Car insurance data processing method and device, equipment, storage medium and program product
The method improves auto insurance risk assessments by cleansing, converting, and dimensionally reducing data using chi-square and information gain models to train a feature engineering model, addressing the inaccuracy of traditional assessments.
Patent Information
- Application Number
- CN202411396669.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-08
- Publication Date
- 2025-01-03
AI Technical Summary
Traditional auto insurance assessment methods rely on subjective agent experience and historical vehicle claims data, leading to inaccurate risk assessments of policyholders.
A method involving data cleansing, conversion, encoding, screening, and dimensionality reduction using chi-square and information gain models to train a feature engineering model for improved risk assessment.
Enhances the accuracy of auto insurance risk assessments by filtering and verifying data to extract relevant features, resulting in more precise policyholder evaluations.
Smart Images
Figure CN119250988A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, device, equipment, storage medium and program product for processing auto insurance data. Background Art
[0002] As the number of cars continues to increase, users' demand for auto insurance is also gradually increasing. Traditional auto insurance business mainly relies on the experience of insurance agents to assess the risk of policyholders and insurance premiums, which is subjective and inaccurate.
[0003] In the existing technology, auto insurance companies use big data technology to collect and analyze policyholders' data information, conduct risk assessments on policyholders, and thus formulate insurance premiums.
[0004] However, the existing technical solutions only collect the vehicle's historical accident records when conducting risk assessment on the insured, and conduct risk assessment on the bidder based on the vehicle's historical accident records, resulting in inaccurate risk assessment of the insured by the auto insurance company. Summary of the invention
[0005] The embodiments of the present application provide a method, device, equipment, storage medium and program product for processing auto insurance data, which are used to solve the problem of inaccurate risk assessment of policyholders by auto insurance companies in the prior art.
[0006] In a first aspect, an embodiment of the present application provides a method for processing auto insurance data, which is applied to a computer device, comprising:
[0007] In response to the operation of the staff inputting the vehicle insurance data, obtaining the vehicle insurance data;
[0008] Determining the availability of the auto insurance data according to the characteristic dimensions of the auto insurance data;
[0009] If the availability of the vehicle insurance data meets the preset availability limit, the vehicle insurance data is cleansed to generate cleaned vehicle insurance data;
[0010] According to the data format of the feature engineering model, data conversion is performed on the cleaned auto insurance data to generate converted auto insurance data;
[0011] Performing data encoding on the converted auto insurance data to generate training data;
[0012] Screening the training data according to a relevance principle to obtain screened training data;
[0013] Performing dimensionality reduction processing on the screened training data to obtain dimensionality-reduced training data;
[0014] Verifying the reduced-dimensional training data according to a chi-square test model to generate a first verification result;
[0015] Verifying the reduced-dimensional training data according to the information gain model to generate a second verification result;
[0016] If the first verification result and the second verification result meet the preset limit, training the feature engineering model according to the dimensionally reduced training data to generate a trained feature engineering model;
[0017] In response to the operation of inputting the data to be evaluated by the staff, extracting features of the data to be evaluated according to the trained feature engineering model to generate a feature data set;
[0018] The feature data set is input into a risk assessment model to output risk assessment information of the policyholder.
[0019] In a possible implementation, the performing dimensionality reduction processing on the screened training data to obtain the reduced-dimensionality training data includes: obtaining the characteristic dimension of the screened training data; and performing dimensionality reduction processing on the characteristic dimension of the screened training data according to a dimensionality reduction model to obtain the reduced-dimensionality training data.
[0020] In a possible implementation, the dimension reduction model is:
[0021]
[0022] In the formula, represents the variance of the training data after dimensionality reduction; X and Y represent feature dimension data; and Represents the average value of feature dimension data; Represents the variance of feature dimension X; It represents the variance of feature dimension Y; Mean(X) represents the mean of feature dimension data X.
[0023] In a possible implementation, the chi-square test model is:
[0024]
[0025] Where X represents the chi-square statistic; A represents the training data after dimensionality reduction; T represents the training data before dimensionality reduction.
