Population income identification method and equipment based on dynamic monitoring model
Through a multi-dimensional data collection and weighted scoring system, combined with the support vector machine model, the problem of neglecting regional economic differences in the existing technology is solved, the precise identification of low-income population and the fairness of resource allocation is achieved, and the accuracy of monitoring is improved sex.
Patent Information
- Application Number
- CN202411465691.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-21
- Publication Date
- 2024-11-19
AI Technical Summary
The prior art ignores socio-economic differences between regions when regional mergers and low-income rank districts, resulting in unfair resource allocation and insufficient characteristic analysis, affecting the effectiveness of monitoring of low-income populations.
A multi-dimensional data collection and weighted scoring system is adopted to establish a support vector machine model through statistical and machine learning methods, consider regional economic differences and feature integration, and ensure that the feature set fully reflects the characteristics of low-income population.
It improves the accuracy of identification and monitoring of low-income populations, and ensures the fairness of resource allocation and the scientific nature of characteristic analysis.
Smart Images

Figure CN118982448A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of social public services, and in particular to a method and device for identifying population income based on a dynamic monitoring model. Background Art
[0002] With the rapid development of social economy, today's society is paying more and more attention to low-income people. In order to better serve low-income people, relevant departments need to timely understand the dynamic distribution and related information of low-income people to facilitate the subsequent dissemination of relevant policies.
[0003] For example, a method for dynamic monitoring of personnel based on model analysis, disclosed in the prior art with publication number CN 114677253A, sets a screening model according to the identified data format of population information to ensure that the established screening model can screen the data in the corresponding format in the population information database in a targeted manner, thereby improving the screening speed and efficiency; avoiding the impact of format incompatibility on data screening; and establishing low-income level areas to facilitate relevant departments to intuitively understand the distribution and dynamic changes of low-income population in each area.
[0004] However, when merging regions and setting up low-income zones, the socioeconomic differences between regions are ignored. Simple mergers may lead to unfair treatment of some regions in resource allocation and service access. At the same time, the merging process of feature vectors may be at risk of information loss or inaccuracy. Improper feature selection leads to insufficient feature analysis of low-income populations, affecting the effectiveness of overall monitoring. Summary of the invention
[0005] In view of the shortcomings of the existing technology, the present invention provides a population income identification method and equipment based on a dynamic monitoring model, which has the advantages of introducing economic differences between regions when identifying low-income population to make the calculation results more accurate, strengthen the characteristic analysis of low-income population and improve the accuracy of monitoring, thereby solving the problems of the above-mentioned technology.
[0006] To achieve the above object, the present invention provides the following technical solution: a population income identification method based on a dynamic monitoring model, the identification method comprising the following steps: Step 1: Multi-dimensional data collection: Collect multi-dimensional data including personal income, population distribution, employment rate, education level, medical resources, infrastructure and living cost. The multi-dimensional data acquisition method includes government public data, community surveys and questionnaires; clean the collected data, process missing values and outliers, and convert the data into a standardized format; Step 2: Establish a multi-dimensional indicator system: Based on the collected multi-dimensional indicators, establish a weighted scoring system to score each region to reflect its comprehensive socio-economic status; Step 3: Regional merging and stratification: Based on the scoring system, similar regions are merged and stratified. Each stratum represents a certain range of socioeconomic conditions, ensuring that each stratum reflects different economic levels and needs. Step 4: Feature selection and merging: Use the stepwise regression feature selection technique in statistics and machine learning to screen out the key features that best explain the low-income status. Through cluster analysis, integrate the features to ensure that the feature set fully reflects the characteristics of the low-income population and reduces redundancy. Step 5: Model training: Using the features selected in step 4, use the support vector machine machine learning algorithm to train the low-income population identification model; Step 6: Model verification: Verify the accuracy and stability of the model through cross-validation and other methods; The types of data collected in step 1 include: personal income , population distribution D, employment rate E, education level Ed, medical resources H, infrastructure B, cost of living C.
