Auto insurance matching method based on customer historical data and vehicle risk score

Abstract

The present invention discloses a method for matching auto insurance based on customer historical data and vehicle risk scores, which belongs to the field of data processing and includes: collecting historical auto insurance data of target vehicles of target customers; constructing an evaluation parameter matrix and a vehicle parameter matrix, and analyzing and calculating to obtain multiple user scores and multiple risk handling scores; performing combined retrieval analysis respectively to obtain multiple combined special degrees; setting multiple retrieval quantities, and retrieving multiple combined sample auto insurance data sets in the auto insurance database; analyzing and calculating to obtain multiple auxiliary user score sets and multiple auxiliary risk handling score sets; calculating to obtain multiple comprehensive scores, and matching multiple historical auto insurances to obtain matching results based on the multiple comprehensive scores. The present application solves the technical problems of the lack of personalization and precision in the auto insurance matching process in the prior art, and achieves the technical effect of improving the personalization and precision of auto insurance matching based on customer historical data and vehicle risk scores.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a method for matching vehicle insurance based on customer historical data and vehicle risk scores. Background Art

[0002] In the existing technology, auto insurance matching mainly adopts a rule matching method based on static vehicle information and basic driver attributes, and screens and matches auto insurance products by setting thresholds for parameters such as vehicle model, displacement, driver occupation, age, etc. However, the current auto insurance matching method still has some shortcomings. First, it lacks personalization and has not fully integrated the customer's historical auto insurance data, making it difficult to accurately grasp the customer's risk characteristics and insurance preferences; second, the matching rules are relatively static and lack a dynamic adjustment mechanism, which cannot adapt to changes in customer needs and risk conditions; third, data utilization is insufficient, and the risk information and behavioral characteristics contained in the customer's historical auto insurance data have not been effectively mined and utilized, resulting in low matching accuracy. Therefore, the existing auto insurance matching process lacks personalization and lacks accuracy. Summary of the invention

[0003] This application aims to solve the technical problems of lack of personalization and insufficient accuracy in the existing auto insurance matching process by providing an auto insurance matching method based on customer historical data and vehicle risk scores.

[0004] The present application discloses a method for matching auto insurance based on customer historical data and vehicle risk scores, comprising: collecting historical auto insurance data of a target vehicle of a target customer, wherein the historical auto insurance data includes multiple user evaluation data and multiple vehicle risk processing data of multiple historical auto insurances of the target vehicle; constructing an evaluation parameter matrix and a vehicle parameter matrix according to the multiple user evaluation data and the multiple vehicle risk processing data, and analyzing and calculating to obtain multiple user scores and multiple risk processing scores; performing combined search and analysis in an auto insurance database according to multiple combinations of the target vehicle and multiple historical auto insurances, respectively, to obtain multiple combined specialities; setting multiple search quantities according to the multiple combined specialities, and searching in the auto insurance database to obtain multiple combined sample auto insurance data sets, each combined sample auto insurance data including sample user evaluation data and sample vehicle risk processing data; analyzing and calculating to obtain multiple auxiliary user score sets and multiple auxiliary risk processing score sets according to the multiple combined sample auto insurance data sets; calculating to obtain multiple comprehensive scores according to the multiple combined specialities, combining multiple user scores, multiple risk processing scores, multiple auxiliary user score sets and multiple auxiliary risk processing score sets; and matching to obtain matching results according to the multiple comprehensive scores.

[0005] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0006] Since the historical auto insurance data of the target vehicle of the target customer is collected, wherein the historical auto insurance data includes multiple user evaluation data and multiple vehicle risk treatment data of multiple historical auto insurances of the target vehicle, the user evaluation and risk treatment data of the customer's past auto insurance are collected to lay a data foundation for the follow-up; based on the multiple user evaluation data and multiple vehicle risk treatment data, an evaluation parameter matrix and a vehicle parameter matrix are constructed, and multiple user scores and multiple risk treatment scores are obtained by analysis and calculation to provide a basis for subsequent matching; based on multiple combinations of the target vehicle and multiple historical auto insurances, combined retrieval analysis is performed in the auto insurance database to obtain multiple combination specificities, and the frequency of occurrence of the combination of vehicles and auto insurance in the database is retrieved and analyzed to calculate the degree of specificity of different combinations, so as to prepare for setting the retrieval quantity according to the specificity in the follow-up; based on the multiple combination specificities, multiple retrieval quantities are set, and multiple combination sample vehicles are retrieved in the auto insurance database. The invention provides a technical scheme for obtaining matching results by combining multiple historical auto insurance policies and obtaining matching results based on multiple combined sample auto insurance data sets, wherein each combined sample auto insurance data includes sample user evaluation data and sample vehicle risk treatment data, so as to enrich the sample size of matching analysis and improve matching reliability; based on multiple combined sample auto insurance data sets, multiple auxiliary user score sets and multiple auxiliary risk treatment score sets are analyzed and calculated to assist in evaluating the experience and risk performance of target customers under different auto insurance combinations; based on multiple combined special degrees, multiple user scores, multiple risk treatment scores, multiple auxiliary user score sets and multiple auxiliary risk treatment score sets are combined to obtain multiple comprehensive scores, and obtain comprehensive matching scores for different auto insurance combinations; based on multiple comprehensive scores, a technical scheme for obtaining matching results by matching within multiple historical auto insurance policies solves the technical problems of lack of personalization and insufficient precision in the auto insurance matching process in the prior art, and achieves the technical effect of improving the personalization and precision of auto insurance matching based on customer historical data and vehicle risk scores.

[0007] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 A flow chart of a method for matching auto insurance based on customer historical data and vehicle risk scores is provided for an embodiment of the present application;

[0009] Figure 2 A flow chart of obtaining multiple comprehensive scores in a car insurance matching method based on customer historical data and vehicle risk scores is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0010] The overall idea of ​​the technical solution provided by this application is as follows:

[0011] The embodiment of the present application provides a method for matching auto insurance based on customer historical data and vehicle risk scores. First, the historical auto insurance data of the target vehicle of the target customer is collected, including user evaluation data and vehicle risk treatment data. Then, an evaluation parameter matrix and a vehicle parameter matrix are constructed based on the collected data, and the user score and the risk treatment score are obtained by analysis and calculation. Next, based on the combination of the target vehicle and the historical auto insurance, the combination specificity is retrieved and analyzed to obtain the combination specificity, and the number of retrievals is set according to the combination specificity, and the combination sample auto insurance data set is obtained, and then the auxiliary score set is analyzed and calculated. Finally, the combination specificity, user score, risk treatment score and auxiliary score set are combined to obtain a comprehensive score through weighted calculation, and the optimal auto insurance combination is matched according to the comprehensive score.

