A car insurance pricing model pre-production risk simulation evaluation system and method

Abstract

The present application relates to the technical field of insurance data processing and risk assessment, and specifically discloses a car insurance pricing model pre-production risk simulation and evaluation system and method. The technical scheme of the present application generates a version snapshot and records it in the database when the risk cost configuration is released, compares the current and to-be-online versions field by field before going online and outputs a difference list, calculates the ultimate claim ratio and difference value of the base fee of the two versions based on the preset period of policy data, compares the actual pricing data cost rate change with the pre-production evaluation results after going online to determine the online effect and decompose the influence degree, solves the problems of chaotic management in the online process of the traditional car insurance pricing model version, inconvenient change tracing, frequent configuration omission and misoperation, missing pre-production evaluation, limited manual sampling calculation dimension and long cycle, lack of verification after going online, improves the version management standardization and pre-production evaluation efficiency, and reduces the version iteration management cost and online risk.

Description

Technical Field

[0001] This invention relates to the field of insurance data processing and risk assessment technology, and in particular to a pre-production risk simulation assessment system and method for auto insurance pricing models. Background Technology

[0002] In the property insurance industry, the accuracy of auto insurance pricing models directly impacts an insurance company's underwriting profits and market competitiveness. Risk cost is a core parameter in auto insurance pricing, and its configuration is frequently updated, with each version potentially having a significant impact on pricing results and the overall claims ratio.

[0003] Traditional auto insurance pricing model version deployment processes have significant technical shortcomings. Risk cost configuration versions lack a systematic management mechanism, making it difficult to trace changes between versions and prone to configuration omissions or errors. Before version deployment, there is a lack of effective pre-launch performance evaluation methods, making it impossible to predict the impact of version changes on the overall quoted cost rate, and deployment risks are difficult to control. Existing evaluation methods mostly rely on manual sampling calculations, which have limited evaluation dimensions and long calculation cycles, typically requiring several days, making it difficult to meet the business needs of high-frequency iterations. After version deployment, there is a lack of a closed-loop verification mechanism, making it impossible to quantify the deviation between the actual effect and the pre-launch performance expectations, resulting in significant adjustment risks.

[0004] These issues have led to chaotic management, delayed assessments, and insufficient decision-making basis during the iteration of auto insurance pricing models, thus hindering insurance companies' ability to dynamically optimize pricing strategies. It is necessary to introduce risk simulation assessment methods to improve the standardization of version management and the scientific rigor of deployment decisions. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a pre-production risk simulation assessment system and method for auto insurance pricing models.

[0006] Firstly, the present invention provides a pre-production risk simulation and assessment system for auto insurance pricing models, the technical solution of which is as follows: Configure the version management module to generate a version snapshot and save it to the database each time the risk cost configuration is released; The version difference comparison module is used to obtain the configuration data of the current online version and the version to be launched before the car insurance pricing model goes online, based on the version snapshot recorded in the database, and compare them field by field, and output a structured difference list combined by business segment, institution level, insurance type and new energy identification. The pre-production risk simulation assessment module is used to calculate the base premium ultimate payout ratio and ultimate payout ratio based on the policy data within a preset time period and the configuration data of the version to be launched, according to the combination of dimensions, based on the configuration data of the current online version and the configuration data of the version to be launched, and outputs the base premium ultimate payout ratio and ultimate payout ratio of each version and the difference value between versions. The online effect verification module is used to obtain the actual quotation data after the car insurance pricing model is launched, compare the cost rate change of the actual quotation data with the base premium ultimate loss ratio and the ultimate loss ratio of each version to determine whether the version launch meets expectations, and decompose the impact of the version launch on the quotation cost rate according to the dimensions. The data storage module is used to store configuration version data, policy data, simulation evaluation results, and online verification results.

[0007] The beneficial effects of the pre-production risk simulation assessment system for auto insurance pricing models of the present invention are as follows: The system of this invention generates a version snapshot and records it in the database when the risk cost configuration is released. Before going live, it compares the current and the version to be launched field by field and outputs a list of differences. Based on the policy data of a preset period, it calculates the final loss ratio, final loss ratio and difference value of the base premium for the two versions respectively. After going live, it compares the change of the actual quoted data cost ratio with the pre-production assessment results to determine the effect of going live and decompose the degree of impact. This solves the problems of chaotic management, inconvenient change tracking, frequent configuration omissions and misoperations, lack of pre-production assessment, limited dimensions and long cycle of manual sampling calculation, and lack of verification after going live in the traditional auto insurance pricing model version launch process. It improves the standardization of version management and the efficiency of pre-production assessment, and reduces the cost of version iteration management and the risk of going live.

[0008] Based on the above scheme, the pre-production risk simulation assessment system for auto insurance pricing model of the present invention can be further improved as follows.

[0009] In one alternative approach, the pre-production risk simulation assessment module is specifically used for: A full-scale evaluation method is adopted, using the configuration data of the current online version and the configuration data of the version to be launched to calculate the full policy data one by one, and output the base premium ultimate payout ratio, the ultimate payout ratio and the difference value between the versions for each version.

[0010] The advantages of adopting the above optional methods are as follows: by further adopting the full-scale evaluation method, using the two versions of configuration data to calculate the base premium ultimate loss ratio and ultimate loss ratio for each of the full policies and output the difference value, the evaluation covers all policies, avoids sampling bias, and the results are more comprehensive and accurate.

[0011] In one alternative approach, the pre-production risk simulation assessment module is specifically used for: An incremental evaluation method is adopted. Based on the structured difference list, the combination of dimensions that have changed is selected and the policy data corresponding to the changed dimensions is incrementally calculated. The calculation results of the policy data corresponding to the unchanged dimensions are read from the historical cache data and reused. The results of the incremental calculation are combined with the results read from the historical cache data to obtain the base fee ultimate payout ratio, the ultimate payout ratio, and the difference value between versions for each version.

[0012] The advantages of adopting the above optional method are as follows: by further adopting the incremental evaluation method, the combination of dimensions with configuration changes is selected based on the structured difference list for incremental calculation, the historical cached results of unchanged dimensions are reused, and the base fee final payout ratio, final payout ratio and difference value are output after merging, thereby reducing duplicate calculations and shortening the evaluation cycle.

[0013] In one alternative approach, the structured difference list includes the change type, change fields, and change magnitude for each dimension combination. The version difference comparison module is further configured to: compare the change magnitude with a preset threshold, and classify and label the degree of change for each dimension combination based on the comparison results.

[0014] The advantages of adopting the above optional method are: further recording the change type, change field and change range of each dimension combination in the structured difference list, and comparing the change range with the preset threshold and then classifying and labeling it, making the version differences more intuitive and facilitating the rapid identification of high-risk change points.

[0015] In one alternative approach, the online performance verification module is specifically used for: The cost rate change of the actual quoted data is compared with the base fee ultimate payout rate and the ultimate payout rate of each version to calculate the deviation rate. The deviation rate is the ratio of the absolute value of the difference between the cost rate change and the predicted cost rate to the predicted cost rate. When the deviation rate is not greater than the preset tolerance range, the version launch is determined to be in line with expectations; when the deviation rate is greater than the preset tolerance range, an abnormal warning is triggered.

[0016] The advantages of adopting the above optional method are as follows: further compare the cost rate change of the actual quotation data with the pre-production assessment results and calculate the deviation rate. Combine the preset tolerance range to automatically determine whether the version launch meets expectations. When the deviation rate exceeds the tolerance, trigger an abnormal warning to achieve quantitative verification of the launch effect.

[0017] In an alternative approach, the online performance verification module is further used for: When a version launch is deemed not to meet expectations, the cost rate change of each dimension combination before and after the version launch is determined according to the aforementioned dimension combination. The contribution of each dimension combination is determined based on the proportion of the cost rate change of each dimension combination to the overall cost rate change, and a contribution ranking and abnormal fluctuation warning are output.

[0018] The beneficial effects of adopting the above optional methods are as follows: when the version launch does not meet expectations, the cost rate change before and after the launch can be determined by the combination of dimensions, the contribution can be determined according to the proportion of the cost rate change of each dimension combination to the overall change, and the contribution ranking and abnormal fluctuation warning can be output to help locate the root cause.

[0019] In one alternative approach, the configuration data includes risk cost coefficients, base premiums, and rate adjustment factors.

[0020] The beneficial effects of adopting the above optional methods are: further incorporating risk cost coefficients, basic premiums, and rate adjustment factors into the scope of configuration data, enabling version snapshots to fully record core pricing parameters, ensuring the integrity and consistency of configuration version data, and providing a comprehensive data foundation for difference comparison and risk simulation.

