Group insurance renewal data intelligent comparison method and system

Abstract

The invention discloses a group insurance renewal data intelligent comparison method and system, and mainly relates to the technical field of application of big data analysis in insurance business. Comprising the following steps that multi-dimensional data of group insurance personnel is acquired, and the multi-dimensional data comprises basic information, historical insurance renewal records, non-insurance renewal reasons, occupational types, health states and claim settlement records; preprocessing the multi-dimensional data, including data cleaning, duplicate removal, formatting and data desensitization, and constructing a three-level label system; based on the preprocessed data, performing insurance renewal behavior analysis by using a machine learning model, including training an insurance renewal trend analysis model, calculating an insurance renewal influence coefficient and generating an insurance renewal prediction result; and according to an analysis result, outputting an insurance renewal rate report, attribute analysis of the non-insurance-renewal personnel and a service strategy suggestion. The method has the beneficial effects that dynamic prediction of group insurance renewal behaviors, accurate tracing of multi-dimensional attributes and automatic generation of personalized service strategies are realized.

Description

Technical Field

[0001] This invention relates to the field of big data analytics in insurance business, specifically a method and system for intelligent comparison of group insurance renewal data. Background Technology

[0002] As the digital transformation of the insurance industry deepens, group insurance businesses face challenges such as large data volumes, complex behavioral patterns, and difficulty in tracing the reasons for churn when managing the renewal of large numbers of policyholders. Existing technologies mainly focus on risk assessment and user profile construction during the underwriting stage, such as using machine learning models like support vector machines to classify and grade insured groups, or using multi-dimensional information to build a policyholder tagging system for intelligent underwriting.

[0003] However, existing technologies have the following drawbacks: 1. Lack of dynamic analysis and prediction capabilities for group policy renewal behavior, making it impossible to provide effective early warnings of future changes in renewal rates; 2. The lack of a correlation analysis model between renewal behavior and the insured's multi-dimensional attributes (such as occupation, health, and claims history) makes it impossible to accurately trace the reasons for non-renewal; 3. There is a lack of an automatic business strategy generation mechanism based on the analysis results, making it difficult to directly translate the analysis conclusions into actionable business actions.

[0004] The reasons for the above defects are as follows: Traditional methods often focus on static data statistics and post-event analysis, failing to introduce predictive models that enable continuous learning. Due to the dispersed dimensions of data and insufficient correlation mining, it is difficult to establish effective attribute-behavior mapping relationships. Existing systems are disconnected from data analysis and business decision support, failing to form a closed-loop output from insight to strategy.

[0005] Therefore, there is an urgent need for a method and system for intelligent comparison of group insurance renewal data to solve the above problems. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for intelligent comparison of group insurance renewal data. This invention realizes dynamic prediction of group insurance renewal behavior, accurate tracing of multi-dimensional attributes, and automatic generation of personalized business strategies.

[0007] To achieve the above objectives, the present invention employs the following technical solution: On the one hand, this invention provides a method for intelligent comparison of group insurance renewal data, including the following steps: Step S1: Obtain multi-dimensional data of group policy participants, including basic information, historical renewal records, reasons for non-renewal, occupation type, health status, and claims records; Step S2: Preprocess the multi-dimensional data, including data cleaning, deduplication, formatting and data anonymization, and construct a three-level tagging system, which includes basic attribute tags, risk attribute tags and behavioral attribute tags; Step S3: Based on the preprocessed data, use a machine learning model to analyze renewal behavior, including training a renewal trend analysis model, calculating the renewal impact coefficient, and generating renewal prediction results; Step S4: Based on the analysis results, output a renewal rate report, attribute analysis of non-renewing personnel, and business strategy recommendations.

[0008] Preferably, in step S2, the data desensitization specifically involves: retaining the first 6 digits and the last 4 digits of the ID number, and anonymizing the name.

[0009] Preferably, in step S3, the three-level labeling system includes: Basic attribute tags include age and gender; Risk attribute tags include occupational risk and claims frequency; Behavioral attribute tags include renewal history and reasons for non-renewal.