[0026] In a possible implementation, the information gain model is:
[0027] g(D,A)=H(D)-H(D|A)
[0028]
[0029] In the formula, g(D,A) represents information gain; D represents the training data after dimensionality reduction; A represents the application scenario of auto insurance data; H(D|A) represents the information entropy of the training data after dimensionality reduction in application scenario A; H(D) represents the information entropy of the training data after dimensionality reduction; C k represents the characteristic parameters in the training data after dimensionality reduction; W i represents the weight; i and k represent the number of features of the training data after dimensionality reduction.
[0030] In a possible implementation, if the first verification result and the second verification result meet the preset limit values, then after training the feature engineering model according to the reduced-dimensional training data to generate the trained feature engineering model, it also includes: configuring a model monitoring probe; verifying the feature data set output by the feature engineering model through the model monitoring probe; if the feature quality in the feature data set does not reach a preset feature quality threshold, obtaining feature data that does not reach the feature quality threshold; obtaining the corresponding feature weights in the feature engineering model according to the feature data that does not reach the feature quality threshold; and adjusting the corresponding feature weights in the feature engineering model to generate an adjusted feature engineering model.
[0031] In a second aspect, an embodiment of the present application provides a vehicle insurance data processing device, which is applied to a computer device, including:
[0032] A first acquisition module, configured to acquire the vehicle insurance data in response to the operation of inputting the vehicle insurance data by the staff;
[0033] A judgment module, used to judge the availability of the vehicle insurance data according to the characteristic dimensions of the vehicle insurance data;
[0034] A cleaning module, configured to clean the vehicle insurance data if the availability of the vehicle insurance data meets a preset availability limit, so as to generate cleaned vehicle insurance data;
[0035] A conversion module, used for performing data conversion on the cleaned auto insurance data according to the data format of the feature engineering model to generate converted auto insurance data;
[0036] An encoding module, used for performing data encoding on the converted auto insurance data to generate training data;
[0037] A screening module, used for screening the training data according to the relevance principle to obtain screened training data;
[0038] A dimension reduction module, used for performing dimension reduction processing on the screened training data to obtain dimension-reduced training data;
[0039] A first generating module is used to verify the reduced-dimensional training data according to a chi-square test model to generate a first verification result;
[0040] A second generating module is used to verify the reduced-dimensional training data according to the information gain model to generate a second verification result;
[0041] A training module, configured to train a feature engineering model according to the reduced-dimensional training data to generate a trained feature engineering model if the first verification result and the second verification result meet a preset limit;
[0042] An extraction module, configured to extract features of the data to be evaluated according to the trained feature engineering model in response to the operation of the staff inputting the data to be evaluated, so as to generate a feature data set;
[0043] The output module is used to input the feature data set into the risk assessment model to output the risk assessment information of the insured.
[0044] In a third aspect, an embodiment of the present application provides a vehicle insurance data processing device, including:
[0045] at least one processor and memory;
[0046] The memory stores computer-executable instructions;
[0047] The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the above first aspect and / or various possible implementations of the first aspect.
[0048] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the first aspect above and / or various possible implementations of the first aspect.
[0049] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the above first aspect and / or various possible implementation methods of the first aspect.
[0050] The motor vehicle insurance data processing method, device, equipment, storage medium and program product provided in the embodiments of the present application judge the availability of the motor vehicle insurance data. If the preset availability limit is met, the motor vehicle insurance data is cleaned and converted, the converted motor vehicle insurance data is encoded, training data is generated, the training data is screened according to the relevance principle, the screened training data is subjected to dimensionality reduction processing to remove noise and redundancy, and the reduced-dimensional training data is verified by a chi-square test model and an information gain model. If the reduced-dimensional training data meets the preset limit, a feature engineering model is trained with the reduced-dimensional training data, the data to be evaluated is input into the feature engineering model to output a feature data set, the feature data set is input into a risk assessment model to output the risk assessment information of the policyholder, and the feature engineering model is used to extract a feature data set representing user characteristics, thereby improving the accuracy of the motor vehicle insurance company's risk assessment of the policyholder. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.
