Machine learning methods dealing with the spatial auto-correlation of the response variable have garnered significant attention in the context of spatial prediction.Nonetheless,under these methods,the relationship bet...Machine learning methods dealing with the spatial auto-correlation of the response variable have garnered significant attention in the context of spatial prediction.Nonetheless,under these methods,the relationship between the response variable and explanatory variables is assumed to be homogeneous throughout the entire study area.This assumption,known as spatial stationarity,is very questionable in real-world situations due to the influence of contextual factors.Therefore,allowing the relationship between the target variable and predictor variables to vary spatially within the study region is more reasonable.However,existing machine learning techniques accounting for the spatially varying relationship between the dependent variable and the predictor variables do not capture the spatial auto-correlation of the dependent variable itself.Moreover,under these techniques,local machine learning models are effectively built using only fewer observations,which can lead to well-known issues such as over-fitting and the curse of dimensionality.This paper introduces a novel geostatistical machine learning approach where both the spatial auto-correlation of the response variable and the spatial non-stationarity of the regression relationship between the response and predictor variables are explicitly considered.The basic idea consists of relying on the local stationarity assumption to build a collection of local machine learning models while leveraging on the local spatial auto-correlation of the response variable to locally augment the training dataset.The proposed method’s effectiveness is showcased via experiments conducted on synthetic spatial data with known characteristics as well as real-world spatial data.In the synthetic(resp.real)case study,the proposed method’s predictive accuracy,as indicated by the Root Mean Square Error(RMSE)on the test set,is 17%(resp.7%)better than that of popular machine learning methods dealing with the response variable’s spatial auto-correlation.Additionally,this method is not only valuable for spatial prediction but also offers a deeper understanding of how the relationship between the target and predictor variables varies across space,and it can even be used to investigate the local significance of predictor variables.展开更多
On the basis of spatial-temporal perspective,by using the data of farmers' net income per capita from 1996 to 2007 in counties of Guangxi coupled with the global and local spatial auto-correlation analysis of ESDA...On the basis of spatial-temporal perspective,by using the data of farmers' net income per capita from 1996 to 2007 in counties of Guangxi coupled with the global and local spatial auto-correlation analysis of ESDA and GIS technology,we conduct the empirical research on the rural economic developmental disparity of counties in Guangxi and the evolving characteristics of local spatial heterogeneity.The results show that the rural economic developmental disparity of counties in Guangxi from 1996 to 2007 varies infinitesimally on the whole,and the regions with similar rural economic developmental level have spatial concentrated distribution.Based on these,the local MORAN'S I scatter diagram and LISA concentration diagram are drawn.In comparison with the traditional analytical method,the spatial analytical method of ESDA-GIS can explain the problem of spatial heterogeneity of rural economic development clearly,and have direct visual effect.展开更多
With the introduction of powerful and high-speed personal computers, proficient techniques for infrastructure development and management have advanced, of which Geoinformatics technology is of great significance. An a...With the introduction of powerful and high-speed personal computers, proficient techniques for infrastructure development and management have advanced, of which Geoinformatics technology is of great significance. An attempt has been made for broad mapping and analysis of existing infrastructures in the context of planning scheme in Paschim Medinipur district, and to delineate the development zones of educational infrastructure facilities. The thematic layers considered in this study are infrastructure accessibility, type and condition of classroom and number of classroom allocated for the educational system at primary and upper primary level. Moran’s I statistics was used to estimate the spatial distribution of elementary infrastructure across the district. All these themes and their individual features were then assigned weights according to their relative importance in educational development and corresponding normalized weights were obtained based on the Saaty’s analytical hierarchy process. The thematic layers were finally integrated in GIS software based on multi-criteria approach to yield educational development infrastructure index. Moran’s I statistics shows girl’s toilet, electric and boundary wall facility within the district are clustered in pattern at primary level. At the upper primary level, only electric and computer facilities shows the clustered distribution across the district. However, four different zones have been delineated, namely ‘very good’, ‘good’, ‘moderate’ and ‘poor’. The block covered by very good elementary educational infrastructure facility is Daspur –I and Dantan –II at primary level and Keshiary block at upper primary level in Paschim Medinipur district. Finally, it is concluded that the Geoinformatics technology is very efficient and useful for the identification of infrastructure development.展开更多