[0012] Among them, by analyzing the customer's historical auto insurance data, the user's preference for different auto insurances and the risk characteristics of the vehicle are comprehensively evaluated; the introduction of combined specificity helps to dynamically adjust the matching strategy according to the uniqueness of the auto insurance combination; the combined sample auto insurance data is used to assist in the evaluation of the matching degree of the target customers, thereby improving the matching accuracy; in the matching process, multiple factors such as user ratings, risk ratings, specificity, auxiliary ratings, etc. are comprehensively considered to form a comprehensive, dynamic and personalized auto insurance matching mechanism, which overcomes the problems of insufficient personalization and low accuracy of auto insurance matching in the existing technology, and realizes the personalization and precision of auto insurance matching.

[0013] After introducing the basic principles of the present application, the non-limiting implementation methods of the present application will be specifically introduced below in conjunction with the drawings in the specification.

[0014] like Figure 1 As shown, the embodiment of the present application provides a method for matching auto insurance based on customer historical data and vehicle risk scores, the method comprising:

[0015] S1: Collect historical auto insurance data of a target vehicle of a target customer, wherein the historical auto insurance data includes a plurality of user evaluation data and a plurality of vehicle risk processing data of a plurality of historical auto insurances of the target vehicle.

[0016] Specifically, first, the historical auto insurance data of the target customer's target vehicle is collected. Among them, the target customer refers to the specific customer for whom the matching auto insurance is to be recommended, and the target vehicle is the vehicle that the target customer intends to insure. The historical auto insurance data includes relevant information on multiple historical auto insurances purchased by the target customer for the target vehicle in the past period of time (such as in recent years), including customer evaluation data on multiple historical auto insurances, such as ratings or feedback on auto insurance service attitude, service efficiency, price, etc., forming multiple user evaluation data, and also including the auto insurance company's risk handling data on the target vehicle, such as the auto insurance company's response speed and compensation ratio when the target vehicle is in danger, forming multiple vehicle risk handling data. If the target vehicle of a historical auto insurance has not been in danger during its underwriting period, the average risk handling data of the historical auto insurance is used as a substitute for subsequent analysis.

[0017] Through data collection, we obtained rich historical data, including multiple user evaluation data of multiple historical auto insurance policies of the target vehicle and multiple vehicle risk processing data, which provided a basis for achieving accurate matching of auto insurance.

[0018] S2: constructing an evaluation parameter matrix and a vehicle parameter matrix according to the plurality of user evaluation data and the plurality of vehicle risk processing data, and analyzing and calculating to obtain a plurality of user scores and a plurality of risk processing scores.

[0019] Specifically, first, an evaluation parameter matrix is ​​constructed based on the collected multiple user evaluation data. The evaluation parameters of multiple evaluation indicators of each historical auto insurance are used as matrix elements, and are arranged according to the two dimensions of auto insurance and indicators to form a matrix. By analyzing and calculating the matrix, the user score of each historical auto insurance can be obtained, reflecting the customer's overall evaluation of the auto insurance, thereby obtaining multiple user scores. At the same time, a vehicle parameter matrix is ​​constructed based on multiple vehicle risk processing data. Similarly, the parameters of each auto insurance on multiple risk processing indicators are used as matrix elements, and are arranged according to the two dimensions of auto insurance and indicators. After analysis and calculation, the risk processing score of each historical auto insurance can be obtained, reflecting the claim service level of the auto insurance when the target vehicle is in danger, thereby obtaining multiple risk processing scores. In the process of constructing the matrix, the original user evaluation data and vehicle risk processing data are pre-processed such as normalization to eliminate the dimensional differences between different indicators and different auto insurances, so as to facilitate unified analysis and calculation.

[0020] Through matrix construction and computational analysis, the scattered evaluation data and processing data are summarized into intuitive user scores and risk management scores, which quantitatively characterize the performance of the target vehicle's historical auto insurance in terms of user experience and risk management, providing important reference indicators for subsequent auto insurance matching.

[0021] S3: According to the multiple combinations of the target vehicle and multiple historical auto insurances, combination retrieval and analysis are performed in the auto insurance database to obtain multiple combination specificities.

[0022] Specifically, by analyzing the uniqueness of the combination of the target vehicle and multiple historical auto insurance policies in the entire auto insurance database, a unique measure reflecting the rarity of each combination is obtained.

[0023] Specifically, first, multiple combinations of the target vehicle and multiple historical auto insurances are enumerated, such as "target vehicle + auto insurance A", "target vehicle + auto insurance B", etc. Then, for each combination, a search is performed in the entire auto insurance database, and the proportion of the combination is counted as the combination rate of the combination. Next, the combination rate of each combination is compared with the overall average combination rate to calculate the specificity of each combination. For each combination, its combination rate is divided by the average combination rate to obtain a ratio, and the reciprocal of the ratio is used as the specificity of the combination. In this way, if the combination rate of a combination is much lower than the average level, its ratio will be very small, the reciprocal will be very large, and the combination specificity will be higher. Conversely, if the combination rate of a combination is close to or higher than the average level, its ratio is close to or greater than 1, the reciprocal will be relatively small, and the combination specificity will be low. For example, assuming that the frequency of occurrence of the combination "target vehicle + auto insurance A" in the entire auto insurance database is 1%, and the average frequency of occurrence of all combinations is 5%, then the ratio of the combination is 1% / 5%=0.2, and its reciprocal is 5, which is the combination specificity of the combination.