[0052] Figure 1 A schematic diagram of the system structure of a computer device provided in an embodiment of the present application;
[0053] Figure 2 A flowchart of the method for processing auto insurance data provided in this application;
[0054] Figure 3 A schematic diagram of correlation analysis of different feature dimensions provided in an embodiment of the present application;
[0055] Figure 4 A schematic diagram of the structure of the vehicle insurance data processing device provided in this application;
[0056] Figure 5 This is a schematic diagram of the structure of the vehicle insurance data processing device provided in this application.
[0057] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION
[0058] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.
[0059] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards, and provide corresponding operation entrances for users to choose to authorize or refuse.
[0060] First, the terms involved in this application are explained:
[0061] Feature engineering: refers to the process of processing and transforming raw data in machine learning tasks to extract features that are more representative of the problem. In practical applications, raw data contains a lot of useless information, noise, and redundant features, which will have a negative impact on the performance of machine learning algorithms. The goal of feature engineering is to select, create, or transform features from raw data so that the extracted features are more representative and interpretable, and can better reflect the essence of the problem.
[0062] One-hot encoding: One-hot encoding, also known as one-bit effective encoding, uses an N-bit state register to encode N states. Each state has its own independent register bit, and at any time, only one of them is valid.
[0063] As the number of cars continues to increase, users' demand for auto insurance is also gradually increasing. Traditional auto insurance business mainly relies on the experience and judgment of insurance agents to assess the risks and insurance premiums of policyholders, which is subjective and inaccurate. In the prior art, auto insurance companies use big data technology to collect and analyze policyholders' data information, conduct risk assessments on policyholders, and thus formulate insurance premiums. However, when conducting risk assessments on policyholders, the existing technical solutions only collect the historical claims of the vehicle, and conduct risk assessments on bidders based on the historical claims of the vehicle, resulting in inaccurate risk assessments of policyholders by auto insurance companies.
[0064] In order to solve the above technical problems, the embodiment of the present application proposes the following technical concept: the inventor considers the availability of the auto insurance data input by the judgment staff, if the availability of the auto insurance data meets the preset availability limit, the auto insurance data is cleaned, the cleaned auto insurance data is converted according to the data format of the feature engineering model, the converted auto insurance data is generated, and the training data is generated by encoding the converted auto insurance data; the training data is screened and dimensionally reduced according to the correlation principle to obtain the reduced dimension training data, considering the chi-square test model and the information gain model to verify whether the reduced dimension training data meets the preset limit, if the limit is met, the reduced dimension training data is used to train the feature engineering model, the trained feature engineering outputs the feature data set, the feature data set is input into the risk assessment model, and the risk assessment information of the insured is output. The following is a detailed description using a detailed embodiment.
[0065] Figure 1 A schematic diagram of the system structure of a computer device provided in an embodiment of the present application. Figure 1 As shown, the computer device includes: a receiving device 101, a processor 102 and a display device 103.
[0066] It is understandable that the structure illustrated in the embodiment of the present application does not constitute a specific limitation on the object identification method. In other feasible implementations of the present application, the above architecture may include more or fewer components than shown in the figure, or combine certain components, or split certain components, or arrange the components differently, which can be determined according to the actual application scenario and is not limited here. Figure 1 The components shown may be implemented in hardware, software, or a combination of software and hardware.
[0067] In a specific implementation process, the receiving device 101 may be an input / output interface or a communication interface, and may acquire the vehicle insurance data.
[0068] Processor 102 can generate a trained feature engineering model and output risk assessment information of the insured.
[0069] The display device 103 can be used to display the risk assessment information of the above-mentioned policyholder.
[0070] The display device may also be a touch display screen, which is used to receive user instructions while displaying the above-mentioned content, so as to realize operational interaction with the user.