[0021] Secondly, this invention provides a method for simulating and assessing the pre-production risk of an auto insurance pricing model. The technical solution of this method is as follows: Generate a version snapshot and save it to the database each time the risk cost configuration is released; Based on the version snapshot recorded in the database, before the car insurance pricing model goes online, the configuration data of the current online version and the version to be launched are obtained and compared field by field, and a structured difference list is output with business segment, institution level, insurance type and new energy identification as dimensions. Using policy data within a preset time period as the evaluation benchmark, and based on the aforementioned dimension combination, the base premium ultimate payout ratio and ultimate payout ratio are calculated respectively based on the configuration data of the current online version and the configuration data of the version to be launched, and the base premium ultimate payout ratio and ultimate payout ratio of each version, as well as the difference value between versions, are output. Obtain the actual quote data after the car insurance pricing model is launched, compare the cost rate change of the actual quote data with the base premium ultimate loss ratio and the ultimate loss ratio of each version to determine whether the version launch meets expectations, and decompose the impact of the version launch on the quote cost rate according to the dimensions. Store configuration version data, policy data, simulation evaluation results, and online verification results.

[0022] The beneficial effects of the pre-production risk simulation assessment method for a car insurance pricing model of the present invention are as follows: The method of this invention generates a version snapshot and records it in the database when the risk cost configuration is released. Before going live, it compares the current and the version to be launched field by field and outputs a list of differences. Based on the policy data of a preset period, it calculates the final loss ratio, final loss ratio and difference value of the base premium for the two versions respectively. After going live, it compares the change of the actual quoted data cost ratio with the pre-production assessment results to determine the effect of going live and decompose the degree of impact. This method solves the problems of chaotic management, inconvenient change tracking, frequent configuration omissions and misoperations, lack of pre-production assessment, limited dimensions and long cycle of manual sampling calculation, and lack of verification after going live in the traditional auto insurance pricing model version launch process. It improves the standardization of version management and the efficiency of pre-production assessment, and reduces the cost of version iteration management and the risk of going live.

[0023] Thirdly, the technical solution of an electronic device according to the present invention is as follows: It includes a memory, a processor, and a program stored in the memory and running on the processor, wherein the processor executes the program to implement the steps of the pre-production risk simulation assessment method for the vehicle insurance pricing model of the present invention.

[0024] Fourthly, the technical solution of a computer-readable storage medium provided by the present invention is as follows: The computer-readable storage medium stores instructions that, when read, cause the computer-readable storage medium to perform the steps of the pre-production risk simulation assessment method for the vehicle insurance pricing model of the present invention.

[0025] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other objects, features and advantages of the present invention more obvious and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0026] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a schematic diagram of the structure of a first embodiment of a pre-production risk simulation and assessment system for a vehicle insurance pricing model according to the present invention; Figure 2 This is a schematic diagram of the structure of a second embodiment of a pre-production risk simulation and assessment system for a vehicle insurance pricing model according to the present invention; Figure 3 This is a schematic diagram of the overall process; Figure 4 This is a flowchart illustrating the process of version difference comparison and full-scale simulation evaluation. Figure 5 This is a flowchart illustrating the process of incremental simulation evaluation and online effect verification. Figure 6 This is a flowchart illustrating an embodiment of a pre-production risk simulation assessment method for a vehicle insurance pricing model according to the present invention. Figure 7 This is a schematic diagram of an embodiment of an electronic device according to the present invention. Detailed Implementation

[0027] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0028] Figure 1 This diagram illustrates the structure of a first embodiment of a pre-production risk simulation assessment system 100 for auto insurance pricing models provided by the present invention. Figure 1 As shown, the pre-production risk simulation assessment system 100 for the auto insurance pricing model includes: Configure version management module 101 to generate version snapshots and save them to the database each time the risk cost configuration is released.

[0029] Among them, risk cost configuration refers to the set of risk cost-related parameters set for different combinations of dimensions in the auto insurance pricing model, including risk cost coefficient, basic premium and rate adjustment factor; for example, the pricing team set the risk cost coefficient to 1.25, the basic premium to 3200 yuan and the rate adjustment factor to 0.95 for the combination of L1 provincial branch, C1 insurance type and new energy vehicle identification "yes" in the P1 business segment.

[0030] The version snapshot refers to a complete record automatically generated by the system each time the risk cost configuration is released, containing the version number, release time, publisher, change description, and a full snapshot of the configuration. For example, if the pricing team releases version V2.3.1 at 10:00 AM on July 1, 2026, the system automatically generates a snapshot of this version, recording the version number as V2.3.1, the release time as July 1, 2026 at 10:00 AM, the publisher as Pricing Officer A, the change description as adjusting the risk cost coefficient for new energy vehicle insurance in the P1 business segment, and the full snapshot of the configuration containing all parameters for all dimensions of this version. The database persistence refers to the operation of persistently storing the version snapshot in a relational database, supporting retrieval and tracing by version number, time range, publisher, and other dimensions. For example, the snapshot of version V2.3.1 is stored in the configuration version management database, and the complete configuration content of this version can be queried subsequently using the version number V2.3.1 or the release time 2026-07-01.

[0031] Specifically, a version snapshot is generated and persisted to the database each time the risk cost configuration is released. After the risk cost configuration is approved, a version number is automatically generated and a version snapshot is created. The version snapshot records the version number, release time, publisher, change description, and a full snapshot of the configuration. The version snapshot is persistently stored in a relational database, supporting multi-dimensional retrieval and tracing by version number, time range, and publisher.

[0032] The version difference comparison module 102 is used to obtain the configuration data of the current online version and the version to be launched before the car insurance pricing model is launched, based on the version snapshot recorded in the database, and to compare them field by field, and output a structured difference list combined by business segment, institution level, insurance type and new energy identification.

[0033] Among them, the auto insurance pricing model refers to a mathematical model that calculates auto insurance premiums and loss ratios based on risk cost configuration, policy data and other pricing factors. For example, the auto insurance pricing model used by the insurance company receives new energy vehicle policy data from the L1 provincial branch of the P1 business segment, and calculates the base premium ultimate loss ratio and ultimate loss ratio of the policy by combining the risk cost configuration of version V2.3.1.

[0034] The "current online version" refers to the risk cost configuration version used by the auto insurance pricing model currently providing quoting services in the production environment. For example, before version V2.3.1 was launched, version V2.3.0 was running in the production environment, and V2.3.0 is the current online version. The "version to be launched" refers to the risk cost configuration version of the auto insurance pricing model that has undergone configuration changes and is ready to be released to the production environment to replace the current online version. For example, the pricing team is preparing to release version V2.3.1 of the risk cost configuration to replace version V2.3.0, and V2.3.1 is the version to be launched.

[0035] The configuration data refers to the complete parameter information recorded in the risk cost configuration, including the risk cost coefficient, basic premium and rate adjustment factor under each dimension combination; for example, the configuration data of version V2.3.1 includes a risk cost coefficient of 1.25, a basic premium of 3200 yuan and a rate adjustment factor of 0.95 for the new energy vehicle dimension of the P1 business segment L1 provincial branch.

[0036] Here, "business segment" refers to the operational management unit of an insurance company, divided according to business attributes, used to differentiate pricing strategies for different business lines or regional markets. For example, an insurance company might divide its auto insurance business into P1, P2, P3, and P4 business segments, with P1 encompassing regions A1, A2, A3, and A4. "Organizational level" refers to the internal organizational structure of an insurance company, used to differentiate organizational modules at different administrative or management levels. For example, an insurance company's internal structure might include four levels: head office, provincial branch, municipal branch, and sub-branch. The A1 regional branch of the P1 business segment belongs to the provincial branch level and is denoted as L1. "Insurance type" refers to the classification of different insurance liability categories within auto insurance products, used to differentiate product types with different coverage and risk attributes. For example, auto insurance products might include various insurance types such as compulsory traffic accident liability insurance, motor vehicle damage insurance, third-party liability insurance, passenger liability insurance, and vehicle theft insurance, with motor vehicle damage insurance denoted as C1 and third-party liability insurance as C2. The new energy vehicle label refers to a mark used to distinguish whether an insured vehicle is a new energy vehicle. For example, if an insured vehicle is a pure electric vehicle, its new energy vehicle label is "yes"; if an insured vehicle is a gasoline vehicle, its new energy vehicle label is "no".