[0024] Through the combination of specialness, we can identify whether the combination of the target vehicle and certain historical auto insurance policies is special. The specialness may come from the characteristics of the target vehicle itself (such as vehicle model, age, etc.), or from the personal preferences of the insured. For combinations with high combination specialness, their historical data is relatively sparse, and the historical auto insurance data of the target vehicle itself is not enough to support subsequent analysis and calculation. Therefore, it is necessary to analyze the combination within a larger database to mine information with reference value.

[0025] By calculating the combination specificity, the rarity of different vehicle and auto insurance combinations is quantitatively characterized, providing a basis for subsequent external data supplementation and analysis. For combinations with higher specificity, the scale of retrieval data will be expanded in a targeted manner to make up for the problem of data sparsity, while for combinations with lower specificity, more reliance can be placed on the historical data of the target vehicle itself, thereby improving the accuracy and reliability of auto insurance matching overall.

[0026] S4: according to the plurality of combined specificities, a plurality of search quantities are set, and a plurality of combined sample auto insurance data sets are retrieved in the auto insurance database, each combined sample auto insurance data set includes sample user evaluation data and sample vehicle risk processing data.

[0027] Specifically, after obtaining the specificity of multiple combinations of target vehicles and historical auto insurance, other data similar to each combination is retrieved and collected from the auto insurance database according to the specificity of the combination, forming multiple combined sample auto insurance data sets. For each combination of "target vehicle-historical auto insurance", a search quantity is first determined based on its combination specificity. Among them, the higher the combination specificity, the rarer the combination is in the database, and the less similar data is available for reference. At this time, it is necessary to expand the search scope and obtain more sample data; conversely, for combinations with lower combination specificity, the number of searches should be appropriately reduced to save computing resources.

[0028] After determining the number of retrievals, conditional retrieval is performed in the auto insurance database using each combination as the retrieval condition. For each combination, a certain number of auto insurance data of other customers are randomly sampled. These data are similar to the characteristics of the combination and are regarded as neighborhood samples of the combination. The sampled data is a combined sample auto insurance dataset. Each dataset contains multiple auto insurance purchase records of other customers, as well as corresponding user evaluation data and vehicle risk processing data. By searching multiple combinations separately, multiple combined sample auto insurance datasets are obtained, thereby specifically supplementing the information of the target vehicle's historical auto insurance data, especially for rare combinations, enriching its analysis samples. These additionally collected multiple combined sample auto insurance datasets reflect the auto insurance choices and usage experience of other car owners with similar conditions to the target vehicle, providing a reference for subsequent user scoring and risk assessment.

[0029] Through targeted data retrieval and sampling, external reference information of historical vehicle insurance data of the target vehicle is supplemented in the context of the entire vehicle insurance database, which not only alleviates the problem of insufficient rare combination data, but also provides a more comprehensive and reliable basis for subsequent user scoring and risk assessment.

[0030] S5: Analyze and calculate, based on the multiple combined sample auto insurance data sets, to obtain multiple auxiliary user score sets and multiple auxiliary risk management score sets.

[0031] Specifically, based on the collected multiple combined sample auto insurance data sets, by constructing an evaluation parameter matrix and a vehicle parameter matrix, and performing analysis and calculation, multiple auxiliary user scores and multiple auxiliary risk management scores corresponding to the historical auto insurance of the target vehicle are obtained.

[0032] First, a sample evaluation parameter matrix is ​​constructed based on the user evaluation data in each combined sample auto insurance data set. Similar to the evaluation parameter matrix, the evaluation parameters of multiple evaluation indicators of each auto insurance are used as matrix elements and arranged according to the two dimensions of auto insurance and indicators. By analyzing and calculating the matrix, the auxiliary user score of each auto insurance under the combined sample is obtained. At the same time, a sample vehicle parameter matrix is ​​constructed based on the vehicle risk treatment data in each combined sample auto insurance data set. Similarly, the parameters of each auto insurance on multiple risk treatment indicators are used as matrix elements and arranged according to the two dimensions of auto insurance and indicators. After analysis and calculation, the auxiliary risk treatment score of each auto insurance under the combined sample is obtained. Since there are multiple combined sample data sets collected, after the above processing, multiple auxiliary user score sets and multiple auxiliary risk treatment score sets are obtained.

[0033] By utilizing the collected external data, an auxiliary evaluation of the target vehicle's historical auto insurance is conducted from multiple perspectives, providing a reference for the target vehicle's auto insurance matching.

[0034] S6: Calculate and obtain multiple comprehensive scores based on the multiple combined special degrees, in combination with the multiple user scores, the multiple risk management scores, the multiple auxiliary user score sets, and the multiple auxiliary risk management score sets.

[0035] Specifically, for each historical auto insurance to be evaluated, first obtain its original user score and risk handling score and use them as the basic score. Then, from the obtained multiple auxiliary user score sets and multiple auxiliary risk handling scores, select the auxiliary user score and auxiliary risk handling score corresponding to the current auto insurance. Next, combined with the calculated combination specificity, different weights are assigned to the basic score and the auxiliary score. Generally speaking, if the combination specificity of the current auto insurance and the target vehicle is high, it means that the historical data of the target vehicle itself is relatively sparse, and a higher weight is given to the auxiliary score at this time; conversely, if the combination specificity is low, it means that the historical data of the target vehicle itself is relatively rich, and a higher weight is given to the basic score at this time. After determining the weight, the basic user score and the auxiliary user score are weighted averaged to obtain the comprehensive user score; at the same time, the basic risk handling score and the auxiliary risk handling score are weighted averaged to obtain the comprehensive risk handling score. These two scores respectively reflect the comprehensive performance of the auto insurance in the two dimensions of user experience and risk management. Afterwards, the comprehensive user score and the comprehensive risk management score are weighted and summed to obtain the final comprehensive score of the auto insurance, which takes into account the user's subjective feelings and objective risk performance, and is a comprehensive assessment of the overall matching degree of the auto insurance.

[0036] Through the above calculation process, a comprehensive score for each historical auto insurance is obtained, thereby obtaining multiple comprehensive scores. These comprehensive scores take into account the historical data of the target vehicle itself and the external reference data, and make adaptive weight adjustments according to the sparsity of the data, which can more accurately and comprehensively reflect the pros and cons of each auto insurance.