[0071] It should be understood that the above-mentioned processor can be implemented by the processor reading instructions in the memory and executing the instructions, or it can be implemented by a chip circuit.
[0072] In addition, the network architecture and business scenarios described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Ordinary technicians in this field can know that with the evolution of network architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.
[0073] Figure 2 This is a flow chart of the method for processing auto insurance data provided by this application. The execution subject of this embodiment can be a computer device, and this embodiment is not particularly limited here. Figure 2 As shown, the method includes:
[0074] S201: In response to a worker inputting the vehicle insurance data, the vehicle insurance data is obtained.
[0075] In this embodiment, the vehicle insurance data is vehicle insurance characteristic data.
[0076] In this embodiment, the auto insurance data includes but is not limited to user data, vehicle data, auto insurance application and claims data.
[0077] User data includes, but is not limited to, user age, individual or group user, and user identification.
[0078] Among them, vehicle data includes but is not limited to vehicle identification, whether it is a flooded vehicle, whether it is on the blacklist or gray list, vehicle attributes, vehicle usage nature, vehicle type, number of seats, tonnage, manufacturer, brand code, car series code, model name, purchase price, actual value, vehicle age, displacement, fuel type, power, number of airbags and whether it is a non-local vehicle, etc.
[0079] Among them, the auto insurance bidding and claims data include but are not limited to the compulsory traffic insurance claims adjustment coefficient, underwriting institution, commercial insurance NCD discount coefficient, compulsory traffic insurance premium, whole order risk premium, the compulsory traffic insurance platform return value of the previous year, the compulsory traffic insurance claims adjustment coefficient of the previous year, the amount of compensation paid by the compulsory traffic insurance platform in the previous year, the reason for the non-floating of commercial insurance in the previous year, the level of the commercial insurance worry-free coefficient in the previous year, whether there were any night accidents in the previous year, whether there were any zero case settlements in the previous year, the commercial insurance premium of the previous year, the amount of commercial insurance compensation paid in the previous year, the number of compulsory traffic insurance claims in the previous year, the number of commercial insurance claims in the previous year and the commercial insurance insured group in the previous year.
[0080] S202: Determine the availability of the auto insurance data based on the characteristic dimensions of the auto insurance data.
[0081] In this embodiment, the characteristic dimensions of the auto insurance data include, but are not limited to, the coverage rate of the auto insurance data, the accuracy rate of the auto insurance data, and the data source of the auto insurance data.
[0082] S203: If the availability of the vehicle insurance data meets the preset availability limit, the vehicle insurance data is cleansed to generate cleansed vehicle insurance data.
[0083] In this embodiment, data cleaning methods include but are not limited to null value deletion, outlier removal, and missing value supplementation.
[0084] For example, if the vehicle manufacturer data in the auto insurance data is a null value, the vehicle manufacturer item in the auto insurance data is deleted; if the data appearing in the car power characteristic item is "manual transmission", the abnormal value appearing in the car power characteristic item is eliminated; for data items that are missing NAN floating-point type data, the data value is recorded as 0.
[0085] S204: Performing data conversion on the cleaned auto insurance data according to the data format of the feature engineering model to generate converted auto insurance data.
[0086] In this embodiment, the data format of the feature engineering model includes but is not limited to a numerical data format, a categorical data format, and a text data format.
[0087] In this embodiment, the data conversion method includes but is not limited to desensitizing the auto insurance data, grouping feature values, and scaling the features.
[0088] Among them, the desensitization methods include but are not limited to encryption desensitization, data perturbation desensitization and data shielding desensitization.
[0089] Exemplarily, the method for grouping characteristic values is: grouping user age characteristic data, setting different characteristic values according to different ages, setting the characteristic value for those under 18 years old to 1, the characteristic value for those between 18 and 30 years old to 2, the characteristic value for those between 30 and 50 years old to 3, the characteristic value for those between 50 and 60 years old to 4, and the characteristic value for those over 60 years old to 5.