[0007] Preferably, the data cleaning calculation in step 1 is: Assume the data set is ,in Indicates The sample features, missing values are Indicates that the missing values are processed using mean value interpolation calculation: ;in: Represents the processed data value; Indicates missing values; For indicators The average value of For Region In the The value of an indicator.
[0008] Preferably, the outlier processing calculation in step 1 is: For features Value , use Z-score method for outlier detection ;if , it is considered as an outlier, where Features The standard deviation of .
[0009] Preferably, the calculation for establishing the multi-dimensional indicator system in step 2 is: Features Regions, each region of Indicators ; ;in is an index of indicators: personal income, population distribution, employment rate, education level, medical resources, infrastructure and cost of living; the weight of each indicator is defined as , let the sum of these weights be 1.
[0010] Preferably, the establishment of the multi-dimensional indicator system in step 2 also includes comprehensive score calculation: Each area Overall rating It is expressed as the following weighted summation formula: ;in: For Region The overall rating of It is a region In the The value of an indicator.
[0011] Preferably, the region merging and layering calculation in step 3 is as follows: Definition of similarity measure: Use Euclidean distance to measure the similarity between regions. For two regions and , its similarity It is expressed as: ;in: and The regions are and In the The values of the indicators; using the hierarchical clustering method to merge regions, first calculate the distance matrix between all regions , and then clustering is performed based on the distance matrix.
[0012] Preferably, the clustering result is ; ;in: Indicates clustering levels; each level The socioeconomic status of the region is expressed as a composite score for all regions within the tier. The mean of .
[0013] Preferably, the feature selection and merging calculation in step 4 is as follows: Features Features and the target variable : Low-income status, the goal of stepwise regression is to find the best explanation Feature subset: Calculate each feature Importance Rating , using the regression coefficients of the linear regression model: ;in: It is a feature The regression coefficient of Use cluster analysis to integrate features and reduce redundancy. Use K-means clustering algorithm to divide features into Cluster Each cluster Represented as the average vector of all features in the cluster: ;in: Representation Cluster The number of features in ; Final feature set: Select each cluster Representative features in , the features with the highest importance scores: ;in: represents the final selected feature set; In each cluster In the feature importance score Maximum feature index .
[0014] Preferably, the model training calculation in step 5 is: Select from step 4 Features , and each sample The eigenvector of ,Label Representation sample The low-income status is 1 or -1, where 1 represents low income and -1 represents non-low income. The decision function of the SVM model is expressed as: ;in: is the weight vector; is the bias term; is the equation of the hyperplane; Represents a symbolic function, used for classification.
[0015] A low-income population identification device based on a dynamic monitoring model comprises a shell installed on the outside of the device for protection, the device is used to implement the population income identification method based on a dynamic monitoring model as described in claim 1, a single-chip microcomputer is installed inside the device, and the single-chip microcomputer has a built-in program for loading the population income identification method based on the dynamic monitoring model.
[0016] Compared with the prior art, the present invention provides a population income identification method and device based on a dynamic monitoring model, which has the following beneficial effects: 1. The present invention establishes a weighted scoring system to score each region to reflect its comprehensive socio-economic status, assigns weights to each indicator, and calculates a comprehensive score, so that the economic differences between different regions can be quantified. The calculation method of the comprehensive score takes into account the specific value of each indicator and its corresponding weight, thereby ensuring the scientificity and rationality of the score. The hierarchical clustering method is used to merge regions, and the Euclidean distance is used to measure the similarity between regions. Regions with similar economic characteristics are grouped together, while highlighting the economic differences between different levels. The support vector machine (SVM) machine learning algorithm is used to train a low-income population identification model, and the accuracy and stability of the model are verified by cross-validation and other methods to ensure the reliability of the model in practical applications, thereby achieving a more accurate calculation result of the economic differences between regions when identifying low-income populations.