Among cancers, lung cancer is the most common cause of death in China. For the prevention and control of lung cancer, it is necessary to investigate the spatial and temporal distribution of lung cancer mortality, as w...Among cancers, lung cancer is the most common cause of death in China. For the prevention and control of lung cancer, it is necessary to investigate the spatial and temporal distribution of lung cancer mortality, as well as the changes in the trend and the affecting mechanism. Based on statistics and auto-correlation analysis, this paper studied the spatial and temporal distribution of lung cancer mortality in Yuhui District, Bengbu, Huaihe River Basin, from 2017 to 2020. In addition, Spearman’s Rank Correlation Assessment Model and Geographic Detector Model were used to examine the relationship between environmental factors and lung cancer mortality to identify impact factors and their mechanisms. The findings indicated that: 1) from the characteristics of temporal distribution, the number of lung cancer deaths exhibited a linear growth tendency, with the highest mortality in winter;2) from the characteristics of spatial distribution, lung cancer mortality showed a strong spatial agglomeration form, concentrating on two clustering areas, located in the old city and the central city of Bengbu, near the Huaihe River;3) from the point of view of the whole research area, there were 15 impact factors with significant correlation in the built and natural environment factors. The significant impacting factors in the built environment included land use, road traffic, spatial form and blue-green space, which could indirectly affect lung cancer mortality, while air pollution and temperature constituted the significant impacting factors in the natural environment;4) the influence of screened environmental factors on lung cancer mortality was different. Spatial stratified heterogeneity assessment, the interaction among environmental factors demonstrated statistical significance, it was found that the interaction between environmental factors in pairs had a significant enhancement effect on lung cancer mortality. To some extent, urban planning and policies could reduce lung cancer mortality.展开更多
Micromotion is the daily tiny vibration of the earth</span><span style="font-family:Verdana;">’</span><span style="font-family:Verdana;">s surface. Micromotional exploratio...Micromotion is the daily tiny vibration of the earth</span><span style="font-family:Verdana;">’</span><span style="font-family:Verdana;">s surface. Micromotional exploration can use the surface wave information of micro motion to study the shallow structure of underground media. In this study, we collected microtremor data at 68 stations in the Middle-Lower Yangtze Metallogenic Belt (MLYMB) and determined the resonant frequency and obtained the distribution of sedimentary thickness in this area by using H/V spectral ratio. According to the results of H/V, the sedimentary layer in the basin is thick, and the predominant frequency of the basin is 0.05</span><span style="font-family:""> </span><span style="font-family:Verdana;">-</span><span style="font-family:""> </span><span style="font-family:Verdana;">0.1</span><span style="font-family:""> </span><span style="font-family:Verdana;">Hz. There are no obvious lateral changes in the impedance interface between bedrock and sedimentary layer in this area. The basement of Tongling, Anqing and Luzhou mining areas and their adjacent areas is Kongling-Dongling type basement, which is composed of a set of metamorphic core complex. The predominant frequency is 0.05</span><span style="font-family:""> </span><span style="font-family:Verdana;">-</span><span style="font-family:""> </span><span style="font-family:Verdana;">0.1</span><span style="font-family:""> </span><span style="font-family:Verdana;">Hz. The sedimentary thickness gradually thinned from 3800</span><span style="font-family:""> </span><span style="font-family:Verdana;">m in the west to 2100</span><span style="font-family:""> </span><span style="font-family:Verdana;">m in the East. Moreover, this article used SPAC (spatial autocorrelation) method to obtain the S-wave velocity structure of the mining area near Luzong. The SPAC method reveals that the depth of the interface between the loose sediments and the volcanic rocks is about 600 m in the study area near the Luzhou mining area in the Middle-Lower Yangtze Metallogenic Belt, and the average depth of the interface between the volcanic rock section and the intrusive complex section is about 1000</span><span style="font-family:""> </span><span style="font-family:Verdana;">m. The thickness of the intrusive rock is more than 2500</span><span style="font-family:""> </span><span style="font-family:Verdana;">m. Tourmaline is developed in the interior of the intrusive rock, which may have better exploration value.展开更多