[0037] S7: According to the multiple comprehensive scores, matching is performed within the multiple historical auto insurance policies to obtain a matching result.

[0038] Specifically, after obtaining the comprehensive scores of each historical auto insurance, the comprehensive scores are sorted to obtain matching results, thus completing the auto insurance matching process.

[0039] First, sort the comprehensive scores of all historical auto insurances obtained, and arrange the scores from high to low, which reflects the matching quality of each auto insurance relative to the target vehicle. The higher the score, the higher the fit between the auto insurance and the target vehicle, and the more suitable it is for recommendation. Next, according to actual needs, select one or several auto insurances with the highest scores from the sorted auto insurance list as the matching results recommended to the target customers. For example, select the auto insurance with the highest comprehensive score as the first recommendation, and select the other auto insurances ranked in the top few as alternatives, so as to obtain an auto insurance recommendation set containing one or more auto insurances as the matching result.

[0040] When showing the recommendation results to customers, not only the name and basic information of the recommended auto insurance are given, but also the comprehensive score of each auto insurance, as well as its performance scores in the two dimensions of user evaluation and risk management, so as to help customers more comprehensively understand the advantages and disadvantages of each recommended plan, and make comparisons and choices. In addition, the matching results can be further personalized according to the specific needs and preferences of customers. For example, if the customer pays more attention to user reputation, the weight of the user score can be appropriately increased; if the customer pays more attention to risk protection, the weight of the risk handling score can be appropriately increased. In this way, based on the comprehensive score, different customers can be provided with matching results that are more in line with their needs.

[0041] By obtaining matching results based on multiple comprehensive scores, we provide reference and guidance for customers' auto insurance selection, thereby improving the accuracy, comprehensiveness and personalization of auto insurance matching.

[0042] Furthermore, the embodiment of the present application also includes:

[0043] Collect multiple historical auto insurance policies purchased by the target customer's target vehicle within a historical period;

[0044] Collecting a plurality of user evaluation data of the target customers on the plurality of historical auto insurances, wherein each user evaluation data includes evaluation parameters of a plurality of evaluation indicators;

[0045] Collecting a plurality of risk processing data of the plurality of historical vehicle insurances on the target vehicle, wherein each risk processing data includes processing parameters of a plurality of types of processing indicators;

[0046] The plurality of historical auto insurances, the plurality of user evaluation data and the plurality of risk processing data are integrated to obtain historical auto insurance data.

[0047] In a feasible implementation, first, a historical time range is determined, such as the past three or five years. Then, all the auto insurance records of the target customer for the target vehicle during this historical time are retrieved from the auto insurance sales and management system to obtain multiple historical auto insurances, including the types of auto insurances, coverage items, insured amounts, premiums, effective time and other basic elements. After obtaining the basic information of the historical auto insurances, the evaluation feedback of the target customers on these historical auto insurances is further collected. By accessing the customer evaluation system or questionnaires, the scores or comments of the target customers on multiple evaluation indicators of each historical auto insurance are collected, such as the scores and opinions on the service quality, claims efficiency, cost-effectiveness and other aspects of the auto insurance, and multiple user evaluation data are obtained. These user evaluation data reflect the subjective feelings and satisfaction of the target customers on different auto insurances. Among them, each user evaluation data includes evaluation parameters of multiple evaluation indicators, such as service quality score, claims efficiency score, cost-effectiveness score, etc., which express the customer's evaluation in a quantitative form.

[0048] In addition to the subjective evaluation of the target customers, it is also necessary to collect the risk handling of historical auto insurance in actual operation. By accessing the auto insurance claims system or risk management system, the risk accidents and their handling data of the target vehicle during the coverage period of each historical auto insurance are collected, such as accident frequency, loss amount, compensation amount, case processing time and other indicators. These risk handling data reflect the performance of different auto insurance in objective risk management. Among them, each risk handling data includes processing parameters of multiple types of processing indicators, such as accident frequency, average loss amount, average compensation amount, average case processing time, etc., which express various risk handling indicators in a quantitative form. After collecting the basic information of historical auto insurance, user evaluation data and risk handling data respectively, the multiple historical auto insurances, multiple user evaluation data and multiple risk handling data are integrated, and organized into structured historical auto insurance data according to the hierarchical structure of auto insurance-evaluation-risk, which comprehensively records the insurance information, customer experience and risk performance of the target vehicle under each historical auto insurance, providing a complete data basis for subsequent auto insurance matching analysis.

[0049] Furthermore, the embodiment of the present application also includes:

[0050] Performing normalization and maximization processing on the evaluation parameters and processing parameters in the plurality of user evaluation data and the plurality of vehicle risk processing data to obtain a plurality of processing evaluation parameter sets and a plurality of processing vehicle parameter sets;

[0051] According to the multiple processing evaluation parameter sets, an evaluation parameter matrix is ​​constructed as follows:

[0052]

[0053] Among them, A is the evaluation parameter matrix, m is the number of evaluation indicators in the evaluation parameter set, and n is the number of multiple historical auto insurances. is the processing evaluation parameter of the first evaluation index of the first historical auto insurance, is the processing evaluation parameter of the first evaluation index of the nth historical auto insurance, is the processing evaluation parameter of the mth evaluation index of the first historical auto insurance, is the processing evaluation parameter of the mth evaluation index of the nth historical auto insurance;

[0054] Analyze and calculate to obtain multiple user ratings according to the evaluation parameter matrix;

[0055] A vehicle parameter matrix is ​​constructed based on the multiple processing vehicle parameter sets, and multiple risk processing scores are obtained through analysis and calculation.

