Method for analyzing spatial heterogeneity of energy ecological efficiency influencing factors

Abstract

This application relates to a spatial heterogeneity analysis method for factors influencing energy and ecological efficiency, aiming to address the problem that traditional methods cannot simultaneously consider the differentiated spatial scales of various factors and the attribution of efficiency gaps between regions. It belongs to the field of geographic information and science and technology consulting technology. This method utilizes a multi-scale spatial regression model to perform geographic weighted fitting on panel data of each unit in the target region, obtaining the local influence coefficients of each influencing factor in the spatial dimension. It then combines Shapley value decomposition to calculate the original contribution value of each factor to the energy and ecological efficiency gap between regions, and finally aggregates these to obtain a global impact intensity index and a global difference contribution degree. Based on this, a two-dimensional geographic classification framework is constructed to classify influencing factors into multiple categories of differentiated policy-oriented factors. Finally, the analysis results are comprehensively output as a two-dimensional scatter plot, a standardized classification report, and a regional policy priority zoning map, providing quantitative consulting support for the formulation of regional energy and environmental management strategies from a geospatial perspective.

Description

Technical Field

[0001] This application relates to the fields of geographic information technology and scientific and technological consulting technology, and in particular to a method for analyzing the spatial heterogeneity of factors affecting energy and ecological efficiency. Background Technology

[0002] Energy-ecological efficiency is a comprehensive indicator that measures the degree of coordination between regional economic development and energy consumption and environmental load, and it has important reference value in regional sustainable development assessment and policy formulation. Quantitative analysis of the influencing factors and spatial distribution characteristics of energy-ecological efficiency has become a key technical requirement supporting the design of zoned and classified management policies.

[0003] Currently, various quantitative analysis tools have been developed in the field of energy and ecological efficiency influencing factor analysis, which can be adapted to basic regional energy efficiency driving factor assessment. However, the overall adaptability of existing analytical methods is limited, making it difficult to meet the integrated analysis needs of complex and heterogeneous regions, which simultaneously consider the analysis of factor-driven laws and the assessment of the causes of regional energy efficiency imbalances. The analytical dimensions are one-sided, and the results are not closely aligned with the actual governance scenarios of the regions. Overall, the analytical accuracy and application limitations are significant, failing to provide comprehensive and reliable data analysis support for tiered and site-specific regional energy and ecological management policies. Therefore, there is an urgent need for a specialized analytical method for energy and ecological efficiency influencing factors that is adaptable to complex spatial regions, has more comprehensive analytical dimensions, and yields more reliable results. Summary of the Invention

[0004] Based on this, the purpose of this application is to provide a spatial heterogeneity analysis method for factors affecting energy and ecological efficiency. By configuring differentiated spatial scales for each factor and decomposing regional pairing contributions, it can simultaneously estimate local effects and attribute efficiency gaps between regions.

[0005] The spatial heterogeneity analysis method for energy and ecological efficiency influencing factors described in this application includes the following steps:

[0006] Acquire panel data for each unit area within the target area; the panel data includes spatial location information, energy and ecological efficiency parameters as the analyzed indicators, and indicator data for several influencing factors; Based on the spatial location information of each unit region, corresponding spatial scales are configured for each influencing factor, and corresponding spatial weight matrices are constructed. Using the energy-ecological efficiency parameters of each unit region as the dependent variable and the index data of the influencing factors as the independent variables, and based on the spatial weight matrices corresponding to each influencing factor, local spatial weighted regression fitting is performed on each unit region to obtain a spatial regression model and the local influence coefficient for each unit region and each influencing factor. Based on the local influence coefficients of the influencing factors in each unit region, the corresponding global influence intensity index is obtained. By iterating through all pairwise combinations of unit regions within the target area, and based on the energy efficiency prediction results of the spatial regression model, the contribution of each influencing factor to the energy-ecological efficiency difference of any pair is determined, and the original contribution value of each influencing factor under each pair is obtained; based on the original contribution value of each influencing factor in all pair combinations, the global difference contribution degree corresponding to each influencing factor is obtained. Based on the global impact intensity index and global difference contribution of each influencing factor, the analysis results of the factors affecting the energy and ecological efficiency of the target area are obtained.

[0007] This application's embodiments configure independent spatial scales for each influencing factor and construct corresponding spatial weight matrices. Then, based on different spatial weight matrices, local spatial weighted regression is performed separately. This allows factors with different spatial radiation ranges to be estimated at reasonable spatial scales, avoiding the problem of underestimation or overestimation of local effects of some factors due to the use of a uniform bandwidth in traditional geographic weighted regression. This results in local influence coefficients for each unit region and each influencing factor that better reflect the true patterns of regional geographic differentiation, improving the data credibility of geospatial quantitative analysis. Furthermore, the local influence coefficients of each unit region are aggregated into a global influence intensity index, objectively reflecting the average influence of each factor within the overall region, providing basic spatial quantitative parameters for geographic science and technology consulting in the fields of land space and energy ecology. Further, by traversing all pairwise combinations of unit regions, energy efficiency prediction is performed for each pair using a spatial regression model. The difference in prediction results is used as a measure of efficiency gap, thereby decomposing the contribution of each factor to this gap, obtaining the original contribution value, and aggregating it into a global difference contribution degree. This process breaks down the sources of efficiency disparities between regions into individual influencing factors. This allows the analysis to depict the distribution of the local effects of each factor at different spatial locations, while also identifying the core factors driving the differentiation of energy and ecological efficiency between regions. Thus, within a single analytical framework, it simultaneously locates local effects and traces the sources of regional disparities, improving the standardized technical chain for spatial heterogeneity diagnosis and attribution of regional development imbalances in geographic science and technology consulting. Ultimately, by combining the overall effect intensity index with the overall difference contribution, decision-makers can clearly distinguish between factors that, while having a high average effect intensity across the entire region, do not significantly widen regional disparities, and factors that, while not necessarily having the strongest average effect, continuously drive regional imbalances. This aligns with the tiered analysis needs of land planning and ecological energy-specific geographic consulting services, providing a complete quantitative basis for the formulation of zoning control policies that simultaneously covers both "local effect characterization" and "regional disparity attribution." This effectively supports the implementation of geographic science and technology consulting results into practical solutions for differentiated spatial governance and integrated energy and ecological management.

[0008] To better understand and implement this application, the following detailed description is provided in conjunction with the accompanying drawings. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating the spatial heterogeneity analysis method for energy and ecological efficiency influencing factors according to an embodiment of this application. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings. Wherein, when the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements.

[0011] It should be understood that the embodiments described below do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this application.

[0012] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application are also intended to include the plural forms unless the context clearly indicates otherwise. Furthermore, in the description of this application, unless otherwise stated, “a plurality” means two or more. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more associated listed items, for example, A and / or B, which can represent: A alone, A and B together, and B alone; the character “ / ” generally indicates that the preceding and following objects are in an “or” relationship.

[0013] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, this information should not be limited to these terms, and these terms are only used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances. Depending on the context, the word "if" as used in this application can be interpreted as "when," "when," or "in response to determination."