文摘Machine learning methods dealing with the spatial auto-correlation of the response variable have garnered significant attention in the context of spatial prediction.Nonetheless,under these methods,the relationship between the response variable and explanatory variables is assumed to be homogeneous throughout the entire study area.This assumption,known as spatial stationarity,is very questionable in real-world situations due to the influence of contextual factors.Therefore,allowing the relationship between the target variable and predictor variables to vary spatially within the study region is more reasonable.However,existing machine learning techniques accounting for the spatially varying relationship between the dependent variable and the predictor variables do not capture the spatial auto-correlation of the dependent variable itself.Moreover,under these techniques,local machine learning models are effectively built using only fewer observations,which can lead to well-known issues such as over-fitting and the curse of dimensionality.This paper introduces a novel geostatistical machine learning approach where both the spatial auto-correlation of the response variable and the spatial non-stationarity of the regression relationship between the response and predictor variables are explicitly considered.The basic idea consists of relying on the local stationarity assumption to build a collection of local machine learning models while leveraging on the local spatial auto-correlation of the response variable to locally augment the training dataset.The proposed method’s effectiveness is showcased via experiments conducted on synthetic spatial data with known characteristics as well as real-world spatial data.In the synthetic(resp.real)case study,the proposed method’s predictive accuracy,as indicated by the Root Mean Square Error(RMSE)on the test set,is 17%(resp.7%)better than that of popular machine learning methods dealing with the response variable’s spatial auto-correlation.Additionally,this method is not only valuable for spatial prediction but also offers a deeper understanding of how the relationship between the target and predictor variables varies across space,and it can even be used to investigate the local significance of predictor variables.
文摘On the basis of spatial-temporal perspective,by using the data of farmers' net income per capita from 1996 to 2007 in counties of Guangxi coupled with the global and local spatial auto-correlation analysis of ESDA and GIS technology,we conduct the empirical research on the rural economic developmental disparity of counties in Guangxi and the evolving characteristics of local spatial heterogeneity.The results show that the rural economic developmental disparity of counties in Guangxi from 1996 to 2007 varies infinitesimally on the whole,and the regions with similar rural economic developmental level have spatial concentrated distribution.Based on these,the local MORAN'S I scatter diagram and LISA concentration diagram are drawn.In comparison with the traditional analytical method,the spatial analytical method of ESDA-GIS can explain the problem of spatial heterogeneity of rural economic development clearly,and have direct visual effect.
文摘With the introduction of powerful and high-speed personal computers, proficient techniques for infrastructure development and management have advanced, of which Geoinformatics technology is of great significance. An attempt has been made for broad mapping and analysis of existing infrastructures in the context of planning scheme in Paschim Medinipur district, and to delineate the development zones of educational infrastructure facilities. The thematic layers considered in this study are infrastructure accessibility, type and condition of classroom and number of classroom allocated for the educational system at primary and upper primary level. Moran’s I statistics was used to estimate the spatial distribution of elementary infrastructure across the district. All these themes and their individual features were then assigned weights according to their relative importance in educational development and corresponding normalized weights were obtained based on the Saaty’s analytical hierarchy process. The thematic layers were finally integrated in GIS software based on multi-criteria approach to yield educational development infrastructure index. Moran’s I statistics shows girl’s toilet, electric and boundary wall facility within the district are clustered in pattern at primary level. At the upper primary level, only electric and computer facilities shows the clustered distribution across the district. However, four different zones have been delineated, namely ‘very good’, ‘good’, ‘moderate’ and ‘poor’. The block covered by very good elementary educational infrastructure facility is Daspur –I and Dantan –II at primary level and Keshiary block at upper primary level in Paschim Medinipur district. Finally, it is concluded that the Geoinformatics technology is very efficient and useful for the identification of infrastructure development.