[0037] The dimension combination refers to a primary key combination composed of four dimensions: business segment, institutional level, insurance type, and new energy identifier. This combination uniquely identifies the business scope to which a pricing configuration applies. For example, P1 business segment, L1 provincial branch, C1 insurance type, and new energy identifier being "yes" constitute a dimension combination, which corresponds to an independent set of risk cost configuration parameters. The structured difference list refers to the difference data set output by the version difference comparison module 102, organized with the dimension combination as the primary key. It includes the change type, change fields, and change magnitude information for each dimension combination. For example, after comparing version V2.3.1 with version V2.3.0, the structured difference list shows that the change type for the new energy dimension combination of the C1 insurance type in the L1 provincial branch of the P1 business segment is "modified," the change field is the risk cost coefficient, the value before the change is 1.18, the value after the change is 1.25, and the change magnitude is an absolute difference of 0.07 and a relative difference of 5.93%.

[0038] Specifically, based on the version snapshot of the data entry records, configuration data of the current online version and the version to be launched are obtained before the auto insurance pricing model goes live, and a field-by-field comparison is performed. A structured list of differences is output, with business segments, institutional levels, insurance types, and new energy vehicle identifiers as dimensions. Using the dimension combination as the comparison key, the configuration parameters of the two versions are compared row by row and field by field. The change type, changed fields, and change magnitude for each dimension combination are marked. Change types include additions, deletions, and modifications; changed fields include risk cost coefficients, basic premiums, and rate adjustment factors; and change magnitudes are expressed as absolute and relative differences.

[0039] The pre-production risk simulation assessment module 103 is used to calculate the base premium ultimate payout ratio and ultimate payout ratio based on the policy data within a preset time period and the configuration data of the current online version and the version to be launched, respectively, according to the dimension combination, and output the base premium ultimate payout ratio and ultimate payout ratio of each version and the difference value between versions.

[0040] The preset time period refers to the length of the historical data time window pre-set by the system for risk assessment. For example, the system defaults to using policy data from the 365 days prior to the assessment start date as the assessment benchmark. Policy data refers to the complete information record of all policies underwritten within the preset time period, including fields such as policy number, policyholder information, vehicle information, insurance type information, premium, and claims records. For example, all policies underwritten between July 1, 2025 and July 1, 2026 are selected as assessment data, totaling 1.2 million policies. Each policy record includes information such as whether the vehicle is a new energy vehicle, the insurance type is C1, the premium amount, and the historical claims amount. The evaluation benchmark refers to the data basis on which the pre-production risk simulation evaluation module 103 calculates indicators, namely, all policy data within the preset time period; for example, using 1.2 million policies from July 1, 2025 to July 1, 2026 as the evaluation benchmark, the configuration data of versions V2.3.0 and V2.3.1 are loaded respectively to calculate the base premium ultimate loss ratio and ultimate loss ratio of each version.

[0041] The base premium ultimate loss ratio refers to the ultimate loss ratio calculated based on the base premium under a given risk cost configuration. It is the ratio of the policy's expected claims cost to its base premium income. For example, using 1.2 million policies in the assessment benchmark as data, under the V2.3.1 configuration, the total base premium income is 384 million yuan, the total expected claims cost is 288 million yuan, and the base premium ultimate loss ratio is 75%. The ultimate loss ratio refers to the final loss ratio after considering the rate adjustment factor under a given risk cost configuration. It is the ratio of the policy's expected claims cost to the actual premium income collected. For example, using 1.2 million policies in the assessment benchmark as data, under the V2.3.1 configuration, the total actual premium income collected is 404 million yuan, the total expected claims cost is 288 million yuan, and the ultimate loss ratio is 71.3%.

[0042] The inter-version difference refers to the numerical difference between the version to be launched and the current online version on the same evaluation indicator under the same evaluation benchmark, including both absolute and relative differences. For example, with 1.2 million policies as the evaluation benchmark, the ultimate payout ratio of version V2.3.1 is 71.3%, and the ultimate payout ratio of version V2.3.0 is 69.5%. The absolute difference between the versions is 1.8 percentage points, and the relative difference is 2.59%.

[0043] Specifically, using policy data within a preset time period as the evaluation benchmark, and based on dimensional combinations, the base premium ultimate payout ratio and ultimate payout ratio are calculated using both the current online version's configuration data and the configuration data of the version to be launched. The base premium ultimate payout ratio and ultimate payout ratio for each version, along with inter-version differences, are then output. When using a full-scale evaluation method, the configuration data of both the current online version and the version to be launched are used to calculate the base premium ultimate payout ratio, ultimate payout ratio, and inter-version differences for each version. When using an incremental evaluation method, the dimensional combinations whose configurations have changed are selected based on a structured difference list. Incremental calculations are performed on the policy data corresponding to the changed dimensions. The calculation results for policy data corresponding to unchanged dimensions are read from historical cached data and reused. The results of the incremental calculations are merged with the results read from historical cached data to obtain the base premium ultimate payout ratio, ultimate payout ratio, and inter-version differences for each version.

[0044] The online effect verification module 104 is used to obtain the actual quotation data after the car insurance pricing model is launched, compare the cost rate change of the actual quotation data with the base premium ultimate loss ratio and the ultimate loss ratio of each version to determine whether the version launch meets expectations, and decompose the impact of the version launch on the quotation cost rate according to the dimensions.

[0045] The actual quote data refers to the premium and cost rate data corresponding to the quotes provided to policyholders in the real production environment after the auto insurance pricing model is launched. For example, after version V2.3.1 was launched, the production system collected 4,523 quotes for new energy vehicles under the C1 insurance category of the L1 provincial branch of the P1 business segment from July 1, 2026 to July 15, 2026, with an average actual cost rate of 71.8%.

[0046] The degree of impact refers to the effect of the version launch on the overall cost rate of quotes. After decomposition by dimension combination, the contribution of each dimension combination reflects the contribution share of that combination to the change in the overall cost rate. For example, after the launch of version V2.3.1, the overall cost rate of quotes increased from 68.2% to 69.0%. Among them, the cost rate of the new energy dimension of the C1 insurance product of the L1 provincial branch of the P1 business segment increased from 70.5% to 73.2%. The contribution of this dimension combination is (73.2%-70.5%) / (69.0%-68.2%)=3.375, that is, this dimension explains 337.5% of the increase in the overall cost rate. It is the dimension combination with the greatest impact and needs to be investigated in detail.

[0047] Specifically, the system acquires actual quote data after the auto insurance pricing model is launched, compares the cost rate changes of the actual quote data with the base premium final loss ratio and final loss ratio of each version, and calculates the deviation rate. The deviation rate is the ratio of the absolute value of the difference between the cost rate change and the predicted cost rate to the predicted cost rate. When the deviation rate is not greater than the preset tolerance range, the version launch is considered to meet expectations; when the deviation rate is greater than the preset tolerance range, an anomaly warning is triggered. The impact of the version launch on the quoted cost rate is decomposed by dimension combination; when the version launch is deemed not to meet expectations, the cost rate change of each dimension combination before and after the version launch is determined separately, and the contribution of each dimension combination is determined based on the proportion of the cost rate change of each dimension combination to the overall cost rate change, and the contribution ranking and anomaly fluctuation warning are output.

[0048] The data storage module 105 is used to store configuration version data, policy data, simulation evaluation results, and online verification results.

[0049] The simulation evaluation results refer to all calculation results output by the pre-production risk simulation evaluation module 103, including the base cost ultimate payout ratio, ultimate payout ratio, and inter-version differences for each version. For example, the simulation evaluation results include the base cost ultimate payout ratio of 73.2% and the ultimate payout ratio of 69.5% for version V2.3.0, the base cost ultimate payout ratio of 75% and the ultimate payout ratio of 71.3% for version V2.3.1, and the absolute difference of 1.8 percentage points and the relative difference of 2.46% between the two versions in terms of the base cost ultimate payout ratio, and the absolute difference of 1.8 percentage points and the relative difference of 2.59% in terms of the ultimate payout ratio.

[0050] The online verification results refer to all judgment conclusions and analysis data output by the online effect verification module 104, including the deviation rate calculation results, the judgment conclusions on whether it meets expectations, the contribution ranking of each dimension combination, and the abnormal fluctuation warning information. For example, the online verification results show that the deviation rate of version V2.3.1 is (71.8%-71.3%) / 71.3%=0.7%, which is less than the preset tolerance range of ±5%, and the version is judged to meet expectations. Among the various dimension combinations, the one with the highest contribution is the new energy dimension of the C1 insurance product of the L1 provincial branch of the P1 business segment, and no abnormal fluctuation warning was triggered.

[0051] Specifically, the system stores configuration version data, policy data, simulation evaluation results, and online verification results. Configuration version data is stored in a relational database, policy data is stored in a data warehouse, simulation evaluation results and online verification results are stored in an analytical database, and historical calculation cache is stored in an in-memory database.

