D) modeling of mineral prospectivity plays a pivotal role in the mineral exploration process, enabling explorers to identify, evaluate, and prioritize deep-seated targets more effectively.Accurately predicting the 3D distribution of mineral prospectivity relies on effectively quantifying correlations to mineralization. The ore formation process is extremely intricate, marked by spatial non-stationary correlations. While state-of-the-art 3D mineral prospectivity modeling methods use machine learning (including deep learning) to capture correlation between geological evidence and mineralization hidden in the geosciences data, limited attention has been given to quantifying the local non-stationary nature in ore formation. This paper proposes a novel method that adapts to local spatial non-stationarity in 3D mineral prospectivity modeling. The approach integrates a prospectively model with a local regression model, creating a global–local collaborative model to quantify spatial non-stationary associations with mineralization. Random Forest (RF) and Reset are employed to construct mineral prospectivity models that quantify the global associations with mineralization. Simultaneously, the geographically weighted regression (GWR) model serves as a local model to capture the local non-stationarity within mineralization association. An Expectation Maximization algorithm is tailored to bridge the gap between global and local models, resulting in an iterative framework that progressively fuses global spatial associations and local non-stationarity related to mineralization. The iterative process generates a mineral perspectivity model that adapts to local non-stationarity in predictive areas. A case study on the Dingjiashan Pb-Zn deposit validates the effectiveness of the proposed method. The results demonstrate that the global–local collaborative mineral perspectivity model significantly boosts prediction accuracy in predictive areas. Moreover, the regression parameters of GWR derived from the model construction process effectively reveal the spatial non-stationary impacts of different ore-controlling factors, thereby improving the interpretability and reliability of predicted mineral prospectivity.

Spatio-temporal evolution and influencing factors of high quality economic development: Case study of Guangdong-Hong Kong-Macao Greater Bay Area

With the increase in global economic integration, high-quality economic development (HQED) has become a common goal of all countries. Based on these five development concepts, this paper uses the Gini coefficient, trend surface analysis, geographically weighted regression (GWR), the entropy weighting method, and standard deviation ellipse analysis to study the spatio-temporal pattern and driving mechanism of HQED in the Guangdong-Hong Kong-Macao Greater Bay Area (GBA). This paper examines the spatial and temporal patterns and driving mechanisms of HQED in the Guangdong-Hong Kong-Macao Greater Bay Area (GBA) based on five development concepts. The study revealed that (1) HQED is on the rise overall, but at the same time, it highlights the uneven development of multiple dimensions, especially in terms of significant differences in innovation, openness, and sharing. (2) HQED shows a clear center-periphery structure, with Guangzhou, Shenzhen, and Hong Kong as the core high-value areas; the growth rate of HQED capacity in the internal areas is significantly greater than that in the external areas, and HQED is prominent in the cities around Guangzhou, Shenzhen, and Hong Kong. (3) Factors such as agglomeration level, human capital, foreign investment, infrastructure development, financial and environmental protection expenditures, and financial inputs, and scientific, and technological inputs have a significant positive effect on HQED, and their interactions are further strengthened. This study reveals the importance of the realization of HQED in the GBA and the promotion of the overall development of the region.

Investigating the spatial distribution mechanisms of traditional villages from the human geography region: A case study of Jiangnan, China

Investigating the spatial distribution and adaptation mechanisms of traditional villages is crucial for their development and human well-being. However, relatively few studies have focused on the interaction and comprehensive influence of historical-cultural, natural-geographic and socio-economic factors on traditional villages in Jiangnan region. Employing geographically weighted regression (GWR) and Geodetector methods, we developed an integrated framework. We found that (1) Traditional villages exhibit a “one-core, multiple-periphery” clustering pattern (nearest neighbor index <1), with significant high-high and low-low local spatial clustering characteristics and preferring gentle and relatively Complex terrains to adapt to the need to avoid warfare. (2) Urbanization negatively affects traditional village density patterns, while natural-geographic complexity has a positive impact. Their interaction nonlinearly amplifies this effect, with the urbanization-GDP interplay being particularly significant (q = 0.765). (3) Factors exhibit significant spatial heterogeneity in their impact on traditional village density, with the explanatory power increasing by 20% (GWR/R2 = 0.65). The urbanization rate and slope are the most critical socio-economic and natural-geographic factors, respectively (local regression coefficient: 23.31 and 11.58). Our study transcends administrative boundaries, revealing how historical-cultural, natural-geographical, and socio-economic factors collectively influence the distribution of traditional villages. This not only provides a new theoretical foundation for the conservation and development of traditional villages but also provides a new methodology basis for research on traditional villages.