[0010] Preferably, in step S3, the renewal trend analysis model is trained using a support vector machine algorithm, specifically as follows: Historical renewal data was divided into a training set and a test set, with a ratio of 7:3. The support vector machine model is trained on the training set and the model accuracy is verified on the test set. When the accuracy is lower than 95%, the model parameters are re-optimized. An online learning mechanism is introduced to regularly update model parameters to maintain prediction accuracy.

[0011] Preferably, the step of calculating the renewal impact coefficient is implemented through a renewal impact coefficient model, which calculates the renewal impact coefficient using the following formula. : ; in, Indicates the first The factors include occupation type, health status, and claims history. Indicates the first The weights of each influencing factor are obtained through training on historical data, and are as follows: The occupation type factor is assigned a value based on the occupational risk level, with a value range of 0-1. The health status factor is assigned a value based on the health assessment results, with a value range of 0-1; The claims record factor is assigned a value based on the claims frequency and amount, with a value range of 0-1. Weight The prediction error is minimized by optimizing using gradient descent.

[0012] On the other hand, the present invention provides a group insurance renewal data intelligent comparison system for implementing the group insurance renewal data intelligent comparison method as described above, including: The data acquisition module is used to obtain multi-dimensional data of group insurance participants; The data processing module is used to preprocess the multi-dimensional data and construct labels; The intelligent analysis module is used to analyze renewal behavior using machine learning models; The business application module is used to output analysis results and business strategy recommendations.

[0013] Preferably, the data acquisition module is used for: The basic information, historical renewal records, reasons for non-renewal, occupation type, health status, and claims records of the group's insured individuals are collected. Sensitive information is processed using data anonymization techniques, including retaining the first 6 and last 4 digits of the ID number and anonymizing the name.

[0014] Preferably, the data processing module is specifically used for: The collected data is cleaned, deduplicated, and formatted. Construct a three-level tagging system, including basic attribute tags, risk attribute tags, and behavioral attribute tags; The basic attribute tags include age and gender, the risk attribute tags include occupational risk and claim frequency, and the behavioral attribute tags include renewal history and reasons for non-renewal.

[0015] Preferably, the intelligent analysis module is specifically used for: The renewal trend analysis model was trained using the support vector machine algorithm. Historical renewal data was used as the training set, and the ratio of the training set to the test set was 7:3. The accuracy of the model was verified, and the threshold was set to 95%. Calculate the renewal impact factor using the formula. Quantify the impact of occupation, health, and claims factors on renewal behavior; Generate renewal prediction results, output the probability of future renewal, and issue risk warnings for low-probability group policies.

[0016] Preferably, the business application module is specifically used for: Output renewal rate analysis reports, attribute tags for those who did not renew their policies, and profiles of new policyholders; Generate business strategy recommendations, including optimizing coverage for younger groups and providing customized renewal plans for high-risk occupations; It supports model iteration and optimization, and regularly updates model parameters and labeling system through an online learning mechanism.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. Enabled dynamic analysis and accurate prediction of renewal behavior: By constructing a renewal trend analysis model based on support vector machines and introducing an online learning mechanism, the system can continuously learn and dynamically predict group insurance renewal rates, effectively overcoming the lag of traditional static statistical methods and providing forward-looking guidance for business decisions.

[0018] 2. A multi-dimensional attribute correlation analysis mechanism was established: The system constructs a three-level tagging system that includes basic attributes, risk attributes, and behavioral attributes, and combines it with a renewal impact coefficient model (quantifying the degree of influence of each factor) to achieve accurate tracing of non-renewing policyholders, revealing the intrinsic relationship between multi-dimensional attributes such as occupation type, health status, and claims records and renewal behavior.

[0019] 3. A closed-loop output from data analysis to business strategy has been formed: After completing the data analysis, the system can automatically output analysis results including renewal rate reports, attribute analysis of non-renewing personnel, and profiles of new policyholders. Based on these results, it can generate personalized business strategy suggestions, such as "optimizing coverage for younger groups" and "providing exclusive renewal plans for high-risk occupations," effectively solving the problem of disconnect between analysis conclusions and business practices.

[0020] 4. Ensures the security and compliance of data processing: Data anonymization technology is used at the data collection layer (the first 6 and last 4 digits of the ID number are retained, and the name is anonymized), and a hierarchical permission mechanism is established to ensure the compliant use of personal privacy data while achieving data analysis goals.