[0052] The technical solution of this embodiment generates a version snapshot and records it in the database when the risk cost configuration is released. Before going live, it compares the current and the version to be launched field by field and outputs a list of differences. Based on the policy data of a preset period, it calculates the final loss ratio, final loss ratio and difference value of the base premium for the two versions respectively. After going live, it compares the change of the actual quoted data cost ratio with the pre-production assessment results to determine the effect of going live and decompose the degree of impact. This solves the problems of chaotic management, inconvenient change tracking, frequent configuration omissions and misoperations, lack of pre-production assessment, limited dimensions and long cycle of manual sampling calculation, and lack of verification after going live in the traditional auto insurance pricing model version launch process. It improves the standardization of version management and the efficiency of pre-production assessment, and reduces the version iteration management cost and launch risk.

[0053] In an alternative embodiment, the pre-production risk simulation assessment module 103 is specifically used for: A full-scale evaluation method is adopted, using the configuration data of the current online version and the configuration data of the version to be launched to calculate the full policy data one by one, and output the base premium ultimate payout ratio, the ultimate payout ratio and the difference value between the versions for each version.

[0054] The full-scale evaluation method refers to loading the configuration data of the current online version and the version to be launched separately for all policy data, calculating each policy individually, and outputting the complete evaluation indicators of the two versions and the evaluation mode of the differences between the versions. For example, the system loads the configuration data of versions V2.3.0 and V2.3.1, calculates the base premium ultimate payout ratio and ultimate payout ratio for each of the 1.2 million policies according to the parameters of the two versions, and completes all calculations in 3.5 hours, outputting the indicators and differences between the two versions.

[0055] Specifically, a full-scale evaluation approach is adopted, which is suitable for scenarios involving initial evaluation or large-scale configuration changes. Using configuration data from both the current online version and the version to be deployed, the full set of policy data is calculated item by item. Under each dimension combination, the premium, base premium ultimate payout ratio, and ultimate payout ratio for both versions are summarized, outputting the base premium ultimate payout ratio, ultimate payout ratio, and the difference between the versions for both versions.

[0056] Among the above optional methods, a full-scale evaluation method is further adopted. The two versions of configuration data are used to calculate the base premium ultimate loss ratio and ultimate loss ratio for each of the full policies and output the difference value. The evaluation covers all policies, avoids sampling bias, and the results are more comprehensive and accurate.

[0057] In an alternative embodiment, the pre-production risk simulation assessment module 103 is specifically used for: An incremental evaluation method is adopted. Based on the structured difference list, the combination of dimensions that have changed is selected and the policy data corresponding to the changed dimensions is incrementally calculated. The calculation results of the policy data corresponding to the unchanged dimensions are read from the historical cache data and reused.

[0058] The incremental assessment method refers to the following: based on the structured difference list, the system filters out the dimension combinations that have changed, calculates only the policy data corresponding to the changed dimensions, and directly reuses historical cached data for the unchanged dimensions. Then, the two results are merged to obtain the complete assessment indicators and the assessment mode of version difference values. For example, the structured difference list shows that only the risk cost coefficient of the dimension combination of new energy insurance for the C1 insurance product of the L1 provincial branch of the P1 business segment has changed. The system filters out 85,000 policies corresponding to this dimension for incremental calculation, and the calculation results of the remaining 1,115,000 policies are read from the historical cached data. The two results are merged to obtain the complete assessment indicators, and the entire assessment is completed in 0.8 hours.

[0059] Among them, a "changed dimension" refers to a dimension combination marked in the structured difference list as having at least one changed configuration field. For example, the risk cost coefficient of the new energy dimension combination for the C1 insurance product of the L1 provincial branch in the P1 business segment changed from 1.18 to 1.25, and this dimension combination is marked as a changed dimension. All configuration fields of the new energy dimension combination for the C1 insurance product of the L1 provincial branch in the P2 business segment remain unchanged, and this dimension combination is not a changed dimension. An "unchanged dimension" refers to a dimension combination in the structured difference list where all configuration fields have not changed. For example, the risk cost coefficient, basic premium, and rate adjustment factor of the new energy dimension combination for the C1 insurance product of the L1 provincial branch in the P2 business segment are exactly the same in both versions V2.3.0 and V2.3.1, and this dimension combination is classified as an unchanged dimension.

[0060] Historical cached data refers to the base premium ultimate payout ratio and ultimate payout ratio calculation results of each dimension combination under each version, generated by the system in historical assessment calculations and stored in the in-memory database. These results are used for incremental assessments where unchanged dimensions can be directly reused. For example, in the full assessment before the launch of version V2.3.0, the system calculated the ultimate payout ratio of the new energy dimension combination of the C1 insurance product of the L1 provincial branch of the P2 business segment to be 68.5% under version V2.3.0, and this result was stored in the cache of the in-memory database. In the incremental assessment of version V2.3.1, since this dimension combination has not changed, the system directly reads the 68.5% result from the cache for reuse.

[0061] Specifically, an incremental evaluation method is adopted, which is suitable for regular version iteration and update scenarios. Based on a structured difference list, dimensional combinations with changed configurations are selected. Changed dimensions are those marked as having at least one changed configuration field, while unchanged dimensions are those where none of the configuration fields have changed. Incremental calculations are performed on the policy data corresponding to changed dimensions, while the calculation results for the policy data corresponding to unchanged dimensions are read from historical cached data and reused.

[0062] The results of the incremental calculation are combined with the results read from the historical cache data to obtain the base fee ultimate payout ratio, the ultimate payout ratio, and the difference value between versions for each version.

[0063] Specifically, the results of incremental calculations are merged with the results read from historical cached data to obtain the base fee ultimate payout ratio and ultimate payout ratio for all dimensions in each version; based on the base fee ultimate payout ratio and ultimate payout ratio for each version under the same dimension combination, the difference value between versions is calculated.

[0064] Among the above-mentioned optional methods, an incremental evaluation method is further adopted. Based on the structured difference list, the combination of dimensions with configuration changes is selected for incremental calculation. The unchanged dimensions reuse historical cache results. After merging, the base fee final payout ratio, final payout ratio and difference value are output, which reduces redundant calculation and shortens the evaluation cycle.

[0065] In one alternative approach, the structured difference list includes the change type, change fields, and change magnitude for each dimension combination, and the version difference comparison module 102 is further configured to: compare the change magnitude with a preset threshold, and classify and label the degree of change for each dimension combination according to the comparison results.

[0066] The "Change Type" refers to the category of changes in the configuration data within a dimension combination relative to the current online version, including three types: addition, deletion, and modification. For example, in version V2.3.0, the new energy dimension combination for the C1 insurance product of the L1 provincial branch in the P1 business segment already existed with a risk cost coefficient of 1.18. In version V2.3.1, the risk cost coefficient changed to 1.25, and the change type is "modification." Similarly, in the P3 business segment, a new branch office was added. The corresponding dimension combination for this branch office did not exist in version V2.3.0 and was added for the first time in version V2.3.1, and the change type is "addition." The "Change Field" refers to the name of the specific configuration parameter field that changed within the dimension combination. For example, in the new energy dimension combination for the C1 insurance product of the L1 provincial branch in the P1 business segment, the risk cost coefficient changed from 1.18 to 1.25, while the basic premium and rate adjustment factor remained unchanged; the changed field is the risk cost coefficient. The magnitude of change refers to the numerical difference between the current online version and the version to be launched for the changed field. It includes two forms: absolute difference and relative difference. The absolute difference is the difference between the value after the change and the value before the change, and the relative difference is the ratio of the absolute difference to the value before the change. For example, the risk cost coefficient of the new energy dimension combination of insurance product C1 in the P1 business segment of the L1 provincial branch changes from 1.18 to 1.25. The absolute difference is 1.25-1.18=0.07, and the relative difference is 0.07 / 1.18=5.93%.

[0067] The preset threshold refers to a pre-set numerical boundary used to determine the degree of change. For example, the system sets the grading boundary of the absolute difference as follows: the slight level corresponds to an absolute difference of less than 0.02 or a relative difference of less than 2%; the significant level corresponds to an absolute difference of greater than or equal to 0.02 and less than 0.05 or a relative difference of greater than or equal to 2% and less than 5%; and the major level corresponds to an absolute difference of greater than or equal to 0.05 or a relative difference of greater than or equal to 5%.

[0068] The degree of change refers to the classification of configuration changes for each dimension combination based on the magnitude of the change compared to a preset threshold. For example, the absolute difference in risk cost coefficients for the new energy dimension combination of the C1 insurance product in the L1 provincial branch of the P1 business segment is 0.07, and the relative difference is 5.93%. After comparison with the preset threshold, the degree of change for this dimension combination is classified as significant. The hierarchical labeling refers to adding corresponding labels to each dimension combination in the structured difference list according to the degree of change. For example, the new energy dimension combination of the C1 insurance product in the L1 provincial branch of the P1 business segment is labeled as significant, the new energy dimension combination of the C1 insurance product in the L1 provincial branch of the P2 business segment is labeled as minor, and the new energy dimension combination of the C1 insurance product in the L1 provincial branch of the P3 business segment is labeled as substantial.

