Search

CN-122022559-A - Sustainability index construction method based on social network analysis and dynamic weighting

CN122022559ACN 122022559 ACN122022559 ACN 122022559ACN-122022559-A

Abstract

The invention relates to the technical field of sustainability evaluation and discloses a sustainability index construction method based on social network analysis and dynamic weighting, which comprises the following steps of S1, data collection and pretreatment, namely, widely collecting data related to sustainability in multiple fields such as economy, environment, society and the like, cleaning the collected original data, removing repeated, wrong and excessive data records of missing values, and supplementing a small quantity of missing values by adopting methods such as mean filling, regression prediction and the like. According to the invention, complex association relations among economic, environmental and social factors are analyzed and mined through a social network, compared with a traditional linear weighting method, the core influence indexes and conduction effects among the indexes can be accurately identified, the unilateral performance of single-dimension evaluation is avoided, and the sustainability index is more attached to the actual situation of the overall collaborative development of the system, and the fitting degree with the actual development quality is improved by more than 12%.

Inventors

  • ZHOU XIANG

Assignees

  • 广州大学

Dates

Publication Date
20260512
Application Date
20260108

Claims (10)

  1. 1. The method for constructing the sustainability index based on social network analysis and dynamic weighting is characterized by comprising the following steps of: s1, data collection and preprocessing, namely, widely collecting data related to sustainability in multiple fields such as economy, environment, society and the like, cleaning the collected original data, removing repeated, erroneous and missing data records, and supplementing a small amount of missing values by adopting methods such as mean filling, regression prediction and the like; S2, constructing a social network model, namely determining network nodes as sustainability related factors in various fields of economy, environment, society and the like, determining the association relation among factors by a method of correlation analysis, grangejie causal test and the like, defining network edges according to the association strength, and constructing the social network model by adopting a complex network modeling algorithm such as a small-world network model, a non-scale network model and the like; S3, designing a dynamic weighting mechanism, namely predicting the development trend of each factor and the potential influence on sustainable development by using methods such as time sequence analysis, machine learning prediction models and the like, setting the frequency of weight updating such as monthly, quarterly or annually according to the prediction result and the change of the influence degree of each factor on the sustainable development in different periods; S4, index calculation and evaluation, namely substituting the processed data into a pre-constructed sustainability index calculation formula, wherein the formula is based on social network analysis results and dynamic weights, and calculates to obtain sustainability indexes by considering the interrelation and importance degree of each factor; and S5, calibrating and verifying the synthesized sustainability index, wherein the calibration adopts a historical data backtracking method, and the index calculation result is compared with the contemporaneous development quality evaluation result.
  2. 2. The method for constructing the sustainability index based on the social network analysis and dynamic weighting according to claim 1, wherein in the step S1, the sustainability evaluation index system comprises an economic development dimension, an ecological environment dimension, a social folk dimension and a resource utilization dimension, the index quantity ratio under each dimension is 2:3:3:2, the indexes are subjected to correlation test and discrimination analysis screening, the absolute value of the correlation coefficient is less than or equal to 0.75, and the discrimination coefficient is more than or equal to 0.3.
  3. 3. The method for constructing a sustainability index based on social network analysis and dynamic weighting according to claim 1, wherein the data preprocessing in step S1 comprises the following sub-steps: s11, filling missing data by adopting a Lagrange interpolation method, and correcting data which are continuously missing for more than 3 time nodes by combining with the contemporaneous data of the same type region; s12, carrying out standardization treatment on the original data by adopting a Z-score standardization method, and unifying index directions of negative indexes by adopting a mode of 'reverse standardization'; S13, identifying abnormal data according to a 3 sigma rule, and adopting Winsorize tail-shrinking treatment on the abnormal data, wherein the tail-shrinking proportion is 5%.
  4. 4. The method for constructing a sustainability index based on social network analysis and dynamic weighting according to claim 1, wherein in the step S2, when the social network model is constructed, each index is used as a network node, the bias correlation coefficient between indexes is used as the weight of the edge, and the calculation of the bias correlation coefficient is based on the multiple linear regression model to strip the interference of the third party index, specifically through the statsmodels library of SPSS or Python.
  5. 5. The method for constructing a sustainability index based on social network analysis and dynamic weighting according to claim 1, wherein the social network analysis in step S2 comprises the following sub-steps: S21, calculating overall network indexes such as network density, average degree, clustering coefficient and the like, and evaluating the overall association degree of an index system; S22, calculating the centrality, the approximate centrality and the intermediate centrality of each node, fusing the three centralities by adopting an entropy weight method to obtain a core index comprehensive score, and selecting an index of 30% before the score as a core index; and S23, drawing a social network map by adopting UCINET software, and visually displaying the association structure of the core index and the non-core index.
  6. 6. The method for constructing a sustainability index based on social network analysis and dynamic weighting according to claim 1, wherein the dynamic weighting mechanism in the step S3 comprises two stages of basic weight calculation and dynamic adjustment, wherein the basic weight is determined based on the product of a core index comprehensive score and an associated strength, the dynamic adjustment is triggered based on sliding of a time window, the length of the time window is set to 3-12 evaluation periods, and when the variation coefficient of an index in the window exceeds 0.2, the self-adaptive adjustment of the weight is triggered.
  7. 7. The method for constructing the sustainability index based on social network analysis and dynamic weighting according to claim 6 is characterized in that an index smoothing method is adopted in the dynamic adjustment stage to calculate a weight adjustment coefficient, the value range of the smoothing coefficient alpha is 0.1-0.3, the specific value is determined according to an evaluation scene, wherein alpha is 0.1-0.2 for macroscopic region evaluation and alpha is 0.2-0.3 for microscopic enterprise evaluation.
  8. 8. The method for constructing the sustainability index based on the social network analysis and the dynamic weighting according to claim 1, wherein in the step S4, a composite mode of core index weighting and non-core index correction is adopted during weighting synthesis, wherein the core index weight is 60% -70%, the non-core index weight is 30% -40%, and the non-core index weight is positively adjusted according to the association strength between the non-core index weight and the core index.
  9. 9. The method for constructing the sustainability index based on the social network analysis and the dynamic weighting according to claim 1, wherein in the step S5, when the error exceeds 5%, the smooth coefficient in the dynamic weighting mechanism is adjusted, and the verification adopts Kappa consistency test, and the consistency coefficient is more than or equal to 0.7 and is regarded as passing the verification.
  10. 10. The method for constructing a sustainability index based on social network analysis and dynamic weighting according to claim 1, wherein in step S2, when the evaluation scene is regional sustainability evaluation, a virtual node of "regional policy fitness" is additionally introduced into the social network model, and the association strength between the node and each index node is quantitatively determined according to the index mentioned frequency in the policy file.