[0021] 5. Supports continuous system optimization and iteration: By introducing an online learning mechanism and regular label verification, the system can continuously optimize model parameters and labeling system based on newly generated business data, ensuring the long-term accuracy and effectiveness of analysis results and enabling the system to self-evolve. Attached Figure Description

[0022] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0023] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined in this application.

[0024] In this invention, terms such as "upper," "lower," "left," "right," "front," "back," "vertical," "horizontal," "side," and "bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used only to facilitate the description of the structural relationships of the various components or elements of this invention and do not specifically refer to any component or element in this invention. They should not be construed as limiting the invention.

[0025] Example: like Figure 1 As shown, this embodiment provides a method for intelligent comparison of group insurance renewal data, including the following steps: Step S1: Obtain multi-dimensional data of group policy participants, including basic information, historical renewal records, reasons for non-renewal, occupation type, health status, and claims records; Step S2: Preprocess the multi-dimensional data, including data cleaning, deduplication, formatting, and data anonymization, and construct a three-level tagging system, which includes basic attribute tags, risk attribute tags, and behavioral attribute tags; Step S3: Based on the preprocessed data, use a machine learning model to analyze renewal behavior, including training a renewal trend analysis model, calculating the renewal impact coefficient, and generating renewal prediction results; Step S4: Based on the analysis results, output a renewal rate report, attribute analysis of non-renewing personnel, and business strategy recommendations.

[0026] like Figure 2 As shown, this embodiment also provides a group insurance renewal data intelligent comparison system, including: The data acquisition module is used to obtain multi-dimensional data of group insurance participants; The data processing module is used to preprocess the multi-dimensional data and construct labels; The intelligent analysis module is used to analyze renewal behavior using machine learning models; The business application module is used to output analysis results and business strategy recommendations.

[0027] Below, this embodiment takes Insurance Company A as an example to describe in detail the deployment of the system in Insurance Company A and the implementation of related methods.

[0028] The specific architecture of the system in this embodiment at Insurance Company A is as follows: 1. Data Acquisition Layer Deployment: Establish a unified data interface platform to connect with data sources such as internal core business systems, claims systems, and customer service systems; Deploy a data anonymization gateway to perform real-time anonymization processing on all sensitive information; Establish data quality standards to ensure the integrity and accuracy of collected data; 2. Data processing layer construction: A big data processing platform was built, adopting a distributed storage architecture, capable of processing terabytes of data per day. A data quality monitoring system was established, with 128 data quality checkpoints set up; A three-tiered tag management system was established, defining a total of 15 basic attribute tags, 12 risk attribute tags, and 9 behavioral attribute tags; 3. Construction of the intelligent analysis layer: Deploy a machine learning platform and integrate multiple algorithm models; Establish a model training pipeline to support automated model training and evaluation; Set up a model performance monitoring and early warning mechanism to track model performance in real time; 4. Implementation of the business application layer: Develop a renewal analysis workbench and provide a visual analysis interface; Establish a strategy recommendation engine to automatically generate business strategy suggestions; It enables seamless integration with business systems, supporting rapid strategy implementation.