[0069] In the above optional methods, the change type, change field and change magnitude of each dimension combination are further recorded in the structured difference list, and the change magnitude is compared with the preset threshold and then graded and labeled to make the version differences more intuitive and facilitate the rapid identification of high-risk change points.

[0070] In one alternative approach, the online performance verification module 104 is specifically used for: The cost rate change of the actual quoted data is compared with the base fee ultimate payout rate and the ultimate payout rate of each version to calculate the deviation rate. The deviation rate is the ratio of the absolute value of the difference between the cost rate change and the predicted cost rate to the predicted cost rate.

[0071] The deviation rate refers to the quantitative measure of the degree of deviation between the cost rate change of the actual quotation data after the version is launched and the predicted cost rate of each version in the pre-production simulation assessment before the launch. The value is the ratio of the absolute value of the difference between the cost rate change and the predicted cost rate to the predicted cost rate. For example, before the launch of version V2.3.1, the predicted final payout rate in the pre-production simulation assessment was 71.3%. After the version was launched, the actual cost rate was calculated to be 71.8% based on the actual quotation data collected from July 1, 2026 to July 15, 2026, and the deviation rate was (71.8%-71.3%) / 71.3%=0.7%. The predicted cost rate refers to the base premium ultimate payout ratio and ultimate payout ratio of each version calculated by the pre-production risk simulation assessment module 103 based on the configuration data of the version to be launched and the policy data in the assessment benchmark, which serves as a comparison benchmark for verifying the launch effect; for example, the pre-production simulation assessment of version V2.3.1 predicts an ultimate payout ratio of 71.3%, which is used as the predicted cost rate after version V2.3.1 is launched.

[0072] When the deviation rate is not greater than the preset tolerance range, the version launch is determined to be in line with expectations; when the deviation rate is greater than the preset tolerance range, an abnormal warning is triggered.

[0073] The preset tolerance range refers to the pre-defined allowable range of deviation rate used to determine whether the version launch meets expectations. For example, the system sets the deviation rate to be no greater than ±5% to determine that the version launch meets expectations, and ±5% is the preset tolerance range. The deviation rate of version V2.3.1 is 0.7%, which is less than 5%, so it is determined to meet expectations.

[0074] In the above-mentioned optional methods, the cost rate change of the actual quotation data is further compared with the pre-production assessment results and the deviation rate is calculated. Combined with the preset tolerance range, it is automatically determined whether the version launch meets the expectations. When the deviation rate exceeds the tolerance, an abnormal warning is triggered to achieve quantitative verification of the launch effect.

[0075] In an alternative embodiment, the online performance verification module 104 is further configured to: When a version launch is deemed not to meet expectations, the cost rate change of each dimension combination before and after the version launch is determined according to the aforementioned dimension combination. The contribution of each dimension combination is determined based on the proportion of the cost rate change of each dimension combination to the overall cost rate change, and a contribution ranking and abnormal fluctuation warning are output.

[0076] The change in cost rate refers to the difference in the quoted cost rate before and after the version launch for the same dimension combination. For example, the actual cost rate of the new energy dimension combination of insurance product C1 in the L1 provincial branch of the P1 business segment was 70.5% after the launch of version V2.3.0, and the actual cost rate was 73.2% after the launch of version V2.3.1. The change in cost rate for this dimension combination is 73.2% - 70.5% = 2.7 percentage points. The change in overall cost rate refers to the change in the overall quoted cost rate before and after the version launch, after summing all dimension combinations. For example, the overall quoted cost rate was 68.2% before the launch of version V2.3.1, and the overall quoted cost rate was 69.0% after the launch. The change in overall cost rate is 69.0% - 68.2% = 0.8 percentage points.

[0077] The contribution ranking refers to the result of sorting the cost rate changes of each dimension combination in descending order of their proportion to the overall cost rate change. For example, after calculating the contribution of each dimension combination, the ranking is as follows: P1 business segment L1 provincial branch C1 insurance new energy dimension (contribution 3.375), P1 business segment L1 provincial branch C2 insurance new energy dimension (contribution 0.625), P2 business segment L1 provincial branch C1 insurance new energy dimension (contribution -1.500). The contribution ranking reflects the magnitude of the contribution of each dimension combination to the overall cost rate change. A positive value indicates that the cost rate is increased, and a negative value indicates that the cost rate is decreased. Abnormal fluctuation warning refers to the risk warning information automatically output when the online effect verification module 104 identifies that the deviation rate of a certain dimension combination exceeds the preset tolerance range, or the absolute value of the contribution of the dimension combination deviates significantly from the normal range. For example, if the contribution of the new energy dimension combination of the C1 insurance product of the L1 provincial branch of the P1 business segment is 3.375, which is significantly higher than other dimension combinations and the absolute value exceeds the experience threshold of 2.0, an abnormal fluctuation warning will be automatically output, prompting that the dimension combination needs to be investigated.

[0078] Specifically, when a version launch is deemed not to meet expectations, the cost rate change of each dimension combination before and after the version launch is determined separately. The proportion of the cost rate change of each dimension combination to the overall cost rate change is calculated, and the proportion is determined as the contribution of each dimension combination. The dimension combinations are sorted from largest to smallest contribution and the contribution ranking is output. An abnormal fluctuation warning is output for dimension combinations whose absolute contribution value deviates significantly from the normal range.

[0079] In the above-mentioned optional methods, if the version launch does not meet expectations, the cost rate change before and after the launch is determined by the combination of dimensions. The contribution is determined according to the proportion of the cost rate change of each dimension combination to the overall change. The contribution ranking and abnormal fluctuation warning are output to help locate the root cause.

[0080] In one alternative approach, the configuration data includes risk cost coefficients, base premiums, and rate adjustment factors.

[0081] The risk cost coefficient refers to the core multiplier parameter used in the auto insurance pricing model to calculate risk costs, reflecting the risk level under a specific combination of dimensions. For example, the risk cost coefficient for the new energy vehicle dimension combination of the C1 insurance product in the L1 provincial branch of the P1 business segment has been adjusted from 1.18 to 1.25. This coefficient is used to calculate the risk cost of each policy under this dimension combination. The basic premium refers to the basic premium amount in the auto insurance pricing model before the application of the rate adjustment factor, serving as the benchmark value for premium calculation. For example, the basic premium for the new energy vehicle product of the C1 insurance product in the L1 provincial branch of the P1 business segment is 3200 yuan, meaning that without considering the rate adjustment factor, the basic premium for a standard policy under this dimension combination is 3200 yuan. The rate adjustment factor refers to the multiplier factor used in the auto insurance pricing model to adjust the basic premium, reflecting the rate discount or premium increase level under a specific combination of dimensions. For example, the rate adjustment factor for new energy vehicles under the C1 insurance product of the L1 provincial branch of the P1 business segment is 0.95, and the actual premium income is the basic premium divided by the rate adjustment factor, i.e., 3200 / 0.95=3368.42 yuan.

[0082] Among the above-mentioned optional methods, risk cost coefficients, basic premiums, and rate adjustment factors are further included in the configuration data scope, so that the version snapshot fully records the core pricing parameters, ensuring the integrity and consistency of the configuration version data, and providing a comprehensive data foundation for difference comparison and risk simulation.

[0083] Figure 2 A schematic diagram of the second embodiment of the pre-production risk simulation assessment system for auto insurance pricing models is shown. Figure 2 As shown, the pre-production risk simulation assessment system for the auto insurance pricing model consists of a configuration version management module 101, a version difference comparison module 102, a pre-production risk simulation assessment module 103, a production effect verification module 104, and a data storage module 105.

[0084] The configuration version management module 101 is responsible for the full lifecycle version management of risk cost configuration and is the foundational layer of the system. Risk cost configuration is a set of risk cost-related parameters set for different combinations of dimensions, including risk cost coefficients, basic premiums, and rate adjustment factors. Each time a risk cost configuration version is released, this module automatically generates a version snapshot. The version snapshot records the version number, release time, publisher, change description, full configuration snapshot, and approval status. The configuration version management module 101 stores the version snapshot in a relational database, supporting multi-dimensional retrieval and traceability by version number, time range, publisher, and other dimensions. This module also provides version comparison and version rollback functions, making configuration changes fully traceable and reversible. When the deviation rate output by the online effect verification module 104 exceeds the preset tolerance range, the configuration version management module 101 responds to the version rollback command, rolling back the configuration of the auto insurance pricing model to the configuration data of the current online version. The configuration data of the current online version is read from the version snapshot stored in the relational database.