[0014] Please refer to Figure 1 The spatial heterogeneity analysis method for energy and ecological efficiency influencing factors described in this application includes the following steps: S101: Obtain panel data for each unit area within the target area; the panel data includes spatial location information, energy and ecological efficiency parameters as the analyzed indicators, and indicator data for several influencing factors; S102: Based on the spatial location information of each unit region, configure corresponding spatial scales for each influencing factor and construct corresponding spatial weight matrices; using the energy and ecological efficiency parameters of each unit region as the dependent variable and the index data of the influencing factors as the independent variables, and based on the spatial weight matrices corresponding to each influencing factor, perform local spatial weighted regression fitting on each unit region to obtain a spatial regression model and obtain the local influence coefficients for each unit region and each influencing factor; based on the local influence coefficients of the influencing factors in each unit region, obtain the corresponding global influence intensity index; S103: Traverse all pairwise combinations of unit regions within the target area, and based on the energy efficiency prediction results of the spatial regression model, determine the contribution of each influencing factor to the energy-ecological efficiency difference of any pair, and obtain the original contribution value of each influencing factor under each pair; based on the original contribution value of each influencing factor in all pair combinations, obtain the global difference contribution degree corresponding to each influencing factor. S104: Based on the global impact intensity index and global difference contribution of each influencing factor, the analysis results of the factors affecting the energy and ecological efficiency of the target area are obtained.

[0015] This application's embodiments configure independent spatial scales for each influencing factor and construct corresponding spatial weight matrices. Then, based on different spatial weight matrices, local spatial weighted regression is performed separately. This allows factors with different spatial radiation ranges to be estimated at reasonable spatial scales, avoiding the problem of underestimation or overestimation of local effects of some factors due to the use of a uniform bandwidth in traditional geographic weighted regression. This results in local influence coefficients for each unit region and each influencing factor that better reflect the true patterns of regional geographic differentiation, improving the data credibility of geospatial quantitative analysis. Furthermore, the local influence coefficients of each unit region are aggregated into a global influence intensity index, objectively reflecting the average influence of each factor within the overall region, providing basic spatial quantitative parameters for geographic science and technology consulting in the fields of land space and energy ecology. Further, by traversing all pairwise combinations of unit regions, energy efficiency prediction is performed for each pair using a spatial regression model. The difference in prediction results is used as a measure of efficiency gap, thereby decomposing the contribution of each factor to this gap, obtaining the original contribution value, and aggregating it into a global difference contribution degree. This process breaks down the sources of efficiency disparities between regions into individual influencing factors. This allows the analysis to depict the distribution of the local effects of each factor at different spatial locations, while also identifying the core factors driving the differentiation of energy and ecological efficiency between regions. Thus, within a single analytical framework, it simultaneously locates local effects and traces the sources of regional disparities, improving the standardized technical chain for spatial heterogeneity diagnosis and attribution of regional development imbalances in geographic science and technology consulting. Ultimately, by combining the overall effect intensity index with the overall difference contribution, decision-makers can clearly distinguish between factors that, while having a high average effect intensity across the entire region, do not significantly widen regional disparities, and factors that, while not necessarily having the strongest average effect, continuously drive regional imbalances. This aligns with the tiered analysis needs of land planning and ecological energy-specific geographic consulting services, providing a complete quantitative basis for the formulation of zoning control policies that simultaneously covers both "local effect characterization" and "regional disparity attribution." This effectively supports the implementation of geographic science and technology consulting results into practical solutions for differentiated spatial governance and integrated energy and ecological management.

[0016] The spatial heterogeneity analysis method for energy and ecological efficiency influencing factors described in this application uses a computer as the execution subject. The following provides a detailed description of each step.

[0017] For step S101, panel data of each unit area within the target area is obtained; the panel data includes spatial location information, energy and ecological efficiency parameters as the analyzed indicators, and indicator data of several influencing factors.

[0018] Energy eco-efficiency parameter is a comprehensive indicator used to quantify the degree of matching between regional energy consumption, economic output, and environmental load. In this embodiment, the comprehensive calculation results of energy consumption per unit GDP and carbon emissions per unit GDP of each unit region are used. The higher the value of this indicator, the lower the energy consumption and environmental cost of the region under the same economic output, that is, the better the energy eco-efficiency.

[0019] Influencing factors refer to external variables that may drive or inhibit energy and ecological efficiency. In this embodiment, several influencing factors include urbanization level, economic development level, industrial structure index, and degree of openness to the outside world. Among them, urbanization level can be measured by the proportion of urban permanent residents to the total population; economic development level can be measured by per capita GDP; industrial structure index can be measured by the proportion of secondary industry added value to GDP; and degree of openness to the outside world can be measured by the proportion of total import and export volume to GDP.

[0020] In this step, panel data for each unit area within the target region is first acquired. Panel data is a longitudinal dataset spanning multiple time segments, simultaneously reflecting the state changes of each unit area at different points in time. Specifically, the panel data acquired in this step includes three parts: first, spatial location information for each unit area, describing its geographical distribution, which can be represented by the latitude and longitude coordinates of the administrative centers of each unit area; second, energy-ecological efficiency parameters as the analyzed indicators, namely the aforementioned indicators that comprehensively reflect the degree of matching between energy consumption, economic output, and environmental load in each unit area; and third, indicator data for several influencing factors, namely the values ​​of four variables—urbanization level, economic development level, industrial structure index, and degree of openness to the outside world—at each unit area and each time segment. The above data can be obtained from statistical yearbooks and relevant public databases of each unit area, and after normalization or standardization processing, constitute a complete panel dataset.

[0021] For step S102, based on the spatial location information of each unit region, a corresponding spatial scale is configured for each influencing factor and a corresponding spatial weight matrix is ​​constructed; the energy and ecological efficiency parameters of each unit region are used as the dependent variable of the model and the index data of the influencing factors are used as the independent variables of the model. Based on the spatial weight matrix corresponding to each influencing factor, local spatial weighted regression fitting is performed on each unit region to obtain a spatial regression model and obtain the local influence coefficient corresponding to each unit region and each influencing factor; based on the local influence coefficient of the influencing factor in each unit region, the corresponding global influence intensity index is obtained.

[0022] The spatial weight matrix is ​​a mathematical matrix that reflects the degree of spatial correlation between unit regions. Each element represents the spatial influence weight of a unit region on another unit region on a specific influencing factor. The weight is determined by the spatial distance between the two regions and the spatial scale of the influencing factor.

[0023] Spatial scale is a parameter that characterizes the size of the spatial radiation range of a certain influencing factor. Different influencing factors have different spatial radiation radii due to their different mechanisms of action. This parameter is used to determine which neighboring regions should be included in the local regression window of the factor when constructing the spatial weight matrix.

[0024] Spatial regression model refers to a regression model obtained by local spatial weighted regression fitting. This model uses the energy and ecological efficiency parameters of each unit region as dependent variables and the index data of several influencing factors as independent variables. It performs weighted estimation based on the spatial weight matrix independently configured for each influencing factor. The model can determine the local influence coefficient corresponding to each influencing factor in each unit region during the fitting process, which is used to reflect the marginal effect of the factor in that spatial location.