[0056] In a preferred embodiment,

[0057] First, the evaluation parameters and processing parameters in the original multiple user evaluation data and multiple vehicle risk processing data are normalized and converted into dimensionless values ​​between 0 and 1. Through normalization, the dimensional differences between different indicators and different auto insurances can be eliminated, making subsequent analysis and calculation more accurate and reliable. Among them, normalization can adopt a variety of methods such as linear normalization and logarithmic normalization. Secondly, the normalized parameters are maximized to unify the pros and cons of each indicator as "the larger the value, the better". For indicators whose smaller values ​​are better, they are converted by subtracting the normalized value from 1. After normalization and maximization, multiple processing evaluation parameter sets and multiple processing vehicle parameter sets are obtained. The parameters in each parameter set are dimensionless and maximized, which is convenient for subsequent matrix construction and analysis and calculation. Then, based on multiple processing evaluation parameter sets, the evaluation parameter matrix is ​​constructed as follows:

[0058]

[0059] Among them, A is the evaluation parameter matrix, m is the number of evaluation indicators in the evaluation parameter set, and n is the number of multiple historical auto insurances. is the processing evaluation parameter of the first evaluation index of the first historical auto insurance, is the processing evaluation parameter of the first evaluation index of the nth historical auto insurance, is the processing evaluation parameter of the mth evaluation index of the first historical auto insurance, is the processing evaluation parameter of the mth evaluation index of the nth historical auto insurance. The m processing evaluation parameter sets are arranged in the form of a matrix to construct an evaluation parameter matrix A with n rows and m columns. Each row of the matrix corresponds to a historical auto insurance, and each column corresponds to an evaluation index. The matrix elements represent the processing evaluation parameters of the nth historical auto insurance on the mth evaluation index. Through matrix construction, the scattered processing evaluation parameter sets are integrated into a regular matrix structure, which facilitates the subsequent user rating calculation.

[0060] Next, the evaluation parameter matrix A is used as input to design a scoring function and calculate the user score of each historical auto insurance. The scoring function can be a weighted average or a nonlinear combination. The design of the scoring function fully considers the importance and relevance of each evaluation indicator, as well as the distribution characteristics of user scores. Through the calculation of the scoring function, an n-dimensional user score vector is obtained. Each element of the vector corresponds to a user score of a historical auto insurance, reflecting the comprehensive performance of the auto insurance in the subjective evaluation of the target customers. At the same time, similar to the construction of the evaluation parameter matrix, the processing vehicle parameter set is integrated into a vehicle parameter matrix, and a risk scoring function is designed to calculate the risk treatment score of each historical auto insurance. Among them, the design of the risk scoring function comprehensively considers the importance and relevance of each risk treatment indicator, as well as the distribution characteristics of the risk score. Through the calculation of the risk scoring function, an n-dimensional risk treatment score vector is obtained, reflecting the comprehensive performance of each historical auto insurance in objective risk management.

[0061] Through the conversion process from original evaluation data and risk treatment data to user scores and risk treatment scores, not only the efficiency and accuracy of data analysis are improved, but also the data foundation is laid for subsequent auto insurance matching.

[0062] Furthermore, the embodiment of the present application also includes:

[0063] According to the evaluation parameter matrix, multiple user scores are obtained by analysis and calculation, as shown in the following formula:

[0064]

[0065] Among them, G i is the user rating of the i-th historical auto insurance, w j is the weight of the jth evaluation index, is the maximum value among the n processing evaluation parameters in the jth evaluation index in the evaluation parameter matrix, is the processing evaluation parameter of the jth evaluation index of the ith historical auto insurance in the evaluation parameter matrix, It is the minimum value among the n processing evaluation parameters in the j-th evaluation index in the evaluation parameter matrix.

[0066] In a preferred embodiment, a user scoring function is designed to obtain multiple user scores according to the evaluation parameter matrix. The user scoring function is: Among them, G i is the user rating of the i-th historical auto insurance. The user rating function comprehensively considers the relative closeness between the performance of each auto insurance on each evaluation index and the best performance and the worst performance. First, two intermediate variables are defined as and represents the weighted Euclidean distance between the processing evaluation parameter of the i-th historical auto insurance on each evaluation index and the maximum value of the n processing evaluation parameters, where, Indicates the maximum value in the jth column of the evaluation parameter matrix (i.e., the jth evaluation index), which is the maximum value of the index; represents the element in the i-th row and j-th column of the evaluation parameter matrix, i.e., the processing evaluation parameter of the i-th historical auto insurance on the j-th indicator; w j represents the weight of the jth indicator, reflecting the importance of the indicator in user evaluation. The calculation formula is: Among them, m is the total number of evaluation indicators. Similarly, It represents the weighted Euclidean distance between the processing evaluation parameter of the i-th historical auto insurance on each evaluation index and the minimum value of the n processing evaluation parameters. It represents the minimum value in the jth column of the evaluation parameter matrix, which is the worst value of the indicator. The calculation formula is:

[0067] get and Afterwards, follow Calculate the user rating G of the i-th historical auto insurance i , G i yes and When the performance of the i-th historical auto insurance on each evaluation index is closer to the maximum value and farther away from the minimum value, will be relatively large, will be relatively small, so G i The larger it is, the maximum is 1; on the contrary, when the performance of the i-th historical auto insurance on each evaluation index is closer to the minimum value and farther away from the maximum value, will be relatively small, will be relatively large, so G i The smaller it is, the smaller it is, and the minimum is 0. Therefore, G i The size of G reflects the quality of the i-th historical auto insurance in terms of user evaluation. i The bigger it is, the more popular and recognized the auto insurance is among users.

[0068] Through the above calculation process, we get the n-dimensional user rating vector G = (G 1 , G 2 , …, G n ), each element of the vector corresponds to a user rating of historical auto insurance, thereby obtaining multiple user ratings.

[0069] Furthermore, the embodiment of the present application also includes:

[0070] Collecting multiple sample combinations, wherein the multiple sample combinations are obtained by traversing multiple sample vehicles and multiple sample auto insurance combinations;

[0071] Obtaining multiple sample combination rates of the multiple sample combinations in the motor vehicle insurance database, and calculating an average combination rate;

[0072] According to the multiple combinations of the target vehicle and multiple historical auto insurances, performing combination searches in the auto insurance database respectively to obtain multiple combination rates;

[0073] A plurality of combination specificities are calculated based on the plurality of combination rates and the average combination rate.