[0085] The version difference comparison module 102 is automatically executed before each deployment of the auto insurance pricing model. The auto insurance pricing model is a mathematical model that calculates auto insurance premiums and loss ratios based on risk cost configuration, policy data, and other pricing factors. This module retrieves the configuration data of the current online version and the version to be deployed based on the version snapshot recorded in the database. The current online version is the risk cost configuration version currently used by the auto insurance pricing model providing quoting services in the production environment. The version to be deployed is the risk cost configuration version that has undergone configuration changes and is ready to be released to the production environment to replace the current online version. The configuration data contains complete parameter information recorded in the risk cost configuration, including risk cost coefficients, basic premiums, and rate adjustment factors under various dimension combinations.

[0086] Version difference comparison module 102 uses dimension combinations as the comparison primary key, comparing the configuration parameters of two versions row by row and field by field. A dimension combination consists of four dimensions: business segment, organizational level, insurance type, and new energy vehicle identifier, uniquely identifying the business scope applicable to a set of pricing configurations. A business segment is an operational management unit divided by the insurance company according to business attributes. An organizational level is the internal organizational structure of the insurance company, including levels such as head office, provincial branch, city branch, and sub-branch. Insurance type is the classification of different insurance liability categories in auto insurance products. The new energy vehicle identifier is used to distinguish whether the insured vehicle is a new energy vehicle. Version difference comparison module 102 outputs a structured difference list organized with dimension combinations as the primary key, containing the change type, change fields, and change magnitude for each dimension combination. Change types include addition, deletion, and modification. Change fields include pricing-related parameters such as risk cost coefficient, basic premium, and rate adjustment factor. Change magnitude includes absolute difference and relative difference; the absolute difference is the difference between the value after the change and the value before the change, and the relative difference is the ratio of the absolute difference to the value before the change. This module also compares the magnitude of changes with preset thresholds, and classifies the degree of change for each dimension combination according to the comparison results. The classification levels include minor, significant, and major. The comparison dimensions can be expanded to vehicle series, channels, and age groups as needed. The structured difference list is output in JSON and Excel formats.

[0087] The pre-production risk simulation assessment module 103 is the core calculation engine of the system. This module uses policy data within a preset time period as the assessment benchmark. The preset time period is the length of the historical data time window used for risk assessment, which is set in advance by the system. By default, it uses policy data within one year prior to the assessment start date as the assessment benchmark. Policy data is a complete information record of all policies underwritten within the preset time period, including fields such as policy number, policyholder information, vehicle information, insurance type information, premium, and claims records. The assessment benchmark is the data foundation upon which the pre-production risk simulation assessment module 103 calculates indicators, namely, all policy data within the preset time period. Based on the combination of dimensions, this module calculates the base premium ultimate loss ratio and ultimate loss ratio for each version, based on the configuration data of the current online version and the configuration data of the version to be launched. The base premium ultimate loss ratio is the ultimate loss ratio calculated according to the basic premium under a given risk cost configuration, that is, the ratio of the expected loss cost of the policy to the basic premium income. The ultimate loss ratio is the final loss ratio after considering the premium adjustment factor under a given risk cost configuration; that is, the ratio of the expected loss cost of the policy to the actual premium income collected. This module outputs the base premium ultimate loss ratio and the ultimate loss ratio for each version, as well as the difference between versions. The difference between versions is the numerical difference between the version to be launched and the current online version on the same evaluation metric under the same evaluation benchmark, including both absolute and relative differences.

[0088] The pre-production risk simulation assessment module 103 offers two modes: full assessment and incremental assessment. The full assessment mode is suitable for initial assessments or scenarios involving large-scale configuration changes. In this mode, the module uses configuration data from both the current online version and the version to be launched, calculating each policy data entry individually. Under each dimension combination, it summarizes the premium, base premium ultimate payout ratio, and ultimate payout ratio indicators for both versions, outputting the base premium ultimate payout ratio, ultimate payout ratio, and inter-version differences for each version. The full assessment mode ensures performance through a distributed parallel computing framework; the full calculation of tens of millions of policy data entries for both versions is completed within 4 hours. The incremental assessment mode is suitable for regular version iteration and update scenarios. In this mode, the module filters dimension combinations where configurations have changed based on the structured difference list output by the version difference comparison module 102. Changed dimensions are those marked in the structured difference list as having at least one changed configuration field. Unchanged dimensions are those where none of the configuration fields in the structured difference list have changed. This module only performs incremental recalculation on policy data corresponding to changed dimensions. The calculation results for policy data corresponding to unchanged dimensions are directly read from and reused from historical cached data. Historical cached data consists of the base premium final payout ratio and final payout ratio calculation results for each dimension combination under each version, generated by the system in historical evaluation calculations and stored in an in-memory database. This module merges the incremental calculation results with the results read from historical cached data to obtain the base premium final payout ratio, final payout ratio, and inter-version differences for each version. The incremental evaluation mode completes the evaluation within 1 hour. The full evaluation mode uses a distributed parallel computing framework to perform sharded parallel processing on all policy data according to dimension combinations. The full calculation of tens of millions of policy data across two versions is completed within 4 hours; in the incremental evaluation mode, historical cached data is stored in an in-memory database, and the reading time for calculation results of unchanged dimensions is in the millisecond range.

[0089] The assessment output includes three indicators: risk cost, base cost ultimate payout ratio, and ultimate payout ratio. Each indicator outputs the values ​​before and after adjustment, as well as the adjustment magnitude, which includes both absolute and relative differences. The system uses preset thresholds to label the adjustment magnitude in three levels: minor, significant, and major, assisting assessors in quickly identifying high-risk dimensions.

[0090] The online effect verification module 104 implements closed-loop verification after the version is launched. This module obtains the actual quotation data after the auto insurance pricing model is launched. The actual quotation data is the premium and cost rate data corresponding to the quotation provided to the policyholder in the real production environment after the auto insurance pricing model is launched. This module compares the cost rate change of the actual quotation data with the base premium ultimate loss ratio and ultimate loss ratio of each version to calculate the deviation rate. The deviation rate is a quantitative measure of the degree of deviation between the cost rate change of the actual quotation data after the version is launched and the predicted cost rate of each version in the pre-production simulation evaluation before the launch. The value is the ratio of the absolute value of the difference between the cost rate change and the predicted cost rate to the predicted cost rate. The predicted cost rate is the base premium ultimate loss ratio and ultimate loss ratio of each version calculated by the pre-production risk simulation evaluation module 103 based on the configuration data of the version to be launched and the policy data in the evaluation benchmark.

[0091] The formula for calculating the deviation rate is: in, Indicates the deviation rate. This represents the cost rate of the actual quoted data. This represents the projected cost rate determined based on the base premium ultimate payout ratio or ultimate payout ratio for each version.

[0092] When the deviation rate is not greater than the preset tolerance range, the online effect verification module 104 determines that the version launch meets expectations. The preset tolerance range is a pre-set allowable range of deviation rates used by the system to determine whether the version launch meets expectations. When the deviation rate exceeds the preset tolerance range, the module triggers an anomaly warning and automatically notifies the relevant responsible persons.

[0093] This module decomposes the impact of version launch on the cost rate by dimension combination. The impact level refers to the change in the overall cost rate caused by the version launch. When a version launch is deemed unsatisfactory, this module determines the cost rate change for each dimension combination before and after the version launch. The cost rate change is the difference in the cost rate before and after the version launch for the same dimension combination. The overall cost rate change is the magnitude of the change in the overall cost rate before and after the version launch, summed for all dimension combinations. This module determines the contribution of each dimension combination based on its proportion of the overall cost rate change and outputs a contribution ranking and an abnormal fluctuation warning. The contribution ranking is the result of sorting each dimension combination from largest to smallest by its proportion of the overall cost rate change; positive values ​​indicate an increase in the cost rate, and negative values ​​indicate a decrease. The abnormal fluctuation warning is a risk alert automatically output by the system when the module identifies that the deviation rate of a certain dimension combination exceeds the preset tolerance range, or that the absolute value of the contribution of that dimension combination significantly deviates from the normal range.

[0094] The formula for calculating contribution is: in, This indicates the contribution of the dimension combination. This indicates the actual cost rate of this dimension combination after the version is launched. This indicates the cost rate of this dimension combination before the version goes live. This indicates the overall cost ratio after the version is launched. This indicates the overall cost rate before the version goes live.

[0095] The data storage module 105 employs a hierarchical storage architecture to store various types of data. Configuration version data is stored in a relational database, supporting transactional consistency and version queries. Basic policy data is stored in a data warehouse, supporting large-scale parallel computing. Simulation evaluation results and deployment verification results are stored in an analytical database, supporting fast multidimensional queries and visualization. Historical calculation caches are stored in an in-memory database, supporting millisecond-level data reuse in incremental mode.