[0025] The local impact coefficient is a regression coefficient estimated in each unit region through local spatial weighted regression. It reflects the direction and magnitude of the marginal effect of the corresponding influencing factor on energy and ecological efficiency at that specific spatial location.

[0026] The global impact intensity index is a scalar index obtained by aggregating the local impact coefficients of a certain influencing factor across all unit areas within a target region. It is used to measure the average impact of the factor throughout the entire target region. The aggregation calculation method can be to take the absolute value of the local impact coefficients and then calculate the mean.

[0027] In this step, due to the different mechanisms of action of various influencing factors, their spatial radiation ranges naturally differ, thus requiring independent configuration for each factor. Specifically, the spatial radiation range of urbanization level is usually wide because the urbanization process generates spillover effects to surrounding areas through infrastructure interconnection and population flow, allowing for a larger spatial scale. The radiation range of economic development level is moderate, allowing for a medium scale. The industrial structure index is affected by upstream and downstream linkages in the industrial chain, resulting in a considerable radiation range. The degree of openness to the outside world is constrained by transportation and trade channels, leading to a relatively concentrated radiation range, allowing for a smaller scale. The spatial scale can be determined through cross-validation, using the maximization of the goodness of fit of the local regression model for each factor as the criterion. After determining the spatial scale of each factor, a corresponding spatial weight matrix is ​​constructed. Each element in this matrix represents the spatial weight of a unit region on another unit region for a specific influencing factor. The weight is determined by the spatial distance between the two regions and the spatial scale of the influencing factor. Common construction methods include calculation based on Gaussian kernel function or inverse distance weight function. When the distance between the two regions exceeds the spatial scale of the factor, the weight is set to zero.

[0028] Subsequently, for each unit region, using that region as the regression center point, the weights of each unit region within the local window are determined using the spatial weight matrix of the corresponding influencing factor. Then, the regression coefficient of the influencing factor at that center point is estimated using weighted least squares, ensuring that each factor is estimated within a window matching its own spatial radiation range. For example, when estimating the local influence coefficient of urbanization level in a unit region, only neighboring unit regions within the spatial distance of the urbanization level's effective scale are included in the regression window, and the weights of each unit region within the window are determined by the spatial weight matrix corresponding to the urbanization level. Similarly, when estimating the local influence coefficient of openness to the outside world in the same unit region, only neighboring unit regions within the spatial distance of the openness level's effective scale are included in the window, and the weights are determined using the spatial weight matrix corresponding to the openness level. Through this method, a local influence coefficient is obtained for each unit region within the target area and for each of the influencing factors. Based on this, the local influence coefficients of each influencing factor across all unit regions are aggregated. For example, the absolute value of the local influence coefficient is taken and the mean is calculated to obtain the global influence intensity index corresponding to the factor, which is used to measure the average influence of the factor in the whole region.

[0029] In one embodiment, step S102, based on the spatial location information of each unit region, configures corresponding spatial scales for each influencing factor and constructs corresponding spatial weight matrices; using the energy-ecological efficiency parameters of each unit region as the model dependent variable and the index data of the influencing factors as the model independent variables, and performing local spatial weighted regression fitting on each unit region based on the spatial weight matrices corresponding to each influencing factor to obtain a spatial regression model, includes: Step S1021: Initialize the spatial scale of each influencing factor and perform the following scale optimization operation: Perform local regression fitting with each unit region as the central region: Based on the spatial scale of any influencing factor, determine several adjacent unit regions in which the influencing factor acts; Based on the several adjacent unit regions and their corresponding spatial location information, construct the local spatial weight matrix corresponding to the influencing factor; Based on the energy and ecological efficiency parameters of the central region, the index data of each influencing factor in its respective adjacent unit regions, and the local spatial weight matrix corresponding to each influencing factor, perform local spatial weighted regression fitting.

[0030] First, initialize the spatial scale of each influencing factor. Initialization can be done by assigning an empirical default value to each factor, such as initializing urbanization level to 150 km, economic development level to 100 km, industrial structure index to 130 km, and openness to the outside world to 80 km. Alternatively, it can be uniformly initialized to a multiple of the average distance between unit regions within the target area. Then, perform scale optimization. Scale optimization involves iteratively adjusting the spatial scale of each influencing factor and performing local regression fitting centered on all unit regions after each adjustment to find the optimal scale combination for the overall model fit. Specifically, in each optimization iteration, local regression fitting is performed with each unit region as the central region. For any influencing factor, based on its current spatial scale, determine several adjacent unit regions affected by the factor—that is, all neighboring unit regions within the scale range. Based on these adjacent unit regions and their corresponding spatial location information, construct a local spatial weight matrix for the influencing factor. The weight of each element in this matrix can be calculated using a Gaussian kernel function; the closer the distance, the greater the weight, and if the distance exceeds the current spatial scale, the weight is set to zero. Subsequently, using the energy-ecological efficiency parameter of the central region as the dependent variable and the corresponding index data of each influencing factor in its adjacent unit region as the independent variable, and using the local spatial weight matrix corresponding to each influencing factor, a local spatial weighted regression fitting is performed on the central region. This process is repeated for all unit regions within the target region, thus completing one round of local regression fitting in the scale optimization.

[0031] Step S1022: Obtain the fitting residual corresponding to the local regression of each unit region; calculate the global correction Akaike information criterion value based on the fitting residual of all unit regions; update the spatial action scale of each influencing factor and continue to perform the scale optimization operation; when the global correction Akaike information criterion value meets the preset optimal condition, determine the spatial action scale corresponding to each influencing factor and obtain the spatial regression model.

[0032] Among them, the global calibration Akaike Information Criterion is an evaluation index that considers both model fit and model complexity. It is used to balance the relationship between fitting accuracy and the number of parameters in model comparison. The smaller the value, the better the model has achieved a balance between fitting accuracy and simplicity. It can be used as the optimal judgment criterion in the scaling optimization process.

[0033] After completing the local regression fitting of all unit regions in one round of scale optimization, the fitting residual corresponding to the local regression of each unit region is obtained, which is the difference between the actual value and the model prediction value of the energy-ecological efficiency parameter of that unit region. Subsequently, based on the fitting residuals of all unit regions, the global correction Akaike Information Criterion value is calculated. The calculation of this value takes into account both the sum of squared residuals of all unit regions and the number of effective parameters of the model, achieving a balance between fitting accuracy and model complexity. After obtaining the global correction Akaike Information Criterion value under the current scale combination, the spatial scale of each influencing factor is updated. For example, gradient descent or grid search methods can be used to adjust the scale of each factor. After the update, the scale optimization operation in step S1021 is continued, that is, the adjacent unit regions of each factor are re-determined with the new scale combination, the local spatial weight matrix is ​​reconstructed, local regression fitting is re-performed on all unit regions, and the global correction Akaike Information Criterion value is recalculated. This process is repeated iteratively until the global correction Akaike information criterion value meets the preset optimal conditions, such as the decrease in the value being less than a set threshold in several consecutive iterations, or the preset maximum number of iterations is reached. At this point, the spatial scale of each influencing factor is the optimal scale, and a multi-scale spatial regression model is obtained based on the combination of these optimal scales. In this model, each influencing factor is locally weighted and estimated using the spatial weight matrix corresponding to its optimal spatial scale.