[0090] S205: Encode the converted auto insurance data to generate training data.
[0091] In this embodiment, the data encoding method includes but is not limited to serial number encoding, one-hot encoding and binary encoding.
[0092] In this embodiment, the data encoded in the motor vehicle insurance data includes: new vehicle purchase price, vehicle age, displacement, tonnage grouping, whole-order risk premium, vehicle type and fuel type.
[0093] S206: Filter the training data according to the relevance principle to obtain filtered training data.
[0094] In this embodiment, the correlation principle is: for feature dimensions with high correlation, only one feature dimension is added to the calculation of the feature engineering model.
[0095] Figure 3 A schematic diagram of correlation analysis of different feature dimensions provided in an embodiment of the present application.
[0096] like Figure 3 As shown, Figure 3 Each pixel in the middle color area is a data point corresponding to each feature dimension in the auto insurance data. The depth of the color represents the correlation of the dimensions. The darker the color, the stronger the correlation.
[0097] Specifically, through the principle of correlation, strongly correlated data in each feature dimension data is screened to reduce information redundancy.
[0098] S207: Perform dimensionality reduction processing on the screened training data to obtain dimensionality reduced training data.
[0099] Specifically, the dimension reduction model is used to reduce the feature dimensions of the screened training data to obtain the reduced-dimensional training data.
[0100] S208: Validate the reduced-dimensional training data according to the chi-square test model to generate a first validation result.
[0101] In this embodiment, the chi-square statistics before and after dimensionality reduction are calculated using a chi-square test model to determine whether the chi-square statistics before and after the calculation are within a preset limit range.
[0102] S209: Verify the reduced-dimensional training data according to the information gain model to generate a second verification result.
[0103] In this embodiment, the information gain in different application scenarios is calculated by using an information gain model to determine whether the data after dimensionality reduction is within a normal range in different application scenarios.
[0104] S210: If the first verification result and the second verification result meet the preset limit, the feature engineering model is trained according to the dimensionally reduced training data to generate a trained feature engineering model.
[0105] Specifically, if the training data after dimensionality reduction meets the preset limit, the feature dimension in the training data after dimensionality reduction is obtained, and the weight of the feature engineering model is configured according to the feature dimension.
[0106] S211: In response to the staff inputting the data to be evaluated, feature extraction is performed on the data to be evaluated according to the trained feature engineering model to generate a feature data set.
[0107] Specifically, the data in the feature data set is converted according to the data format of the risk assessment model to generate a feature data set with a unified format.
[0108] S212: Input the feature data set into the risk assessment model to output the risk assessment information of the policyholder.
[0109] Specifically, the feature data set is input into the risk assessment model, and the risk assessment model calculates the risk assessment value of the policyholder based on the risk characteristics of each dimension in the feature data set.
[0110] It can be seen from the above embodiments that by judging the availability of the auto insurance data, if the preset availability limit is met, the auto insurance data is cleaned and converted, the converted auto insurance data is encoded, training data is generated, the training data is screened according to the relevance principle, the screened training data is reduced in dimension to remove noise and redundancy, and the reduced-dimensional training data is verified by the chi-square test model and the information gain model. If the reduced-dimensional training data meets the preset limit, the feature engineering model is trained with the reduced-dimensional training data, the data to be evaluated is input into the feature engineering model to output a feature data set, the feature data set is input into the risk assessment model to output the risk assessment information of the policyholder, and the feature engineering model is used to extract the feature data set that characterizes the user characteristics, thereby improving the accuracy of the auto insurance company's risk assessment of the policyholder.
[0111] In one embodiment of the present application, step S207 includes:
[0112] S2071: Obtain feature dimensions of the filtered training data.
[0113] In this embodiment, the feature dimensions of the screened training data include, but are not limited to, the individual group nature of the policyholder, the commercial insurance NCD discount coefficient, the insured person identification and the actual value.