[0074] In a preferred embodiment, first, a certain number of sample vehicles and sample auto insurances are randomly selected from the auto insurance database to obtain multiple sample vehicles and multiple sample auto insurances, and these sample vehicles and sample auto insurances are combined in pairs to obtain multiple sample combinations. These sample combinations cover multiple dimensions such as different vehicle models, vehicle ages, and insurance types. Then, for each sample combination, the number of records that fully match it is retrieved in the auto insurance database, and then divided by the total number of records in the auto insurance database to obtain the combination rate of the sample combination, thereby obtaining multiple combination rates. The combination rate reflects the frequency or prevalence of the combination in the entire auto insurance database. After calculating the combination rates of all sample combinations, the arithmetic mean of these combination rates is taken to obtain the average combination rate of the sample combination, which is used as a benchmark value for measuring the prevalence of the combination.

[0075] At the same time, the target vehicle is combined with each historical auto insurance to obtain multiple target combinations. Then, the number of matching records for each target combination is retrieved from the auto insurance database and divided by the total number of records in the auto insurance database to obtain the combination rate of each target combination. These combination rates reflect the frequency of occurrence of the target vehicle and different historical auto insurance combinations in the entire database. After that, the average combination rate is divided by the combination rate of each target combination to obtain the combination specificity of the target combination. The combination specificity is a relative indicator that measures the rarity of a target combination relative to the average level of all combinations. If the combination rate of a target combination is much lower than the average combination rate, then the combination specificity of the target combination is high, indicating that the combination is relatively rare in the entire auto insurance database; conversely, if the combination rate of a target combination is close to or higher than the average level, then the combination specificity of the target combination is relatively low, indicating that this combination is relatively common. Through relative comparison, the specificity value of each target combination is obtained, and multiple combination specificities are obtained.

[0076] By extracting the specific information of the target vehicle and historical auto insurance combinations from the auto insurance database, the specificity of each target combination is quantitatively characterized, providing targeted reference information for subsequent auto insurance matching.

[0077] Furthermore, the embodiment of the present application also includes:

[0078] Get the preset auxiliary analysis quantity;

[0079] According to the multiple combined specificities, the preset auxiliary analysis quantity is adjusted and calculated to obtain multiple retrieval quantities;

[0080] According to the multiple search quantities and multiple combinations, a plurality of combined sample auto insurance data sets that meet the multiple combinations are randomly searched in the auto insurance database, and each combined sample auto insurance data set includes sample user evaluation data and sample vehicle risk processing data.

[0081] In a feasible implementation, first, a preset auxiliary analysis number is set as the basis for adjusting the subsequent search number. The preset auxiliary analysis number is determined according to the size of the motor insurance database, the data processing capacity, and the required auxiliary analysis accuracy. The larger the preset auxiliary analysis number, the more sample data is obtained, and the auxiliary analysis results will be more accurate and stable; but at the same time, too large a number will also bring problems of computing efficiency and resource consumption. Therefore, a balance is made between accuracy and efficiency, and a suitable preset value is selected. Then, for each target combination, the preset auxiliary analysis number is adjusted according to its combination specificity to obtain the actual search number of the target combination. The basic principle of adjustment is: the higher the combination specificity, the larger the search number; the lower the combination specificity, the smaller the search number. This is because combinations with high specificity are relatively rare in the motor insurance database. In order to obtain sufficient auxiliary analysis samples, the search range needs to be expanded; while combinations with low specificity are relatively common, and the sample data is relatively sufficient, and the search amount can be appropriately reduced to improve efficiency. For example, the adjustment calculation adopts the method of linear proportional transformation, that is, the adjusted search number = preset auxiliary analysis number × (1 + k × (combination specificity - 1)). Among them, k is a proportional coefficient, which controls the influence of specificity on the number of searches. The larger the k value, the greater the number of searches for high-specificity combinations, and the smaller the number of searches for low-specificity combinations. The selection of the k value is determined by comprehensively considering factors such as the database size and the distribution of combination specificity. Through the above adjustment calculation, the number of searches for each target combination is formed, and multiple search numbers are obtained.

[0082] Subsequently, for each target combination, according to its corresponding retrieval quantity, the matching auto insurance data is randomly retrieved from the auto insurance database to form a combined sample auto insurance data set. This data set contains auto insurance purchase cases of other users who match the target combination in terms of vehicle model, vehicle age, insurance type, etc. Each case includes corresponding sample user evaluation data and sample vehicle risk treatment data. During the retrieval process, if the actual matching data volume of a target combination is less than its retrieval quantity, all matching data are taken as its combined sample auto insurance data set; if the actual matching data volume is greater than the retrieval quantity, the data of the specified retrieval quantity is randomly selected from it as the combined sample auto insurance data set. By searching all target combinations one by one, a combined sample auto insurance data corresponding to each combination is obtained, forming multiple combined sample auto insurance data sets.

[0083] Furthermore, the embodiment of the present application also includes:

[0084] constructing a plurality of sample evaluation parameter matrices and a plurality of sample vehicle parameter matrices respectively according to the plurality of combined sample auto insurance data sets;

[0085] According to the multiple sample evaluation parameter matrices and the multiple sample vehicle parameter matrices, multiple auxiliary user score sets and multiple auxiliary risk management score sets are obtained through analysis and calculation.

[0086] In a feasible implementation, for each combined sample auto insurance data set, the user evaluation data and vehicle risk processing data therein are extracted respectively, and the corresponding sample evaluation parameter matrix and sample vehicle parameter matrix are constructed in a similar manner to the construction of the evaluation parameter matrix and the construction of the vehicle parameter matrix. Since there are multiple combined sample auto insurance data sets, the matrix construction obtains multiple sample evaluation parameter matrices and multiple sample vehicle parameter matrices equal to the number of data sets, which correspond to different target combinations, respectively, and reflect the statistical characteristics of the target combination in terms of user evaluation and risk processing. Subsequently, for each sample evaluation parameter matrix, the same method as that for obtaining multiple user scores is used to obtain the corresponding auxiliary user score set. At the same time, for each sample vehicle parameter matrix, the same method as that for obtaining multiple risk processing scores is used to obtain the corresponding auxiliary risk processing score set. By calculating all combined samples one by one, multiple auxiliary user score sets and multiple auxiliary risk processing score sets equal to the number of target combinations are obtained, reflecting the statistical distribution characteristics of the corresponding target combination in terms of user evaluation or risk processing, and reflecting the overall performance of the combination.