[0034] This embodiment introduces a scale optimization iterative mechanism, using the globally corrected Akaike Information Criterion as the optimal criterion, to automatically search for the optimal spatial scale for influencing factors such as urbanization level, economic development level, industrial structure index, and degree of openness to the outside world. In each iteration, this process performs local regression fitting centered on the entire unit region and calculates the globally corrected Akaike Information Criterion value based on the fitting residuals of all unit regions. This ensures that the searched scale combination achieves an optimal balance between fitting accuracy and complexity in the spatial regression model, avoiding the subjectivity of manually setting the scale. Based on this, the spatial regression model constructed using the optimal scale combination obtained through optimization estimates the local influence coefficients of each influencing factor within a window that best matches its true spatial radiation range, ensuring the accuracy of the local coefficients.

[0035] In one embodiment, step S102, which obtains the local influence coefficient corresponding to each unit region and each influencing factor, includes: Step S1023: Input the panel data of each unit region within the target area into the spatial regression model; perform local weighted regression calculation on each unit region through the spatial regression model.

[0036] Using a spatial regression model with a determined optimal spatial scale, each unit region is sequentially used as a regression center, and the local regression coefficients of each influencing factor are estimated within that central region. Specifically, for any unit region within the target area, the energy-ecological efficiency parameter of that region is taken as the dependent variable, and the index data of each influencing factor in adjacent unit regions within the spatial scale of that region and each influencing factor are taken as independent variables. Weighted least squares operations are then performed using the local spatial weight matrices corresponding to each influencing factor. After performing the above local weighted regression operation on all unit regions within the target area, the estimated regression coefficients and related statistics of each influencing factor are obtained for each unit region.

[0037] Step S1024: Extract the local influence coefficients for each unit region and each influencing factor from the local weighted regression calculation results corresponding to each unit region.

[0038] Specifically, the results of the local weighted regression calculation performed on each unit region in step S1023 include the estimated regression coefficients of each influencing factor in that region. These estimated regression coefficients can be directly used as the corresponding local influence coefficients. For example, in the local weighted regression calculation results of a certain unit region, the estimated regression coefficient corresponding to the urbanization level is -0.35. Therefore, the local influence coefficient of the urbanization level in that unit region is -0.35, indicating that for every unit increase in the urbanization level in that region, the energy-ecological efficiency parameter decreases by an average of 0.35 units. After extracting the calculation results for all unit regions one by one, a local influence coefficient matrix of all unit regions and all influencing factors is obtained.

[0039] In this embodiment, panel data from each unit region within the target area are uniformly input into a spatial regression model determined by scale optimization. The model then performs local weighted regression calculations on each unit region, ensuring that each influencing factor in each unit region independently completes weighted regression estimation under the local spatial weight matrix corresponding to its optimal spatial scale. This guarantees that the extraction process of local influence coefficients is completely consistent with the optimal spatial configuration determined in the scale optimization stage.

[0040] In one embodiment, step S102, which involves obtaining the corresponding global impact intensity index based on the local impact coefficients of the influencing factors in each unit region, includes: Step S1025: Obtain all local influence coefficients of any influencing factor in all unit regions; take the absolute value of each local influence coefficient and calculate the average of all absolute values ​​to obtain the global effect intensity index corresponding to the influencing factor.

[0041] This embodiment obtains the total local influence coefficients of any influencing factor across all unit regions, and calculates the arithmetic mean of the absolute values ​​of each local influence coefficient to obtain the global influence intensity index corresponding to that factor. By using absolute value aggregation, the global influence intensity index reflects only the average magnitude of the factor's influence across the entire region, without offsetting each other due to some regions having a positive promoting effect and others a negative inhibiting effect. This allows for an objective measurement of the average influence of urbanization level, economic development level, industrial structure index, and degree of openness on energy and ecological efficiency across the entire region.

[0042] For step S103, traverse all pairwise combinations of unit regions within the target area, and based on the energy efficiency prediction results of the spatial regression model, determine the contribution of each influencing factor to the energy ecological efficiency difference of any pairing combination, and obtain the original contribution value of each influencing factor under each pairing combination; according to the original contribution value of each influencing factor in all pairing combinations, obtain the global difference contribution degree corresponding to each influencing factor.

[0043] The original contribution value refers to the specific contribution of a factor to the energy-ecological efficiency difference of a pairing combination under a certain unit area, obtained by using the prediction results of a spatial regression model to decompose the energy-ecological efficiency difference of the pairing combination into each influencing factor.

[0044] The global differential contribution rate is a scalar indicator obtained by aggregating the original contribution values ​​of a certain influencing factor across all pairwise combinations within a target region. It is used to measure the extent to which this factor drives the differentiation of energy and ecological efficiency among regions. The aggregation calculation method can be to sum the absolute values ​​of the original contribution values ​​across all pairwise combinations and then divide by the total number of pairwise combinations.

[0045] In this step, the gap in energy and eco-efficiency between regions is considered as a result driven by various influencing factors. By traversing all paired combinations and decomposing the sources of efficiency differences one by one, the causes of the regional gap are attributed. Specifically, for any pair of unit regions, the aforementioned spatial regression model is first used to predict the energy and eco-efficiency of the two unit regions, obtaining their respective predicted energy efficiency values. The difference between the two is the energy and eco-efficiency difference under that pair, reflecting the efficiency gap between the two regions. Based on this, the contribution of each influencing factor to the efficiency difference is determined, that is, the product of the difference in the local influence coefficient of the factor in the two regions and the difference in the index value of the factor in the two regions, thus obtaining the original contribution value of the factor under that pair. After performing the above calculation on all paired combinations, the original contribution values ​​of a certain factor in all paired combinations are aggregated, for example, by taking the absolute value, summing them, and then dividing by the total number of paired combinations, to obtain the global difference contribution degree corresponding to the factor. The larger the global difference contribution degree, the more significant the difference in the effect of the factor between different regions, and the more it is a driving factor that widens the gap in energy and eco-efficiency between regions.

[0046] In one embodiment, step S103, which involves iterating through all pairwise combinations of unit regions within the target area and determining the contribution of each influencing factor to the energy-ecological efficiency difference of any pair based on the energy efficiency prediction results of the spatial regression model, and obtaining the original contribution value of each influencing factor under each pair, includes: Step S1031: Traverse all pairwise combinations of unit regions within the target area; for any pairwise combination, traverse all subsets of all influencing factors and determine the weight corresponding to each subset; for each influencing factor, calculate the change in the energy-ecological efficiency prediction difference of the pairwise combination output by the spatial regression model before and after the influencing factor is added to each subset, and obtain the marginal contribution of the influencing factor in the corresponding subset; sum the marginal contributions of each subset according to the weights corresponding to each subset to obtain the original contribution value of the influencing factor in the pairwise combination.

[0047] A subset is any combination of several items selected from all influencing factors. Each possible combination is a subset, and all possible combinations constitute the power set of the set of influencing factors.