Analysis of Factors Influencing Housing Prices in Mountain Cities Based on Multiscale Geographically Weighted Regression—Demonstrated in the Central Urban Area of Chongqing

By leveraging a multiscale geographically weighted regression (MGWR) model, this paper delves into the intricate factors that influence housing prices in the prototypical mountainous cityscape of Chongqing’s central urban area. The key findings are as follows: Firstly, the distribution of housing prices in the study region exhibits pronounced spatial heterogeneity, with the core area exhibiting a distinct “high-high” clustering pattern and manifesting characteristics of a multicenter group distribution. Secondly, the MGWR model effectively assigns an individual bandwidth to each feature quantity, allowing for a more nuanced portrayal of the varying influence scales exerted by diverse variables. Lastly, the study reveals that factors such as property cost, greening rate, building age, and proximity to rivers have a notable negative impact on housing prices, whereas, educational facilities exert a marked positive influence. Elevation, floor area ratio, and distance from the Central Business District (CBD) exhibit a more complex influence on housing prices.

Spatial Heterogeneity in Determinants of No Condom Use at First Sex in South Africa

Background Condom use, especially at first sex for unmarried people, is an important behaviour that reduces the risk of HIV transmission, and other Sexually Transmitted Diseases (STDs), as well as unwanted pregnancies. In reviewing the literature, little information was found to elucidate the spatial context of condom use among people who had an early sexual debut. Understanding factors that influence no condom use at first sex in different local or geographical areas can add value in addressing these risky first-sex practices and minimise the risk of getting HIV, STDs, and unwanted pregnancies in South Africa, especially among young people. Therefore, this paper aimed to determine the spatial heterogeneity of factors associated with no condom use at first sex in South Africa using Geographically Weighted Regression (GWR) with Poisson distribution and local bivariate relationships. Methods This article used data from the fifth South African National HIV Prevalence, HIV Incidence, Behaviour and Communication Survey (SABSSM V) of 2017. Spatial autocorrelation, GWR using Poisson distribution, and local bivariate relationships were undertaken using ArcGIS Pro. Results Geographically weighted regression using Poisson distribution results highlighted being female, being young, having never married, being unemployed, having sexual debut at a younger age, having two or more sexual partners, having no risk of being infected with HIV, and having good Self-perceived Health (SPH) to be significant factors influencing no condom use at first sex, with pseudo R2 = 0.96 and AICc = 692. Local bivariate relationships showed the relationship between no condom use at first sex and most explanatory variables to exhibit a positive linear relationship in most districts with the exception of having never married, which showed a positive linear relationship in only 18 districts, while 33 districts showed a concave relationship. These relationships were all significant with p &lt; 0.05. Conclusion This localised analytical approach may contribute to the development of geo-targeted interventions and policies tailored to specific regions within South Africa, thereby improving the effectiveness of efforts to promote safe first sexual practices. Educational programs targeting youth should be developed to further raise awareness of the risks associated with unprotected sex and the importance of condom use. More efforts should be made to break the barriers, such as sexual relationship power, gender, and sociocultural and religious influences, which tend to put young, never-married, and unemployed women at risk of no condom use at first sexual debut. It is further recommended that the type of relationship between no condom use at first sex and each explanatory factor at the local region be considered during geo-targeted interventions as this will assist in the type of intervention and duration of intervention.