[0017] 2. The present invention scores each region by constructing a weighted scoring system and calculates the socioeconomic status of the region through a comprehensive scoring method. This not only quantifies the economic characteristics of each region, but also provides a holistic perspective for the characteristic analysis of low-income groups, and can compare the socioeconomic conditions of different regions. Various features are integrated through cluster analysis, and the most representative features in each cluster are selected. This not only simplifies the feature set, but also ensures that the feature set can fully reflect the diversity and complexity of the low-income population. The support vector machine (SVM) algorithm is used to train the selected features to identify the low-income population. The powerful classification ability of the SVM model makes effective use of the influence of different features, thereby improving the accuracy of monitoring and achieving the beneficial effect of strengthening the characteristic analysis of the low-income population and improving the accuracy of monitoring. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] Figure 1 It is a schematic diagram of the structure of the present invention. DETAILED DESCRIPTION
[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0020] See also Figure 1 ,The population income identification method based on the dynamic monitoring model includes the following steps: Step 1: Multi-dimensional data collection: Collect multi-dimensional data including personal income, population distribution, employment rate, education level, medical resources, infrastructure and living cost. The multi-dimensional data acquisition methods include government public data, community surveys and questionnaires; clean the collected data, handle missing values and outliers, and convert the data into a standardized format; Step 2: Establish a multi-dimensional indicator system: Based on the collected multi-dimensional indicators, establish a weighted scoring system to score each region to reflect its comprehensive socio-economic status; Step 3: Regional merging and stratification: Based on the scoring system, similar regions are merged and stratified. Each stratum represents a certain range of socioeconomic conditions, ensuring that each stratum reflects different economic levels and needs. Step 4: Feature selection and merging: Use the stepwise regression feature selection technique in statistics and machine learning to screen out the key features that best explain the low-income status. Through cluster analysis, integrate the features to ensure that the feature set fully reflects the characteristics of the low-income population and reduces redundancy. Step 5: Model training: Using the features selected in step 4, use the support vector machine machine learning algorithm to train the low-income population identification model; Step 6: Model verification: Verify the accuracy and stability of the model through cross-validation and other methods; The types of data collected in step 1 include: personal income , population distribution D, employment rate E, education level Ed, medical resources H, infrastructure B, cost of living C.
[0021] By constructing a weighted scoring system to score each region and calculating the socioeconomic status of the region through a comprehensive score, not only can the economic characteristics of each region be quantified, but it also provides a holistic perspective for the characteristic analysis of low-income groups, which can compare the socioeconomic conditions of different regions. Various features are integrated through cluster analysis, and the most representative features in each cluster are selected. This not only simplifies the feature set, but also ensures that the feature set can fully reflect the diversity and complexity of the low-income population. The support vector machine (SVM) algorithm is used to train the selected features to identify the low-income population. The powerful classification ability of the SVM model makes effective use of the influence of different features, thereby improving the accuracy of monitoring.
[0022] Specifically, the data cleaning calculation in step 1 is: Assume the data set is ,in Indicates The sample features, missing values are Indicates; processing of missing values, using mean value interpolation calculation: ;in: Represents the processed data value; Represents missing values.
[0023] Specifically, the outlier processing calculation in step 1 is: For features Value , use Z-score method for outlier detection ;if , it is considered as an outlier, where Features The standard deviation of .
[0024] Specifically, the calculation for establishing the multi-dimensional indicator system in step 2 is: Features Regions, each region of Indicators ;in is an index of indicators: personal income, population distribution, employment rate, education level, medical resources, infrastructure and cost of living; the weight of each indicator is defined as , let the sum of these weights be 1.
[0025] Specifically, the establishment of the multi-dimensional indicator system in step 2 also includes the calculation of comprehensive scores: Each area Overall rating It is expressed as the following weighted summation formula: ;in: For Region The overall rating of It is a region In the The value of an indicator.