[0048] Marginal contribution measures the incremental impact of adding a certain influencing factor to a specific subset on the predicted difference in energy and eco-efficiency of the paired combination.

[0049] The original contribution value is the contribution of a factor to the energy-ecological efficiency difference of a pair of factors, determined after comprehensively considering the interaction between the influencing factor and all other factors under a certain pairing combination.

[0050] This embodiment introduces a marginal contribution decomposition mechanism based on subset traversal and weighted aggregation. For each pair of unit regions, it traverses all subsets of all influencing factors and assigns weights to each subset. Then, it calculates the change in the predicted difference in energy and ecological efficiency of the paired combination output by the spatial regression model before and after each influencing factor is added to each subset, using this as the marginal contribution. Finally, it weights and sums the marginal contributions using the weights of each subset to obtain the original contribution value. This process completely decomposes the contribution of each influencing factor to the efficiency difference of the paired combination from all possible combinations of that factor with other factors, avoiding the bias caused by simply approximating the contribution by multiplying a single regression coefficient by the difference in index values ​​in traditional methods. By traversing all subsets and aggregating them with weights, this embodiment incorporates the interaction between factors into the contribution calculation, allowing the original contribution value to more accurately reflect the true contribution of the influencing factor to the efficiency gap between regions after considering the synergistic or offsetting effects with other factors.

[0051] In one embodiment, step S103, which involves obtaining the global difference contribution of each influencing factor based on its original contribution value across all paired combinations, includes: Step S1032: For any pairing combination of unit regions, obtain the original contribution value of each influencing factor under the pairing combination; perform normalization processing on the original contribution values ​​of all influencing factors to obtain the pairing contribution rate of each influencing factor under the pairing combination.

[0052] The paired contribution rate is a relative proportion index obtained by normalizing the original contribution values ​​of each influencing factor under any paired combination. It reflects the relative contribution of each factor to the efficiency difference under that paired combination.

[0053] This step converts the original contribution values ​​of each factor under any pairing combination into relative proportions, eliminating the influence of differences in the absolute values ​​of the original contribution values ​​between different pairing combinations, and making the contributions of each factor under that pairing combination comparable. Specifically, taking a pairing combination consisting of unit region A and unit region B as an example, the original contribution values ​​of the four influencing factors—urbanization level, economic development level, industrial structure index, and degree of openness to the outside world—are first obtained, assumed to be 0.12, -0.08, 0.25, and 0.03, respectively. Subsequently, these four original contribution values ​​are normalized. The normalization method can be to first take the sum of the absolute values ​​of each original contribution value as the denominator, and then divide the absolute value of each factor's original contribution value by the denominator to obtain the pairing contribution rate of each factor. Taking the above values ​​as an example, the sum of the absolute values ​​of the four original contribution values ​​is 0.12 + 0.08 + 0.25 + 0.03 = 0.48. Therefore, the paired contribution rate of urbanization level is 0.12 / 0.48 = 0.25, the paired contribution rate of economic development level is 0.08 / 0.48 ≈ 0.167, the paired contribution rate of industrial structure index is 0.25 / 0.48 ≈ 0.521, and the paired contribution rate of openness to the outside world is 0.03 / 0.48 = 0.0625. In other embodiments, other normalization methods can also be used, such as subtracting the minimum value from each original contribution value and then dividing by the range, as long as the original contribution values ​​can be converted into the relative contribution ratio of each factor under this paired combination.

[0054] Step S1033: Based on the average of the pairing contribution rates of any influencing factor under all pairing combinations, obtain the global difference contribution degree corresponding to the influencing factor.

[0055] In this embodiment, S1032, for each unit region pairing combination, the original contribution values ​​of each influencing factor under the pairing combination are normalized to obtain the pairing contribution rate. This makes the contribution of each factor under the pairing combination present in the form of a relative proportion, eliminating the problem of incomparability of original contribution values ​​caused by the different absolute values ​​of efficiency differences between different pairing combinations. Based on this, S1033, the average of the pairing contribution rates of any influencing factor across all pairing combinations is calculated to obtain the global difference contribution degree corresponding to that factor. This aggregation method allows the global difference contribution degree to reflect the average relative contribution level of the factor to the efficiency difference in all pairing combinations within the whole region, rather than being dominated by the original contribution values ​​of a few extreme pairing combinations. Combining the global impact strength index calculated from the average absolute value of the local impact coefficient in S1025, the final analysis results simultaneously present the average impact of each factor across the entire region and its average relative contribution to driving efficiency differentiation in each regional pairing. Decision-makers can not only grasp the overall impact of urbanization level, economic development level, industrial structure index, and degree of openness, but also identify which factors are the core drivers of energy and ecological efficiency gaps in most regional pairings. This provides a complete quantitative support for the formulation of zoning control policies, covering both the characterization of local effects and the attribution of regional gaps.

[0056] For step S104, the analysis results of the factors affecting the energy and ecological efficiency of the target area are obtained based on the global effect intensity index and the global difference contribution of each influencing factor.

[0057] Specifically, the analysis combines the overall impact strength index and overall differential contribution of various influencing factors, such as urbanization level, economic development level, industrial structure index, and degree of openness to the outside world, to present a combined analysis. Decision-makers can then determine the role of each factor: if a factor has both a high overall impact strength index and a high overall differential contribution, it indicates that the factor has a significant average effect on energy and ecological efficiency across the entire region, but is also a major driver of efficiency disparities between regions; if a factor has a low overall impact strength index but a high overall differential contribution, it indicates that while the factor's overall average effect is weak, its effect varies greatly across different regions, making it a hidden key factor leading to regional development imbalances. The final analysis results can be presented in tabular or visual heatmap form, providing a quantitative basis for formulating differentiated energy and environmental management policies for each unit area within the target region.

[0058] In one embodiment, step S104, which involves obtaining the analysis results of energy and ecological efficiency influencing factors in the target area based on the global impact intensity index and global difference contribution of each influencing factor, includes: Step S1041: Determine the intensity classification benchmark threshold based on the median of the global effect intensity index corresponding to all influencing factors; construct a high effect intensity membership function based on the intensity classification benchmark threshold; input the global effect intensity index of each influencing factor into the high effect intensity membership function to obtain the corresponding high effect intensity membership value.

[0059] The intensity classification benchmark threshold refers to the median of the global effect intensity index corresponding to all influencing factors as the dividing line between high and low effect intensity. This threshold is determined by the distribution characteristics of the data itself, rather than being preset by humans.

[0060] The high-intensity membership function is a mapping function constructed around the intensity classification benchmark threshold. It is used to convert any global intensity index value into the degree of membership of the factor belonging to the high-intensity category. The function value range is 0 to 1, with a larger value indicating that the factor belongs to the high-intensity category. In this embodiment, the high-intensity membership function can be an S-shaped function, a trapezoidal function, or a linear function. For example, it can be set as follows: when the global intensity index is greater than or equal to the intensity classification benchmark threshold, the membership value is 1; when the global intensity index is less than the intensity classification benchmark threshold, the membership value decreases linearly as the difference between the index value and the threshold increases, and when the index value is 0, the membership value is 0.