[0087] Further, such as Figure 2 As shown, the embodiment of the present application also includes:

[0088] Calculate and obtain multiple auxiliary user scores and multiple auxiliary risk management scores according to the multiple auxiliary user score sets and the multiple auxiliary risk management score sets;

[0089] According to the plurality of combined special degrees, configuring a plurality of basic weights and a plurality of auxiliary weights;

[0090] According to the multiple basic weights and the multiple auxiliary weights, the multiple user scores and the multiple auxiliary user scores are weighted and calculated to obtain multiple comprehensive user scores, and the multiple risk treatment scores and the multiple auxiliary risk treatment scores are weighted and calculated to obtain multiple comprehensive risk treatment scores;

[0091] A plurality of comprehensive scores are calculated based on the plurality of comprehensive user scores and the plurality of comprehensive risk handling scores.

[0092] In a preferred embodiment, first, for each target combination, the auxiliary user score and auxiliary risk handling score of the combination are extracted from the corresponding auxiliary user score set and auxiliary risk handling score set, which can be the arithmetic mean of all scores in the set, or other representative values, such as median, mode, etc. The specific comprehensive index selected depends on the distribution characteristics of the score set and the application requirements. At the same time, a weight is set for the user score, risk handling score, auxiliary user score and auxiliary risk handling score of each target combination for subsequent weighted calculation. Among them, the weight of the user score and the risk handling score is called the basic weight, and the weight of the auxiliary user score and the auxiliary risk handling score is called the auxiliary weight. The configuration principle of the weight is: the higher the combination specificity, the greater the auxiliary weight and the smaller the basic weight; conversely, the lower the combination specificity, the greater the basic weight and the smaller the auxiliary weight. This is because, for a combination with high specificity, the historical data of the target vehicle itself is relatively sparse, and the reliability of the score is low, so more reference to the external auxiliary score is needed; while for a combination with low specificity, the historical data of the target vehicle itself is relatively rich, and the reliability of the score is high, so more reliance is placed on the basic score. For example, the specific configuration of the weight is inversely proportional to the combination specificity, that is, basic weight = 1 / combination specificity; auxiliary weight = 1-basic weight. In this way, the greater the combination specificity, the smaller the basic weight, the greater the auxiliary weight, and the sum of the two is always 1.

[0093] Then, for each target combination, multiply the user score by its basic weight, multiply the auxiliary user score by its auxiliary weight, and add the two together to get the comprehensive user score of the combination; at the same time, multiply the risk treatment score by the basic weight, multiply the auxiliary risk treatment score by the auxiliary weight, and add the two together to get the comprehensive risk treatment score of the combination, and get the comprehensive user score and comprehensive risk treatment score of each target combination. After obtaining the comprehensive user score and comprehensive risk treatment score of each target combination, perform weighted average processing to obtain multiple comprehensive scores, thereby comprehensively and accurately reflecting the advantages and disadvantages and adaptability of each historical auto insurance.

[0094] In summary, the auto insurance matching method based on customer historical data and vehicle risk score provided in the embodiment of the present application has the following technical effects:

[0095] The historical auto insurance data of the target vehicle of the target customer is collected, wherein the historical auto insurance data includes multiple user evaluation data and multiple vehicle risk treatment data of multiple historical auto insurances of the target vehicle, providing a data basis for the subsequent analysis of customer preferences and vehicle risk characteristics. Based on multiple user evaluation data and multiple vehicle risk treatment data, an evaluation parameter matrix and a vehicle parameter matrix are constructed, and multiple user scores and multiple risk treatment scores are obtained by analysis and calculation, which quantifies the customer's evaluation level of different auto insurances and the risk performance of the vehicle under different auto insurances, and obtains intuitive numerical indicators for subsequent matching analysis. Based on multiple combinations of the target vehicle and multiple historical auto insurances, combined retrieval analysis is performed in the auto insurance database to obtain multiple combination specificities, which are used to measure the uniqueness of different auto insurance combinations and provide a differentiated basis for subsequent matching. Based on multiple combination specificities, multiple retrieval quantities are set, and multiple combined sample auto insurance data sets are retrieved in the auto insurance database. Each combined sample auto insurance data includes sample user evaluation data and sample vehicle risk treatment data. By expanding the sample size, the reliability of matching analysis is improved. Based on multiple combined sample auto insurance data sets, multiple auxiliary user score sets and multiple auxiliary risk treatment score sets are analyzed and calculated, reflecting the experience and risk performance of other users under the same auto insurance combination, which are used to assist in evaluating the matching degree of target customers. Based on multiple combined special degrees, multiple user scores, multiple risk treatment scores, multiple auxiliary user score sets and multiple auxiliary risk treatment score sets are calculated to obtain multiple comprehensive scores to form a comprehensive and balanced matching metric. Based on multiple comprehensive scores, matching results are obtained within multiple historical auto insurances. Based on the comprehensive scores, the auto insurance with the highest comprehensive score is selected from the historical auto insurance of the target customer as the matching result to complete personalized and precise auto insurance matching.

[0096] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0097] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A car insurance matching method based on customer historical data and vehicle risk score, characterized in that: The method comprises: Collecting historical auto insurance data of a target vehicle of a target customer, wherein the historical auto insurance data includes a plurality of user evaluation data and a plurality of vehicle risk processing data of a plurality of historical auto insurances of the target vehicle; According to the plurality of user evaluation data and the plurality of vehicle risk treatment data, construct an evaluation parameter matrix and a vehicle parameter matrix, and analyze and calculate to obtain a plurality of user scores and a plurality of risk treatment scores; According to the multiple combinations of the target vehicle and the multiple historical auto insurances, combination retrieval analysis is performed in the auto insurance database to obtain multiple combination special degrees, wherein the combination special degree is obtained by analyzing the special degree of the combination of the target vehicle and the multiple historical auto insurances in the entire auto insurance database to obtain a special measure reflecting the rarity of each combination; According to the plurality of combined special degrees, a plurality of search quantities are set, and a plurality of combined sample auto insurance data sets are retrieved in the auto insurance database, each combined sample auto insurance data set includes sample user evaluation data and sample vehicle risk processing data; Analyzing and calculating, based on the multiple combined sample auto insurance data sets, multiple auxiliary user score sets and multiple auxiliary risk management score sets are obtained; Calculate and obtain multiple comprehensive scores based on the multiple combined special degrees, combining the multiple user scores, the multiple risk management scores, the multiple auxiliary user score sets, and the multiple auxiliary risk management score sets; According to the multiple comprehensive scores, matching results are obtained within the multiple historical auto insurance policies.