[0026] Specifically, the regional merging and stratification calculation in step 3 is: Definition of similarity measure: Use Euclidean distance to measure the similarity between regions. For two regions and , its similarity It is expressed as: ;in: and The regions are and In the The values of the indicators; using the hierarchical clustering method to merge regions, first calculate the distance matrix between all regions , and then clustering is performed based on the distance matrix.
[0027] Specifically, the clustering results are ; ;in: Indicates clustering levels; each level The socioeconomic status of the region is expressed as a composite score for all regions within the tier. The mean of .
[0028] Specifically, the feature selection and merging calculation in step 4 is as follows: Features Features and the target variable : Low-income status, the goal of stepwise regression is to find the best explanation Feature subset: Calculate each feature Importance Rating , using the regression coefficients of the linear regression model: ;in: It is a feature The regression coefficient of Use cluster analysis to integrate features and reduce redundancy. Use K-means clustering algorithm to divide features into Cluster Each cluster Represented as the average vector of all features in the cluster: ;in: Representation Cluster The number of features in ; Final feature set: Select each cluster Representative features in , the features with the highest importance scores: ;in: represents the final selected feature set; In each cluster In the feature importance score Maximum feature index .
[0029] Specifically, the model training calculation in step 5 is: Select from step 4 Features , and each sample The eigenvector of ,Label Representation sample The low-income status is 1 or -1, where 1 represents low income and -1 represents non-low income. The decision function of the SVM model is expressed as: ;in: is the weight vector; is the bias term; is the equation of the hyperplane; Represents a symbolic function, used for classification.
[0030] A low-income population identification device based on a dynamic monitoring model comprises a protective shell installed on the outside of the device, the device is used to implement the population income identification method based on a dynamic monitoring model of claim 1, a single-chip microcomputer is installed inside the device, and the single-chip microcomputer has a built-in program for loading the population income identification method based on the dynamic monitoring model.
[0031] By establishing a weighted scoring system, scoring each region to reflect its comprehensive socio-economic conditions, assigning weights to each indicator, and calculating a comprehensive score, the economic differences between different regions can be quantified. The calculation method of the comprehensive score takes into account the specific value of each indicator and its corresponding weight, thereby ensuring the scientificity and rationality of the score.
[0032] Regional mergers are carried out through hierarchical clustering method, and Euclidean distance is used to measure the similarity between regions. Regions with similar economic characteristics are grouped together, while the economic differences between different levels are highlighted. The support vector machine (SVM) machine learning algorithm is used to train the low-income population identification model, and the accuracy and stability of the model are verified through cross-validation and other methods to ensure the reliability of the model in practical applications.
[0033] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A population income identification method based on a dynamic monitoring model, characterized in that: The identification method comprises the following steps: Step 1: Multi-dimensional data collection: Collect multi-dimensional data including personal income, population distribution, employment rate, education level, medical resources, infrastructure and living cost. The multi-dimensional data acquisition method includes government public data, community surveys and questionnaires; clean the collected data, process missing values and outliers, and convert the data into a standardized format; Step 2: Establish a multi-dimensional indicator system: Based on the collected multi-dimensional indicators, establish a weighted scoring system to score each region to reflect its comprehensive socio-economic status; Step 3: Regional merging and stratification: Based on the scoring system, similar regions are merged and stratified. Each stratum represents a certain range of socioeconomic conditions, ensuring that each stratum reflects different economic levels and needs. Step 4: Feature selection and merging: Use the stepwise regression feature selection technique in statistics and machine learning to screen out the key features that best explain the low-income status. Through cluster analysis, integrate the features to ensure that the feature set fully reflects the characteristics of the low-income population and reduces redundancy. Step 5: Model training: Using the features selected in step 4, use the support vector machine machine learning algorithm to train the low-income population identification model; Step 6: Model verification: Verify the accuracy and stability of the model through cross-validation and other methods; The types of data collected in step 1 include: personal income , population distribution D, employment rate E, education level Ed, medical resources H, infrastructure B, cost of living C; The feature selection and merging calculation in step 4 are as follows: Features Features and the target variable : Low-income status, the goal of stepwise regression is to find the best explanation Feature subset: Calculate each feature Importance Rating , using the regression coefficients of the linear regression model: ;in: It is a feature The regression coefficient of Use cluster analysis to integrate features and reduce redundancy. Use K-means clustering algorithm to divide features into Cluster Each cluster Represented as the average vector of all features in the cluster: ;in: Representation Cluster The number of features in ; Final feature set: Select each cluster Representative features in , the features with the highest importance scores: ;in: represents the final selected feature set; In each cluster In the feature importance score Maximum feature index .