[0061] Step S1042: Determine the contribution classification benchmark threshold based on the median of the global difference contribution of all influencing factors; construct a high difference contribution membership function based on the contribution classification benchmark threshold; input the global difference contribution of each influencing factor into the high difference contribution membership function to obtain the corresponding high difference contribution membership value.

[0062] The contribution classification benchmark threshold is the median of the global difference contribution corresponding to all influencing factors, which is used as the dividing line between high difference contribution and low difference contribution. This threshold is also determined by the distribution characteristics of the data itself.

[0063] The high-discrepancy contribution membership function is a mapping function constructed around the contribution classification benchmark threshold. It is used to convert any global discrepancy contribution value into the degree of membership of the factor belonging to the high-discrepancy contribution category. The function value range is 0 to 1, with a larger value indicating that the factor belongs more to the high-discrepancy contribution category. In this embodiment, the high-discrepancy contribution membership function is constructed in the same way as the high-intensity membership function.

[0064] Step S1043: Based on the high intensity membership values ​​and high differential contribution membership values ​​of each influencing factor, all influencing factors are divided into several categories of differentiated policy-oriented factors; based on the differentiated policy-oriented factors corresponding to each influencing factor, the analysis results of the energy and ecological efficiency influencing factors of the target area are obtained.

[0065] Differentiated policy guidance factors are category labels classified based on the combination of high-intensity membership values ​​and high-differentiation contribution membership values ​​of various influencing factors. These labels indicate the policy priority and regulatory direction to be adopted for that factor. In this embodiment, the various categories of differentiated policy guidance factors include global core factors, uniform constraint factors, difference source factors, and secondary factors.

[0066] Among them, the global core factor refers to the influencing factor with a large overall impact and is also the core source of regional energy and ecological efficiency gaps. It has the highest policy priority and requires differentiated intervention based on regional characteristics. The criteria for its determination are that the membership degree of high impact intensity is close to 1 and the membership degree of high difference contribution is close to 1. In one embodiment, "close to 1" means greater than a preset threshold of "close to 1", such as greater than 0.9.

[0067] Uniform constraint factors refer to factors that have a strong impact on all regions but have small differences in their effects between regions and are not the main drivers of regional energy and ecological efficiency disparities. They are suitable for a unified overall optimization policy and do not require differentiated management. The criteria for determining their impact are that the membership degree of high-intensity factors approaches 1 and the membership degree of high-difference contribution factors approaches 0. In one embodiment, "approaching 0" means less than a preset threshold for approaching 0, such as less than 0.1.

[0068] The source of difference refers to the factors that have a low average intensity of influence in a single region but are unevenly distributed across regions. These factors are the main causes of the gap in regional energy and ecological efficiency. The focus of policy regulation is to balance the distribution of indicators among regions rather than simply increasing the absolute level of indicators. The criteria for its determination are that the membership degree of high intensity of influence is close to 0 and the membership degree of high difference contribution is close to 1.

[0069] Secondary factors refer to influencing factors whose performance in both the overall influence intensity and overall difference contribution dimensions is low. These factors have the lowest priority for policy regulation and can be temporarily suspended for key regulation. The criteria for their determination are that the membership degree of high influence intensity is close to 0 and the membership degree of high difference contribution is close to 0.

[0070] In this embodiment, if either the high intensity membership value or the high difference contribution membership value of any influencing factor is in the transition range of 0.4 to 0.6, then the category corresponding to the larger of the two membership values ​​is taken as the final classification result of the influencing factor.

[0071] This embodiment uses the median of the global impact intensity index of all influencing factors as the threshold for intensity classification, and constructs a high impact intensity membership function based on this to map the global impact intensity index of each factor to a high impact intensity membership value. Similarly, it uses the median of the global difference contribution of all influencing factors as the threshold for contribution classification, and constructs a high difference contribution membership function based on this to map the global difference contribution of each factor to a high difference contribution membership value. Both median thresholds are determined by the data distribution itself, avoiding bias caused by subjective human setting. Based on this, all influencing factors are divided into four categories of differentiated policy guidance factors according to the high and low combinations of the two membership values. The analysis results of this embodiment simultaneously cover three levels: characterization of local effects, attribution of regional disparities, and classification of policy priorities. Decision-makers can not only grasp the average influence of urbanization level, economic development level, industrial structure index, and degree of openness to the outside world in the whole region, but also identify which factors are the core drivers of energy and ecological efficiency gaps between regions. Furthermore, they can directly formulate regional control strategies based on the classification results of four types of differentiated policy guidance factors, thereby providing complete quantitative support from analysis to decision-making for energy and environmental governance in the target region.

[0072] In one embodiment, step S1043, which involves obtaining the analysis results of energy and ecological efficiency influencing factors in the target area based on the differentiated policy guidance factors corresponding to each influencing factor, further includes: Step S10431: Randomly extract panel data from a portion of the unit regions to obtain a sample set; based on the sample set, perform the steps of constructing a spatial regression model, solving for local influence coefficients, calculating the global effect intensity index, calculating the global difference contribution, and classifying the results; repeat the above random sampling and calculation process to a preset number of times; Step S10432: Count the frequency with which each influencing factor is classified into the same category of differentiated policy guidance factors; based on the frequency obtained from the statistics, determine the classification confidence level corresponding to each influencing factor; Step S10433: Based on the differentiated policy guidance factors corresponding to each influencing factor and the corresponding classification confidence level, the analysis results of the influencing factors of energy and ecological efficiency in the target area are obtained.

[0073] This embodiment simulates the impact of data sample fluctuations on classification results by performing multiple random samplings of panel data from the target region and independently executing the complete analysis process from spatial regression model construction to differentiated policy guidance factor classification on each sampling sample set. Classification confidence is calculated based on the frequency with which each influencing factor is classified into the same category across multiple samplings, quantifying the stability of the classification results with continuous values ​​from 0 to 1. Finally, the differentiated policy guidance factors and classification confidence are jointly output, ensuring that the final analysis result simultaneously includes two layers of information: "which category does the factor belong to?" and "how reliable is the classification conclusion?". This provides complete quantitative support for energy and environmental governance in the target region, from analysis to decision-making, with accompanying reliability assessment.

[0074] In one embodiment, step S10433, which involves obtaining the analysis results of energy and ecological efficiency influencing factors in the target area based on the differentiated policy guidance factors corresponding to each influencing factor and their corresponding classification confidence levels, includes: Step S104331: Using the global impact intensity index as the horizontal axis and the global difference contribution as the vertical axis, generate a two-dimensional scatter plot of all influencing factors.

[0075] A two-dimensional scatter plot is a visual graph that uses the overall influence intensity index as the horizontal axis and the overall difference contribution as the vertical axis, marking all influencing factors as data points on a plane coordinate system. In one embodiment, the two-dimensional scatter plot includes classification quadrant lines and category labels for each factor. In another embodiment, the size or color of the scatter points can also represent classification confidence, allowing decision-makers to easily identify which factors are the core global factors and which are secondary factors.

[0076] Step S104332: Generate a standardized classification report based on the overall influence intensity index, overall difference contribution, high influence intensity membership value, high difference contribution membership value, differentiated policy guidance factors, and classification confidence of all influencing factors.