基金Under the auspices of Natural Science Foundation of Anhui Province (No. 2008085ME160)Provincial Natural Science Research Projects in Anhui Province-Postgraduate Projects (No. YJS20210500)。
文摘Among cancers, lung cancer is the most common cause of death in China. For the prevention and control of lung cancer, it is necessary to investigate the spatial and temporal distribution of lung cancer mortality, as well as the changes in the trend and the affecting mechanism. Based on statistics and auto-correlation analysis, this paper studied the spatial and temporal distribution of lung cancer mortality in Yuhui District, Bengbu, Huaihe River Basin, from 2017 to 2020. In addition, Spearman’s Rank Correlation Assessment Model and Geographic Detector Model were used to examine the relationship between environmental factors and lung cancer mortality to identify impact factors and their mechanisms. The findings indicated that: 1) from the characteristics of temporal distribution, the number of lung cancer deaths exhibited a linear growth tendency, with the highest mortality in winter;2) from the characteristics of spatial distribution, lung cancer mortality showed a strong spatial agglomeration form, concentrating on two clustering areas, located in the old city and the central city of Bengbu, near the Huaihe River;3) from the point of view of the whole research area, there were 15 impact factors with significant correlation in the built and natural environment factors. The significant impacting factors in the built environment included land use, road traffic, spatial form and blue-green space, which could indirectly affect lung cancer mortality, while air pollution and temperature constituted the significant impacting factors in the natural environment;4) the influence of screened environmental factors on lung cancer mortality was different. Spatial stratified heterogeneity assessment, the interaction among environmental factors demonstrated statistical significance, it was found that the interaction between environmental factors in pairs had a significant enhancement effect on lung cancer mortality. To some extent, urban planning and policies could reduce lung cancer mortality.
文摘Micromotion is the daily tiny vibration of the earth</span><span style="font-family:Verdana;">’</span><span style="font-family:Verdana;">s surface. Micromotional exploration can use the surface wave information of micro motion to study the shallow structure of underground media. In this study, we collected microtremor data at 68 stations in the Middle-Lower Yangtze Metallogenic Belt (MLYMB) and determined the resonant frequency and obtained the distribution of sedimentary thickness in this area by using H/V spectral ratio. According to the results of H/V, the sedimentary layer in the basin is thick, and the predominant frequency of the basin is 0.05</span><span style="font-family:""> </span><span style="font-family:Verdana;">-</span><span style="font-family:""> </span><span style="font-family:Verdana;">0.1</span><span style="font-family:""> </span><span style="font-family:Verdana;">Hz. There are no obvious lateral changes in the impedance interface between bedrock and sedimentary layer in this area. The basement of Tongling, Anqing and Luzhou mining areas and their adjacent areas is Kongling-Dongling type basement, which is composed of a set of metamorphic core complex. The predominant frequency is 0.05</span><span style="font-family:""> </span><span style="font-family:Verdana;">-</span><span style="font-family:""> </span><span style="font-family:Verdana;">0.1</span><span style="font-family:""> </span><span style="font-family:Verdana;">Hz. The sedimentary thickness gradually thinned from 3800</span><span style="font-family:""> </span><span style="font-family:Verdana;">m in the west to 2100</span><span style="font-family:""> </span><span style="font-family:Verdana;">m in the East. Moreover, this article used SPAC (spatial autocorrelation) method to obtain the S-wave velocity structure of the mining area near Luzong. The SPAC method reveals that the depth of the interface between the loose sediments and the volcanic rocks is about 600 m in the study area near the Luzhou mining area in the Middle-Lower Yangtze Metallogenic Belt, and the average depth of the interface between the volcanic rock section and the intrusive complex section is about 1000</span><span style="font-family:""> </span><span style="font-family:Verdana;">m. The thickness of the intrusive rock is more than 2500</span><span style="font-family:""> </span><span style="font-family:Verdana;">m. Tourmaline is developed in the interior of the intrusive rock, which may have better exploration value.