Description

Sustainability index construction method based on social network analysis and dynamic weighting Technical Field The invention relates to the technical field of sustainability evaluation, in particular to a sustainability index construction method based on social network analysis and dynamic weighting. Background The core of the sustainability index construction scheme is to solve the problems of the traditional index neglecting index association and dynamic change, firstly, the complex association among sustainable development indexes such as economy, society and environment is mined through social network analysis, the indexes are regarded as network nodes, the mutual influence degree of the indexes is reflected by the node connection strength, the basic weight is determined according to the mutual influence degree, the dynamic weighting technology is introduced, the weight distribution is adjusted in real time according to the importance change of each index in different time sequence stages instead of a fixed value, and finally, the formed sustainability index is integrated, so that the system association among the indexes can be accurately embodied, the scene change can be dynamically adapted and developed, the evaluation result is more fit with the actual, and the scientificity and timeliness of the sustainable development state evaluation are improved. In most of the current sustainable index construction methods, an index weight adopts a static allocation mode, the index is calculated only according to the initially set weight, dynamic changes of importance of each index in different development stages cannot be captured, the index has poor suitability for scenes, and a real-time state of sustainable development is difficult to reflect. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a sustainable index construction method based on social network analysis and dynamic weighting, and solves the problems that in the existing most sustainable index construction methods, index weights are mostly in a static allocation mode, indexes are calculated only according to the initially set weights, dynamic changes of importance of each index in different development stages cannot be captured, the adaptability of the indexes to scenes is poor, and the real-time state of sustainable development is difficult to reflect. The invention aims at realizing the purposes by adopting the following technical scheme that the sustainability index construction method based on social network analysis and dynamic weighting comprises the following steps: s1, data collection and preprocessing, namely, widely collecting data related to sustainability in multiple fields such as economy, environment, society and the like, cleaning the collected original data, removing repeated, erroneous and missing data records, and supplementing a small amount of missing values by adopting methods such as mean filling, regression prediction and the like; S2, constructing a social network model, namely determining network nodes as sustainability related factors in various fields of economy, environment, society and the like, determining the association relation among factors by a method of correlation analysis, grangejie causal test and the like, defining network edges according to the association strength, and constructing the social network model by adopting a complex network modeling algorithm such as a small-world network model, a non-scale network model and the like; S3, designing a dynamic weighting mechanism, namely predicting the development trend of each factor and the potential influence on sustainable development by using methods such as time sequence analysis, machine learning prediction models and the like, setting the frequency of weight updating such as monthly, quarterly or annually according to the prediction result and the change of the influence degree of each factor on the sustainable development in different periods; S4, index calculation and evaluation, namely substituting the processed data into a pre-constructed sustainability index calculation formula, wherein the formula is based on social network analysis results and dynamic weights, and calculates to obtain sustainability indexes by considering the interrelation and importance degree of each factor; and S5, calibrating and verifying the synthesized sustainability index, wherein the calibration adopts a historical data backtracking method, and the index calculation result is compared with the contemporaneous development quality evaluation result. Further, in step S1, the sustainability evaluation index system includes an economic development dimension, an ecological environment dimension, a social folk dimension and a resource utilization dimension, the index number ratio in each dimension is 2:3:3:2, each index is subjected to correlation test and discrimination analysis screening, the absolute value of the correlation coefficient is