[0077] The standardized classification report is a structured document that summarizes key information such as the overall influence intensity index, overall difference contribution, high influence intensity membership value, high difference contribution membership value, differentiated policy guidance factor category, and classification confidence level of all influencing factors. The information is arranged in a uniform format, which facilitates decision-makers to quickly view and compare it.

[0078] Step S104333: Based on the two-dimensional scatter plot and the standardized classification report, the analysis results of the factors affecting the energy and ecological efficiency of the target area are obtained.

[0079] This embodiment maps the overall impact intensity index and overall difference contribution of all influencing factors into a two-dimensional scatter plot. The overall impact intensity index, overall difference contribution, high impact intensity membership value, high difference contribution membership value, differentiated policy guidance factor, and classification confidence level of each factor are summarized in a standardized classification report in a unified table format, ensuring information completeness and facilitating horizontal comparison. Finally, the scatter plot and classification report are jointly output, forming a complete analysis result with complementary text and graphics. This provides comprehensive quantitative support for energy and environmental governance in the target area, from analysis to decision-making, with rich graphics and text, and includes a reliability assessment.

[0080] In one embodiment, step S104333, which involves obtaining the analysis results of energy and ecological efficiency influencing factors of the target area based on the two-dimensional scatter plot and the standardized classification report, further includes: Step S1043331: For each unit region, the influencing factors whose absolute value of the local influence coefficient is greater than the preset threshold are determined as the high-priority guiding factors of that unit region.

[0081] The preset threshold is a scalar boundary value used to determine whether the absolute value of the local influence coefficient is large enough. This threshold can be set according to actual needs, such as taking the mean or median of the absolute values ​​of all local influence coefficients, or it can be set according to experience to filter the influencing factors that have a prominent effect on a specific unit area.

[0082] High-priority guiding factors refer to factors whose absolute value of local influence coefficient is greater than a preset threshold in a certain unit area. This indicates that the marginal effect of the factor on energy and ecological efficiency in that unit area is prominent, and it is a factor that needs to be focused on in that area.

[0083] Step S1043332: Determine the corresponding intervention urgency level based on the number of high-priority guidance factors in each unit area; where the larger the number of high-priority guidance factors, the higher the corresponding intervention urgency level.

[0084] The intervention urgency level is a classification of the degree of urgency of policy intervention determined by the number of high-priority guiding factors in each unit area. The larger the number of high-priority guiding factors, the more likely the area is driven or constrained by multiple strong factors, and the higher the intervention urgency level. Specifically, in one embodiment, the intervention urgency level can be divided into four levels: Level 1 (lowest urgency), with the number of high-priority guiding factors being a first preset number; Level 2 (lower urgency), with the number being a second preset number; Level 3 (higher urgency), with the number being a third preset number; and Level 4 (highest urgency), with the number being a fourth preset number. The first, second, third, and fourth preset numbers increase sequentially. Of course, the number and thresholds for each level can be adjusted according to actual needs.

[0085] Step S1043333: Generate a regional policy priority zoning map based on the intervention urgency level of each unit area; wherein, unit areas with different intervention urgency levels are matched with corresponding color blocks; The regional policy priority zoning map is a spatial visualization graphic obtained by coloring and rendering all unit areas within the target area according to their intervention urgency level. Unit areas with different intervention urgency levels are matched with corresponding color blocks, enabling decision-makers to intuitively identify which areas need priority intervention.

[0086] This step visualizes the intervention urgency level of each unit area using spatial coloring, enabling decision-makers to intuitively identify which areas require priority intervention and which can be postponed on a single map. Specifically, the geographical boundaries of all unit areas within the target region are drawn on the map, and then each unit area is filled with a corresponding color block according to its intervention urgency level. For example, Level 1 urgency is represented by green, Level 2 by yellow, Level 3 by orange, and Level 4 by red. The darker the color, the higher the intervention urgency level. Taking the data in this embodiment as an example, assuming there are 15 prefecture-level cities in the target region, unit area A is at Level 2 urgency and is filled with yellow; unit area B is at Level 4 urgency and is filled with red; the remaining unit areas are filled with corresponding color blocks according to the number of their respective high-priority guiding factors. In the final generated regional policy priority zoning map, areas with concentrated red areas indicate that these cities are simultaneously affected by multiple strong factors and are key areas for policy intervention; green areas indicate that these cities are less affected by strong factors and can postpone key interventions.

[0087] Step S1043334: Based on the two-dimensional scatter plot, the standardized classification report, and the regional policy priority zoning map, the analysis results of the factors affecting the energy and ecological efficiency of the target region are obtained.

[0088] This embodiment targets each unit region, using a preset threshold to filter out influencing factors with prominent absolute values ​​of local impact coefficients as high-priority guiding factors for that region, thus reducing the analytical granularity from the global factor level to the regional factor level. The intervention urgency level is determined based on the number of high-priority guiding factors in each region; the more factors, the higher the level, quantifying the multi-factor pressure faced by each region. The intervention urgency level of each region is presented as color blocks on a spatial map, generating a regional policy priority zoning map, enabling decision-makers to intuitively identify key intervention areas. Finally, the two-dimensional scatter plot, standardized classification report, and regional policy priority zoning map are integrated and output to form a complete analysis result that simultaneously covers global factor positioning, detailed quantitative indicators, and regional spatial priorities.

[0089] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of this application, and this application also intends to include these modifications and variations.

Claims

1. A method for analyzing the spatial heterogeneity of factors influencing energy and ecological efficiency, characterized in that, Includes the following steps: Acquire panel data for each unit area within the target area; the panel data includes spatial location information, energy and ecological efficiency parameters as the analyzed indicators, and indicator data for several influencing factors; Based on the spatial location information of each unit region, corresponding spatial scales are configured for each influencing factor, and corresponding spatial weight matrices are constructed. Using the energy-ecological efficiency parameters of each unit region as the dependent variable and the index data of the influencing factors as the independent variables, and based on the spatial weight matrices corresponding to each influencing factor, local spatial weighted regression fitting is performed on each unit region to obtain a spatial regression model and the local influence coefficient for each unit region and each influencing factor. Based on the local influence coefficients of the influencing factors in each unit region, the corresponding global influence intensity index is obtained. By iterating through all pairwise combinations of unit regions within the target area, and based on the energy efficiency prediction results of the spatial regression model, the contribution of each influencing factor to the energy-ecological efficiency difference of any pair is determined, and the original contribution value of each influencing factor under each pair is obtained; based on the original contribution value of each influencing factor in all pair combinations, the global difference contribution degree corresponding to each influencing factor is obtained. Based on the global impact intensity index and global difference contribution of each influencing factor, the analysis results of the factors affecting the energy and ecological efficiency of the target area are obtained.