2. The method for matching auto insurance based on customer historical data and vehicle risk score according to claim 1, characterized in that: Collect historical auto insurance data of target customers’ target vehicles, including: Collect multiple historical auto insurance policies purchased by the target customer's target vehicle within a historical period; Collecting a plurality of user evaluation data of the target customers on the plurality of historical auto insurances, wherein each user evaluation data includes evaluation parameters of a plurality of evaluation indicators; Collecting a plurality of risk processing data of the plurality of historical vehicle insurances on the target vehicle, wherein each risk processing data includes processing parameters of a plurality of types of processing indicators; The plurality of historical auto insurances, the plurality of user evaluation data and the plurality of risk processing data are integrated to obtain historical auto insurance data.

3. The method for matching auto insurance based on customer historical data and vehicle risk score according to claim 2, characterized in that: According to the plurality of user evaluation data and the plurality of vehicle risk processing data, an evaluation parameter matrix and a vehicle parameter matrix are constructed, and a plurality of user scores and a plurality of risk processing scores are obtained by analysis and calculation, including: Performing normalization and maximization processing on the evaluation parameters and processing parameters in the plurality of user evaluation data and the plurality of vehicle risk processing data to obtain a plurality of processing evaluation parameter sets and a plurality of processing vehicle parameter sets; According to the multiple processing evaluation parameter sets, an evaluation parameter matrix is ​​constructed as follows: Among them, A is the evaluation parameter matrix, m is the number of evaluation indicators in the evaluation parameter set, and n is the number of multiple historical auto insurances. is the processing evaluation parameter of the first evaluation index of the first historical auto insurance, is the processing evaluation parameter of the first evaluation index of the nth historical auto insurance, is the processing evaluation parameter of the mth evaluation index of the first historical auto insurance, is the processing evaluation parameter of the mth evaluation index of the nth historical auto insurance; Analyze and calculate to obtain multiple user ratings according to the evaluation parameter matrix; A vehicle parameter matrix is ​​constructed based on the multiple processing vehicle parameter sets, and multiple risk processing scores are obtained through analysis and calculation.

4. The method for matching auto insurance based on customer historical data and vehicle risk score according to claim 3, characterized in that: According to the evaluation parameter matrix, multiple user scores are obtained by analysis and calculation, as shown in the following formula: Among them, G i is the user rating of the i-th historical auto insurance, w j is the weight of the jth evaluation index, is the maximum value among the n processing evaluation parameters in the jth evaluation index in the evaluation parameter matrix, is the processing evaluation parameter of the jth evaluation index of the ith historical auto insurance in the evaluation parameter matrix, It is the minimum value among the n processing evaluation parameters in the j-th evaluation index in the evaluation parameter matrix.

5. The method for matching auto insurance based on customer historical data and vehicle risk score according to claim 1, characterized in that: According to the multiple combinations of the target vehicle and multiple historical auto insurances, a combination search and analysis is performed in the auto insurance database to obtain multiple combination specificities, including: Collecting multiple sample combinations, wherein the multiple sample combinations are obtained by traversing multiple sample vehicles and multiple sample auto insurance combinations; Obtaining multiple sample combination rates of the multiple sample combinations in the motor vehicle insurance database, and calculating an average combination rate; According to the multiple combinations of the target vehicle and multiple historical auto insurances, performing combination searches in the auto insurance database respectively to obtain multiple combination rates; A plurality of combination specificities are calculated based on the plurality of combination rates and the average combination rate.

6. The method for matching auto insurance based on customer historical data and vehicle risk score according to claim 1, characterized in that: According to the plurality of combined special degrees, a plurality of search quantities are set, and a plurality of combined sample auto insurance data sets are obtained by searching in the auto insurance database, including: Get the preset auxiliary analysis quantity; According to the multiple combined specificities, the preset auxiliary analysis quantity is adjusted and calculated to obtain multiple retrieval quantities; According to the multiple search quantities and multiple combinations, a plurality of combined sample auto insurance data sets that meet the multiple combinations are randomly searched in the auto insurance database, and each combined sample auto insurance data set includes sample user evaluation data and sample vehicle risk processing data.

7. The method for matching auto insurance based on customer historical data and vehicle risk score according to claim 1, characterized in that: According to the multiple combined sample auto insurance data sets, multiple auxiliary user score sets and multiple auxiliary risk processing score sets are obtained by analysis and calculation, including: constructing a plurality of sample evaluation parameter matrices and a plurality of sample vehicle parameter matrices respectively according to the plurality of combined sample auto insurance data sets; According to the multiple sample evaluation parameter matrices and the multiple sample vehicle parameter matrices, multiple auxiliary user score sets and multiple auxiliary risk management score sets are obtained through analysis and calculation.

8. The method for matching auto insurance based on customer historical data and vehicle risk score according to claim 1, characterized in that: According to the multiple combined special degrees, in combination with the multiple user scores, the multiple risk treatment scores, the multiple auxiliary user score sets and the multiple auxiliary risk treatment score sets, multiple comprehensive scores are calculated, including: Calculate and obtain multiple auxiliary user scores and multiple auxiliary risk management scores according to the multiple auxiliary user score sets and the multiple auxiliary risk management score sets; According to the plurality of combined special degrees, configuring a plurality of basic weights and a plurality of auxiliary weights; According to the multiple basic weights and the multiple auxiliary weights, the multiple user scores and the multiple auxiliary user scores are weighted and calculated to obtain multiple comprehensive user scores, and the multiple risk treatment scores and the multiple auxiliary risk treatment scores are weighted and calculated to obtain multiple comprehensive risk treatment scores; A plurality of comprehensive scores are calculated based on the plurality of comprehensive user scores and the plurality of comprehensive risk handling scores.

Images