[0029] The method of this embodiment is specifically implemented in Insurance Company A as follows: 1. Data preprocessing and anonymization: In practical applications, Company A has refined the data preprocessing steps: Data cleaning: Establish an outlier detection rule base, which includes 68 business rules and 42 statistical rules; Data anonymization: A tiered anonymization strategy is adopted, with the ID number retaining the first 6 and last 4 characters, and the name anonymized using hash encryption; Data standardization: Establish unified data standards and specifications, including date formats, code systems, and numerical units; The specific desensitization process is as follows: Raw data → Format validation → Sensitive information identification → Graded desensitization processing → Quality inspection → Standardized output; 2. Company A expanded and refined its three-tier labeling system based on actual business needs: Basic attribute tags include: demographic tags, occupational characteristic tags, and policy characteristic tags, specifically: Demographic tags: age group (18-25, 26-35, 36-45, 46-55, 55+), gender, region; Occupational characteristic tags: occupational category, job level, length of service, industry type; Policy feature tags: insured years, coverage amount, product type; Risk attribute tags include: occupational risk tags, health risk tags, and claims risk tags, specifically: Occupational risk labels: scored based on dimensions such as occupational disease incidence rate and work injury probability; Health risk labeling: An assessment combining physical examination data, medical records, etc. Claims risk tags: claim frequency, average claim amount, claim trend; Behavioral attribute tags include: renewal behavior tags, interaction behavior tags, and value behavior tags, specifically: Renewal behavior tags: renewal history, renewal timeliness, and preferred renewal channels; Interactive behavior tags: service inquiry frequency, complaint records, satisfaction rating; Value behavior tags: customer value grading, potential demand characteristics; 3. Company A uses the Support Vector Machine (SVM) algorithm to train a renewal trend analysis model. The specific implementation process is as follows: Data preparation stage: Historical renewal data from the past three years were selected as training samples, totaling 2.85 million records. The training set (2019-2021) and the test set (2022) were divided into two sets in chronological order. Feature engineering was performed, and 23 feature variables were ultimately selected for the model. Model training phase: Radial basis functions (RBF) are used as the kernel function; The hyperparameters were optimized using a grid search method, and the penalty parameter C=1.0 and the kernel function parameter γ=0.1 were finally determined. Implement 5-fold cross-validation to ensure model stability; Model validation results: On the test set, the accuracy was 96.3%, the precision was 95.8%, and the recall was 96.1%. The model's AUC value reached 0.98, indicating that the model has high discrimination and classification accuracy; Through business validation, the model's prediction results achieved an accuracy rate of 92% in matching actual business performance. 4. The renewal impact coefficient model is calculated using the following formula: ; The specific definitions and calculation methods for each parameter are as follows: Impact Factor Defined as: Occupational type factor : Assigned based on occupational risk level, with a value range of 0-1; Low-risk occupations (such as administrative staff): 0.8-1.0; Medium-risk occupations (such as technical personnel): 0.5-0.8; High-risk occupations (such as construction workers): 0-0.5; Health status factors Assigning values ​​based on health assessment scores: Excellent (no history of serious illness): 0.8-1.0; Good (with minor illness): 0.6-0.8; General (with chronic diseases): 0.3-0.6; Poor (with serious illness): 0-0.3; Claims record factor A comprehensive score combining claim frequency and amount: Claims frequency score: 1 point for 1 or fewer claims per year, and deduct 0.2 points for each additional claim. Claims Amount Scoring: 1 point is awarded if the annual claims amount is less than or equal to the premium, and 0.3 points are deducted for every 50% exceeding the premium; Overall score = (frequency score + amount score) / 2; Weight Training process: Optimize the weight parameters using gradient descent. The learning rate was set to 0.01, and the number of iterations was 1000. The final weight values ​​are: , , ; 5. Company A has established an online learning and updating mechanism for its models: The model is automatically retrained monthly, using data from the most recent 36 months. Set a model performance degradation warning to immediately trigger optimization when the accuracy drops by more than 2%; Establish model version management to support rapid model rollback.

[0030] Taking a large manufacturing enterprise client, Company A, as an example, this enterprise has 5,200 insured employees, and its annual renewal rate fluctuates around 78%. Through the implementation of this system: 1. Data Acquisition and Processing: (1) Automatically collect the company's insurance data for the past 3 years, including basic information, claims records, service interaction records, etc.; (2) Complete data cleaning and de-identification to generate a standardized dataset; (3) Automatic tagging, generating 36 attribute tags for each insured person; 2. Intelligent analysis and execution: (1) The renewal trend analysis model predicts that the customer's renewal rate for the next year is 76.8%, which is classified as medium risk. (2) Analysis of the impact coefficient of renewal insurance shows that occupational risk is the main influencing factor (contribution rate 42%). (3) Identify 520 high-risk personnel and analyze their common characteristics; 3. Business strategy generation: (1) For high-risk occupational groups, the “Optimized Accident Protection During Work” program is recommended; (2) For groups with poor health, it is recommended to “increase health management services”; (3) Design a “premium discount tiered scheme” for groups with a strong willingness to renew their insurance.