[0096] Figure 3 The overall flowchart of this embodiment is shown. Risk cost configuration is released and stored in the database on a version-by-version basis. Before the version goes live, the version to be launched is compared with the current online version. Then, a pre-production risk simulation assessment is conducted, providing two modes: full deployment and incremental deployment. Differences in three indicators are compared based on a four-dimensional combination of keys. The full deployment mode is completed within 4 hours, and the incremental deployment mode is completed within 1 hour. Finally, the launch effect is verified, including verification of expected compliance and cost rate impact analysis.

[0097] Figure 4 The flowchart illustrates the version difference comparison and full-scale simulation evaluation. The version difference comparison module 102 receives configuration data from the current online version and the version to be launched, and generates a structured difference list through the difference comparison engine. The structured difference list includes dimensional information, change type, change fields, and change magnitude, and classifies the degree of change. The output format supports JSON and Excel. The full-scale simulation evaluation method uses one year's worth of policy data as input, performing full calculations for both version one and version two. The calculations include risk cost, base premium ultimate payout ratio, and ultimate payout ratio. A difference evaluation report is generated through a multi-dimensional comparative analysis engine. The report displays the difference values ​​of the three indicators grouped by four dimensions and classifies them. The full-scale evaluation mode ensures performance through a distributed parallel computing framework; the full calculation of tens of millions of policy data across both versions is completed within 4 hours.

[0098] Figure 5The flowchart illustrates the incremental simulation evaluation and deployment effect verification process. The incremental simulation evaluation method takes a list of version differences as input, filters the changed data, performs incremental calculations only on changed dimensions, and reuses historical results cache for unchanged dimensions, achieving millisecond-level retrieval. The evaluation is completed within one hour, outputting an incremental evaluation report. The report includes details of the changed dimension differences and a comprehensive comparison of all dimensions, including cached and merged results. The deployment effect verification module 104 receives pre-production simulation results and actual post-deployment pricing data, performing expectation compliance verification and pricing cost rate impact analysis. Expectation compliance verification determines whether the version deployment meets expectations by comparing the deviation rate with the tolerance range. Pricing cost rate impact analysis decomposes the contribution of the overall pricing cost rate change after version deployment according to four dimensions, outputting dimension-level impact rankings and abnormal fluctuation warnings. This module ultimately outputs a version deployment evaluation report, which includes expectation compliance determination, cost rate contribution decomposition, high-risk dimension warnings, abnormal fluctuation warnings, and optimization suggestions, providing support for version rollback decisions. Incremental evaluations for regular iterations are completed within one hour.

[0099] In a third embodiment of the pre-production risk simulation assessment system for the vehicle insurance pricing model of the present invention, the system includes: The configuration version management module 101 is used to obtain the risk cost configuration corresponding to the version of the auto insurance pricing model to be launched, and record the version snapshot of the risk cost configuration. The risk cost configuration includes the risk cost coefficient, basic premium and rate adjustment factor. The version snapshot records the version number, release time, publisher, change description and full configuration snapshot. The version snapshot is stored in a relational database.

[0100] The version difference comparison module 102 is used to obtain the configuration data of the current online version and the version to be launched based on the version snapshot before the auto insurance pricing model goes online. It uses the dimension combination composed of business segment, institution level, insurance type and new energy identification as the comparison primary key, compares the configuration parameters of the two versions field by field and generates a structured difference list. The structured difference list includes change type, change field and change magnitude. Change type includes addition, deletion and modification, and change magnitude includes absolute difference and relative difference. The change magnitude is compared with the preset threshold, and the degree of change of each dimension combination is graded and labeled according to the comparison results. The graded labeling level includes slight, significant and major.

[0101] The pre-production risk simulation and assessment module 103 is used to calculate the version change risk coefficient based on the hierarchical labeling results of all dimension combinations in the structured difference list; the version change risk coefficient is determined according to the following formula: in, This indicates the risk factor for version changes. This represents the total number of dimension combinations. Indicates the first The weighting coefficients of the dimension combination, which are based on the weighting coefficients of the first dimension combination. The business scale proportion of each dimension combination is determined. Indicates the first The magnitude of change in the combination of each dimension Indicates the first Preset thresholds for combinations of dimensions, This indicates the number of dimensional combinations that have changed; when the version change risk coefficient is greater than the preset risk threshold, a full evaluation method is used; when the version change risk coefficient is not greater than the preset risk threshold, an incremental evaluation method is used.

[0102] It should be noted that the formula for calculating the version change risk coefficient is based on a preset threshold. The change magnitude is normalized and squared for each dimension combination, and then weighted and averaged according to the business scale ratio of each dimension combination. This method can amplify the impact of large-scale abnormal changes on the overall risk. At the same time, the logarithm of the proportion of the number of changed dimension combinations to the total number of dimension combinations is introduced as a penalty factor for the scope of change coverage, so that the risk coefficient increases as the scope of the change is wider.

[0103] The above formula integrates risk factors from two dimensions: the intensity of change and the scope of change. Its function is to transform the qualitative classification and labeling results in the structured difference list into quantitative risk metrics. By comparing with preset risk thresholds, it automatically determines whether to adopt a full assessment method or an incremental assessment method, thus providing a quantitative decision-making basis for the assessment mode selection.

[0104] The pre-production risk simulation assessment module 104 is also used to calculate the full policy data item by item using the configuration data of the current online version and the configuration data of the version to be launched when adopting the full assessment method, and output the base premium ultimate payout ratio, ultimate payout ratio and inter-version difference value for each version; when adopting the incremental assessment method, it filters the combination of dimensions that have changed configuration according to the structured difference list, performs incremental calculation on the policy data corresponding to the changed dimensions, and reads and reuses the calculation results of the policy data corresponding to the unchanged dimensions from the historical cache data. The results of the incremental calculation are merged with the results read from the historical cache data to obtain the base premium ultimate payout ratio, ultimate payout ratio and inter-version difference value for each version.

[0105] The online effect verification module 105 is used to obtain the actual quotation data after the car insurance pricing model is launched. It compares the cost rate change of the actual quotation data with the base premium final loss ratio and final loss ratio of each version to calculate the deviation rate. The deviation rate is the ratio of the absolute value of the difference between the cost rate change and the predicted cost rate to the predicted cost rate. When the deviation rate is not greater than the preset tolerance range, the version launch is judged to meet expectations. When the deviation rate is greater than the preset tolerance range, an anomaly warning is triggered. The module determines the cost rate change of each dimension combination before and after the version launch, and determines the contribution of each dimension combination according to the proportion of the cost rate change of each dimension combination to the overall cost rate change. It outputs the contribution ranking and anomaly fluctuation warning.

[0106] The technical solution of this embodiment solves the problems of difficulty in tracing version changes during the iteration of auto insurance pricing models, difficulty in quantifying the impact of configuration changes, waste of computing resources due to reliance on experience in manually selecting evaluation methods, and lack of quantitative verification basis after version launch. It improves the objectivity of version change risk assessment and the matching degree of evaluation method selection, reduces the computing resource consumption of full evaluation in low-risk scenarios, and enhances the evaluation efficiency of the version iteration process and the data support capability of version launch decision-making. The configuration version management part records and stores version snapshots of risk cost configuration, the version difference comparison part compares configuration parameters field by field and generates a structured difference list, the pre-production risk simulation assessment part calculates version change risk coefficients based on change magnitude and change scope and automatically selects full evaluation method or incremental evaluation method based on the coefficients, and the online effect verification part compares the cost rate changes of actual quotation data with the pre-production evaluation results and decomposes the contribution of each dimension combination.

[0107] Figure 6 This diagram illustrates an embodiment of a pre-production risk simulation assessment method for an auto insurance pricing model provided by the present invention. The method employs the pre-production risk simulation assessment system 100 for auto insurance pricing models provided by the present invention. Figure 6 As shown, the method includes the following steps: S1. Generate a version snapshot and save it to the database each time the risk cost configuration is released; S2. Based on the version snapshot recorded in the database, before the car insurance pricing model goes online, obtain the configuration data of the current online version and the version to be launched, and compare them field by field. Output a structured difference list with business segments, institutional levels, insurance types and new energy identification as dimensions. S3. Using the policy data within a preset time period as the evaluation benchmark, and based on the dimension combination, calculate the base premium ultimate payout ratio and ultimate payout ratio respectively based on the configuration data of the current online version and the configuration data of the version to be launched, and output the base premium ultimate payout ratio and ultimate payout ratio of each version as well as the difference value between versions; S4. Obtain the actual quote data after the car insurance pricing model is launched, compare the cost rate change of the actual quote data with the base premium ultimate loss ratio and the ultimate loss ratio of each version to determine whether the version launch meets expectations, and decompose the impact of the version launch on the quote cost rate according to the dimensions. S5 stores configuration version data, policy data, simulation evaluation results, and online verification results.