2. The spatial heterogeneity analysis method for factors influencing energy and ecological efficiency according to claim 1, characterized in that, Based on the spatial location information of each unit region, corresponding spatial scales are configured for each influencing factor, and corresponding spatial weight matrices are constructed. The steps include: using the energy-ecological efficiency parameters of each unit region as the dependent variable and the index data of the influencing factors as the independent variables; and performing local spatial weighted regression fitting on each unit region based on the spatial weight matrices corresponding to each influencing factor to obtain the spatial regression model. Initialize the spatial scale of each influencing factor, and perform the following scale optimization operation: Local regression fitting is performed with each unit region as the central region: based on the spatial scale of any influencing factor, several adjacent unit regions in which the influencing factor acts are determined; based on the several adjacent unit regions and their corresponding spatial location information, a local spatial weight matrix corresponding to the influencing factor is constructed; based on the energy and ecological efficiency parameters of the central region, the index data corresponding to each influencing factor in its respective adjacent unit regions, and the local spatial weight matrix corresponding to each influencing factor, local spatial weighted regression fitting is performed. Obtain the fitting residual corresponding to the local regression of each unit region; calculate the global correction Akaike information criterion value based on the fitting residuals of all unit regions; update the spatial action scale of each influencing factor and continue to perform the scale optimization operation; when the global correction Akaike information criterion value meets the preset optimal conditions, determine the spatial action scale corresponding to each influencing factor and obtain the spatial regression model.

3. The spatial heterogeneity analysis method for factors influencing energy and ecological efficiency according to claim 2, characterized in that, The steps to obtain the local influence coefficient for each unit region and each influencing factor include: Input the panel data of each unit region within the target area into the spatial regression model; The spatial regression model is used to perform local weighted regression operations on each unit region. From the local weighted regression results corresponding to each unit region, extract the local influence coefficient corresponding to each influencing factor for each unit region.

4. The spatial heterogeneity analysis method for factors influencing energy and ecological efficiency according to claim 1, characterized in that, The steps of traversing all pairwise combinations of unit regions within the target area, determining the contribution of each influencing factor to the energy-ecological efficiency difference of any pair based on the energy efficiency prediction results of the spatial regression model, and obtaining the original contribution value of each influencing factor under each pair, include: The system iterates through all pairwise combinations of unit regions within the target area. For any pairwise combination, iterates through all subsets of all influencing factors and determines the weight corresponding to each subset. For each influencing factor, it calculates the change in the predicted difference of energy and ecological efficiency of the pairwise combination output by the spatial regression model before and after the factor is added to each subset, thus obtaining the marginal contribution of the influencing factor in the corresponding subset. The marginal contributions of each subset are weighted and summed according to their respective weights to obtain the original contribution value of the influencing factor in the pairwise combination.

5. The spatial heterogeneity analysis method for factors influencing energy and ecological efficiency according to claim 1, characterized in that, The steps for obtaining the corresponding global action intensity index based on the local influence coefficients of the influencing factors in each unit region include: Obtain all local influence coefficients corresponding to any influencing factor in all unit regions; take the absolute value of each local influence coefficient and calculate the average of all absolute values ​​to obtain the global effect intensity index corresponding to the influencing factor.

6. The spatial heterogeneity analysis method for factors influencing energy and ecological efficiency according to claim 1, characterized in that, The steps to obtain the global differential contribution of each influencing factor based on its original contribution value in all paired combinations include: For any pairing combination of unit regions, obtain the original contribution value of each influencing factor under that pairing combination; perform normalization processing on the original contribution values ​​of all influencing factors to obtain the pairing contribution rate of each influencing factor under that pairing combination; The global differential contribution of any influencing factor is obtained by averaging the contribution rates of any pairing factor across all pairing combinations.

7. The spatial heterogeneity analysis method for factors influencing energy and ecological efficiency according to claim 1, characterized in that, The steps for obtaining the analysis results of energy and ecological efficiency influencing factors in the target area based on the global impact intensity index and global difference contribution of each influencing factor include: Based on the median of the global impact intensity index corresponding to all influencing factors, a threshold for intensity classification is determined; based on the threshold for intensity classification, a high impact intensity membership function is constructed; the global impact intensity index of each influencing factor is input into the high impact intensity membership function to obtain the corresponding high impact intensity membership value. Based on the median of the global difference contribution of all influencing factors, a contribution classification benchmark threshold is determined; based on the contribution classification benchmark threshold, a high difference contribution membership function is constructed; the global difference contribution of each influencing factor is input into the high difference contribution membership function to obtain the corresponding high difference contribution membership value. Based on the high intensity membership values ​​and high differential contribution membership values ​​of each influencing factor, all influencing factors are divided into several categories of differentiated policy-oriented factors; based on the differentiated policy-oriented factors corresponding to each influencing factor, the analysis results of the influencing factors of energy and ecological efficiency in the target area are obtained.

8. The spatial heterogeneity analysis method for factors influencing energy and ecological efficiency according to claim 7, characterized in that, The steps for obtaining the analysis results of energy and ecological efficiency influencing factors in the target area based on the differentiated policy guidance factors corresponding to each influencing factor also include: Panel data from a portion of the unit regions are randomly selected to obtain a sample set. Based on the sample set, the following steps are performed: constructing a spatial regression model, solving for the local influence coefficient, calculating the global effect intensity index, calculating the global difference contribution, and classifying the results. The above random sampling and calculation process is repeated a preset number of times. The frequency with which each influencing factor is classified into the same category of differentiated policy guidance factors is statistically analyzed; based on the statistically obtained frequency, the classification confidence level corresponding to each influencing factor is determined. Based on the differentiated policy guidance factors and corresponding classification confidence levels of each influencing factor, the analysis results of the factors affecting the energy and ecological efficiency of the target area are obtained.

9. The spatial heterogeneity analysis method for factors influencing energy and ecological efficiency according to claim 8, characterized in that, The steps for obtaining the analysis results of energy and ecological efficiency influencing factors in the target area based on the differentiated policy guidance factors corresponding to each influencing factor and the corresponding classification confidence level include: A two-dimensional scatter plot of all influencing factors is generated with the overall influence intensity index as the horizontal axis and the overall difference contribution as the vertical axis. A standardized classification report is generated based on the overall influence intensity index, overall difference contribution, high influence intensity membership value, high difference contribution membership value, differentiated policy guidance factors, and classification confidence level of all influencing factors. Based on the two-dimensional scatter plot and the standardized classification report, the analysis results of the factors affecting the energy and ecological efficiency of the target area are obtained.

10. The spatial heterogeneity analysis method for factors influencing energy and ecological efficiency according to claim 9, characterized in that, The step of obtaining the analysis results of energy and ecological efficiency influencing factors in the target area based on the two-dimensional scatter plot and the standardized classification report further includes: For each unit region, influencing factors whose absolute value of local influence coefficient is greater than a preset threshold are identified as high-priority guiding factors for that unit region. The intervention urgency level is determined based on the number of high-priority guidance factors in each unit area; the greater the number of high-priority guidance factors, the higher the corresponding intervention urgency level. A regional policy priority zoning map is generated based on the intervention urgency level of each unit area; unit areas with different intervention urgency levels are matched with corresponding color blocks; Based on the two-dimensional scatter plot, the standardized classification report, and the regional policy priority zoning map, the analysis results of the factors affecting the energy and ecological efficiency of the target region are obtained.

Images