2. The method for identifying population income based on a dynamic monitoring model according to claim 1 is characterized in that: The data cleaning calculation in step 1 is: Assume the data set is ,in Indicates The sample features, missing values are Indicates that the missing values are processed using mean value interpolation calculation: ;in: Represents the processed data value; Indicates missing values; For indicators The average value of For Region In the The value of an indicator.
3. The method for identifying population income based on a dynamic monitoring model according to claim 2 is characterized in that: The outlier processing calculation in step 1 is: For features Value , use Z-score method for outlier detection ;if , it is considered as an outlier, where Features The standard deviation of .
4. The method for identifying population income based on a dynamic monitoring model according to claim 1, characterized in that: The calculation for establishing the multi-dimensional indicator system in step 2 is: Features Regions, each region of Indicators ; ;in is an index of indicators: personal income, population distribution, employment rate, education level, medical resources, infrastructure and cost of living; the weight of each indicator is defined as , let the sum of these weights be 1.
5. The method for identifying population income based on a dynamic monitoring model according to claim 1, characterized in that: The establishment of the multi-dimensional indicator system in step 2 also includes comprehensive score calculation: Each area Overall rating It is expressed as the following weighted summation formula: ;in: For Region The overall rating of It is a region In the The value of an indicator.
6. The method for identifying population income based on a dynamic monitoring model according to claim 1, characterized in that: The region merging and stratification calculation in step 3 is: Definition of similarity measure: Use Euclidean distance to measure the similarity between regions. For two regions and , its similarity It is expressed as: ;in: and The regions are and In the The values of the indicators; using the hierarchical clustering method to merge regions, first calculate the distance matrix between all regions , and then clustering is performed based on the distance matrix.
7. The method for identifying population income based on a dynamic monitoring model according to claim 6 is characterized in that: The clustering result is ; ;in: Indicates clustering levels; each level The socioeconomic status of the region is expressed as a composite score for all regions within the tier. The mean of .
8. The method for identifying population income based on a dynamic monitoring model according to claim 1, characterized in that: The model training calculation in step 5 is: Select from step 4 Features , and each sample The eigenvector of ,Label Representation sample The low-income status is 1 or -1, where 1 represents low income and -1 represents non-low income. The decision function of the SVM model is expressed as: ;in: is the weight vector; is the bias term; is the equation of the hyperplane; Represents a symbolic function, used for classification.
9. A low-income population identification device based on a dynamic monitoring model, comprising a housing installed outside the device for protection, characterized in that: The device is used to implement the population income identification method based on a dynamic monitoring model as described in claim 1. A single-chip microcomputer is installed inside the device, and the single-chip microcomputer has a built-in program for loading the population income identification method based on a dynamic monitoring model.
Citation Information
Patent Citations
A poverty space classification recognition method and device based on big data and machine learning
CN109948737A
County town classification method based on two-step clustering thought
CN110610446A
Dynamic personnel monitoring method based on model analysis
CN114677253A
Urban space income difference measurement method, system and device and storage medium
CN115146945A
Identification method, device and equipment and readable storage medium
CN115587317A