[0031] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A method for intelligent comparison of group insurance renewal data, characterized in that, Includes the following steps: Step S1: Obtain multi-dimensional data of group policy participants, including basic information, historical renewal records, reasons for non-renewal, occupation type, health status, and claims records; Step S2: Preprocess the multi-dimensional data, including data cleaning, deduplication, formatting and data anonymization, and construct a three-level tagging system, which includes basic attribute tags, risk attribute tags and behavioral attribute tags; Step S3: Based on the preprocessed data, use a machine learning model to analyze renewal behavior, including training a renewal trend analysis model, calculating the renewal impact coefficient, and generating renewal prediction results; Step S4: Based on the analysis results, output a renewal rate report, attribute analysis of non-renewing personnel, and business strategy recommendations.

2. The intelligent comparison method for group insurance renewal data according to claim 1, characterized in that, In step S2, the data desensitization specifically involves: retaining the first 6 digits and the last 4 digits of the ID number, and anonymizing the name.

3. The intelligent comparison method for group insurance renewal data according to claim 1, characterized in that, In step S3, the three-level tagging system includes: Basic attribute tags include age and gender; Risk attribute tags include occupational risk and claims frequency; Behavioral attribute tags include renewal history and reasons for non-renewal.

4. The intelligent comparison method for group insurance renewal data according to claim 1, characterized in that, In step S3, the renewal trend analysis model is trained using a support vector machine algorithm, specifically as follows: Historical renewal data was divided into a training set and a test set, with a ratio of 7:

3. The support vector machine model is trained on the training set and the model accuracy is verified on the test set. When the accuracy is lower than 95%, the model parameters are re-optimized. An online learning mechanism is introduced to regularly update model parameters to maintain prediction accuracy.

5. The intelligent comparison method for group insurance renewal data according to claim 4, characterized in that, The step of calculating the renewal impact factor is implemented through the renewal impact factor model, which calculates the renewal impact factor using the following formula. : ； in, Indicates the first The factors include occupation type, health status, and claims history. Indicates the first The weights of each influencing factor are obtained through training on historical data, and are as follows: The occupation type factor is assigned a value based on the occupational risk level, with a value range of 0-1. The health status factor is assigned a value based on the health assessment results, with a value range of 0-1; The claims record factor is assigned a value based on the claims frequency and amount, with a value range of 0-1. Weight The prediction error is minimized by optimizing using gradient descent.

6. A group insurance renewal data intelligent comparison system, used to implement the group insurance renewal data intelligent comparison method as described in any one of claims 1-5, characterized in that, include: The data acquisition module is used to obtain multi-dimensional data of group insurance participants; The data processing module is used to preprocess the multi-dimensional data and construct labels; The intelligent analysis module is used to analyze renewal behavior using machine learning models; The business application module is used to output analysis results and business strategy recommendations.

7. The intelligent comparison system for group insurance renewal data according to claim 6, characterized in that, The data acquisition module is used for: The basic information, historical renewal records, reasons for non-renewal, occupation type, health status, and claims records of the group's insured individuals are collected. Sensitive information is processed using data anonymization techniques, including retaining the first 6 and last 4 digits of the ID number and anonymizing the name.

8. The intelligent comparison system for group insurance renewal data according to claim 6, characterized in that, The data processing module is specifically used for: The collected data is cleaned, deduplicated, and formatted. Construct a three-level tagging system, including basic attribute tags, risk attribute tags, and behavioral attribute tags; The basic attribute tags include age and gender, the risk attribute tags include occupational risk and claim frequency, and the behavioral attribute tags include renewal history and reasons for non-renewal.

9. The intelligent comparison system for group insurance renewal data according to claim 6, characterized in that, The intelligent analysis module is specifically used for: The renewal trend analysis model was trained using the support vector machine algorithm. Historical renewal data was used as the training set, and the ratio of the training set to the test set was 7:

3. The accuracy of the model was verified, and the threshold was set to 95%. Calculate the renewal impact factor using the formula. Quantify the impact of occupation, health, and claims factors on renewal behavior; Generate renewal prediction results, output the probability of future renewal, and issue risk warnings for low-probability group policies.

10. The intelligent comparison system for group insurance renewal data according to claim 6, characterized in that, The business application module is specifically used for: Output renewal rate analysis reports, attribute tags for those who did not renew their policies, and profiles of new policyholders; Generate business strategy recommendations, including optimizing coverage for younger groups and providing customized renewal plans for high-risk occupations; It supports model iteration and optimization, and regularly updates model parameters and labeling system through an online learning mechanism.

Images