[0108] It should be noted that the beneficial effects of the pre-production risk simulation assessment method for the auto insurance pricing model provided in the above embodiments are the same as those of the above-mentioned pre-production risk simulation assessment system 100 for the auto insurance pricing model, and will not be repeated here. Furthermore, the method and system embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the system embodiments, which will not be repeated here.

[0109] The pre-production risk simulation and assessment system 100 for the vehicle insurance pricing model of the present invention can be a computer program (including program code) running on a computer device. For example, the pre-production risk simulation and assessment system 100 for the vehicle insurance pricing model of the present invention is an application software that can be used to execute the corresponding steps in the pre-production risk simulation and assessment method for the vehicle insurance pricing model of the present invention.

[0110] In some embodiments, the pre-production risk simulation assessment system 100 for the vehicle insurance pricing model of the present invention can be implemented in a combination of hardware and software. As an example, the pre-production risk simulation assessment system 100 for the vehicle insurance pricing model of the present invention can be a processor in the form of a hardware decoding processor, which is programmed to execute the pre-production risk simulation assessment method for the vehicle insurance pricing model of the present invention. For example, the processor in the form of a hardware decoding processor can be one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0111] The modules described in the embodiments of this invention can be implemented in software or hardware. The names of the modules are not, in some cases, limiting the scope of the module itself.

[0112] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the above-mentioned methods for pre-production risk simulation assessment of auto insurance pricing models. That is, an electronic device according to an embodiment of the present invention may include, but is not limited to: a processor and a memory; the memory is used to store the computer program; the processor is used to execute the method for pre-production risk simulation assessment of auto insurance pricing models shown in any embodiment of the present invention by calling the computer program.

[0113] In one alternative embodiment, an electronic device is provided, such as Figure 3 As shown, Figure 3 The illustrated electronic device 4000 includes a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, via a bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 4004 is not limited to one type, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0114] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this invention. Processor 4001 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0115] Bus 4002 may include a path for transmitting information between the aforementioned components. Bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 4002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3The bus 4002 is represented by only one thick line, but this does not mean that there is only one bus or one type of bus.

[0116] The memory 4003 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0117] The memory 4003 stores application code (computer program) for executing the present invention, and its execution is controlled by the processor 4001. The processor 4001 executes the application code stored in the memory 4003 to implement the content shown in the foregoing method embodiments.

[0118] Among them, electronic devices can also be terminal devices. A terminal device can be any terminal device that can install applications and access web pages through applications, including at least one of smartphones, tablets, laptops, desktop computers, smart speakers, smartwatches, smart TVs, and smart in-vehicle devices.

[0119] It should be noted that, Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0120] An embodiment of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the above-mentioned methods for pre-production risk simulation assessment of auto insurance pricing models.

[0121] Alternatively, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, a floppy disk, and an optical data storage device, etc.

[0122] In an exemplary embodiment, a computer program product or computer program is also provided, which includes computer instructions stored in a computer-readable storage medium. The processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the aforementioned pre-production risk simulation assessment method for the vehicle insurance pricing model.

[0123] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0124] It should be understood that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0125] The computer-readable storage medium provided in this invention can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0126] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.

[0127] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0128] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0129] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this invention can be specifically implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0130] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A pre-production risk simulation and assessment system for auto insurance pricing models, characterized in that, include: Configure the version management module to generate a version snapshot and save it to the database each time the risk cost configuration is released; The version difference comparison module is used to obtain the configuration data of the current online version and the version to be launched before the car insurance pricing model goes online, based on the version snapshot recorded in the database, and compare them field by field, and output a structured difference list combined by business segment, institution level, insurance type and new energy identification. The pre-production risk simulation assessment module is used to calculate the base premium ultimate payout ratio and ultimate payout ratio based on the policy data within a preset time period and the configuration data of the version to be launched, according to the combination of dimensions, based on the configuration data of the current online version and the configuration data of the version to be launched, and outputs the base premium ultimate payout ratio and ultimate payout ratio of each version and the difference value between versions. The online effect verification module is used to obtain the actual quotation data after the car insurance pricing model is launched, compare the cost rate change of the actual quotation data with the base premium ultimate loss ratio and the ultimate loss ratio of each version to determine whether the version launch meets expectations, and decompose the impact of the version launch on the quotation cost rate according to the dimensions. The data storage module is used to store configuration version data, policy data, simulation evaluation results, and online verification results.

2. The vehicle insurance pricing model pre-production risk simulation and assessment system according to claim 1, characterized in that, The pre-production risk simulation and assessment module is specifically used for: A full-scale evaluation method is adopted, using the configuration data of the current online version and the configuration data of the version to be launched to calculate the full policy data one by one, and output the base premium ultimate payout ratio, the ultimate payout ratio and the difference value between the versions for each version.

3. The pre-production risk simulation and assessment system for auto insurance pricing models according to claim 1, characterized in that, The pre-production risk simulation and assessment module is specifically used for: An incremental evaluation method is adopted. Based on the structured difference list, the combination of dimensions that have changed is selected and the policy data corresponding to the changed dimensions is incrementally calculated. The calculation results of the policy data corresponding to the unchanged dimensions are read from the historical cache data and reused. The results of the incremental calculation are combined with the results read from the historical cache data to obtain the base fee ultimate payout ratio, the ultimate payout ratio, and the difference value between versions for each version.

4. The pre-production risk simulation and assessment system for auto insurance pricing models according to claim 1, characterized in that, The structured difference list includes the change type, change fields, and change magnitude for each dimension combination. The version difference comparison module is also used to: compare the change magnitude with a preset threshold, and classify and label the degree of change for each dimension combination according to the comparison results.

5. The pre-production risk simulation and assessment system for auto insurance pricing models according to claim 1, characterized in that, The online performance verification module is specifically used for: The cost rate change of the actual quoted data is compared with the base fee ultimate payout rate and the ultimate payout rate of each version to calculate the deviation rate. The deviation rate is the ratio of the absolute value of the difference between the cost rate change and the predicted cost rate to the predicted cost rate. When the deviation rate is not greater than the preset tolerance range, the version launch is determined to be in line with expectations; when the deviation rate is greater than the preset tolerance range, an abnormal warning is triggered.

6. The vehicle insurance pricing model pre-production risk simulation and assessment system according to claim 5, characterized in that, The online performance verification module is also used for: When a version launch is deemed not to meet expectations, the cost rate change of each dimension combination before and after the version launch is determined according to the aforementioned dimension combination. The contribution of each dimension combination is determined based on the proportion of the cost rate change of each dimension combination to the overall cost rate change, and a contribution ranking and abnormal fluctuation warning are output.

7. The pre-production risk simulation and assessment system for auto insurance pricing models according to claim 1, characterized in that, The configuration data includes risk cost coefficient, basic premium, and rate adjustment factor.

8. A method for simulating and assessing pre-production risks of a car insurance pricing model, characterized in that, The method employs the pre-production risk simulation assessment system for auto insurance pricing models as described in any one of claims 1 to 7, the method comprising: Generate a version snapshot and save it to the database each time the risk cost configuration is released; Based on the version snapshot recorded in the database, before the car insurance pricing model goes online, the configuration data of the current online version and the version to be launched are obtained and compared field by field, and a structured difference list is output with business segment, institution level, insurance type and new energy identification as dimensions. Using policy data within a preset time period as the evaluation benchmark, and based on the aforementioned dimension combination, the base premium ultimate payout ratio and ultimate payout ratio are calculated respectively based on the configuration data of the current online version and the configuration data of the version to be launched, and the base premium ultimate payout ratio and ultimate payout ratio of each version, as well as the difference value between versions, are output. Obtain the actual quote data after the car insurance pricing model is launched, compare the cost rate change of the actual quote data with the base premium ultimate loss ratio and the ultimate loss ratio of each version to determine whether the version launch meets expectations, and decompose the impact of the version launch on the quote cost rate according to the dimensions. Store configuration version data, policy data, simulation evaluation results, and online verification results.

9. An electronic device, characterized in that, The electronic device includes a processor coupled to a memory, the memory storing at least one computer program, which is loaded and executed by the processor to enable the electronic device to implement the pre-production risk simulation assessment method for the vehicle insurance pricing model as described in claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one computer program, which, when executed by a processor, implements the pre-production risk simulation assessment method for the vehicle insurance pricing model as described in claim 8.

Images