[0114] S2072: Perform dimensionality reduction processing on the feature dimensions of the filtered training data according to the dimensionality reduction model to obtain dimensionality-reduced training data.
[0115] In this embodiment, the dimensionality reduction processing method includes but is not limited to PCA (principal components analysis, PCA for short), LDA (linear discriminant analysis, LDA for short) and ICA (independent component analysis, ICA for short) independent component analysis.
[0116] In this embodiment, the dimensionality reduction model is:
[0117]
[0118] In the formula, represents the training data after dimensionality reduction; X and Y represent the feature dimension data; and Represents the average value of feature dimension data; Represents the variance of feature dimension X; It represents the variance of feature dimension Y; Mean(X) represents the mean of feature dimension data X.
[0119] Specifically, the variance of feature dimension data X and feature dimension data Y is calculated, and the feature dimension is reduced from two dimensions to one dimension through data dimensionality reduction, and the variance of the training data after dimensionality reduction is guaranteed. Try to be close to the variance of the feature dimension data before dimensionality reduction and The sum.
[0120] It can be seen from the above embodiments that the dimension reduction model is used to reduce the dimension of the screened training data, thereby removing useless noise in the training data and improving the efficiency and accuracy of the feature engineering model calculation.
[0121] In one embodiment of the present application, the chi-square test model is:
[0122]
[0123] Where X represents the chi-square statistic; A represents the training data after dimensionality reduction; T represents the training data before dimensionality reduction.
[0124] In this embodiment, the chi-square statistic represents the degree of deviation between the training data before dimensionality reduction and the training data after dimensionality reduction. The larger the chi-square statistic, the greater the degree of deviation between the training data after dimensionality reduction and the training data before dimensionality reduction.
[0125] Specifically, the accuracy of dimensionality reduction is verified based on the chi-square statistic before and after dimensionality reduction. If the chi-square statistic exceeds the preset limit, the training data is re-reduced.
[0126] It can be seen from the above embodiments that by calculating the chi-square statistic before and after dimensionality reduction, verifying the accuracy of data dimensionality reduction according to the chi-square statistic, and adjusting the data dimensionality reduction method according to the chi-square statistic, the accuracy of the training data after dimensionality reduction is improved.
[0127] In one embodiment of the present application, the information gain model is:
[0128] g(D,A)=H(D)-H(D|A)
[0129]
[0130] In the formula, g(D,A) represents information gain; D represents the training data after dimensionality reduction; A represents the application scenario of auto insurance data; H(D|A) represents the information entropy of the training data after dimensionality reduction in application scenario A; H(D) represents the information entropy of the training data after dimensionality reduction; C k represents the characteristic parameters in the training data after dimensionality reduction; W i represents the weight; i and k represent the number of features of the training data after dimensionality reduction.
[0131] Specifically, according to different application scenarios of auto insurance users, the fluctuation of feature data in different application scenarios is calculated, and based on the calculated information gain, it is determined whether the data after dimensionality reduction is within a normal range.
[0132] It can be seen from the above embodiments that by calculating information gain, verifying the accuracy of data dimensionality reduction based on the information gain, and obtaining the fluctuation of feature data based on the calculated information gain, and adjusting the auto insurance plan based on the data fluctuation, the accuracy of risk assessment of the insured is improved.
[0133] In one embodiment of the present application, after step S210, the method further includes:
[0134] S213: Configure the model monitoring probe.
[0135] Specifically, characteristic parameters in the feature engineering model are obtained, the monitoring probe operating environment is configured, different monitoring probes are set according to different characteristic parameters, and the monitoring threshold of each characteristic parameter is set.
[0136] S214: Verify the feature data set output by the feature engineering model through the model monitoring probe.
[0137] Specifically, according to each characteristic parameter in the characteristic data set, each output characteristic parameter is compared with the corresponding monitoring threshold value.
[0138] S215: If the feature quality in the feature data set does not reach a preset feature quality threshold, then acquiring feature data that does not reach the feature quality threshold.
[0139] Specifically, it is determined whether the feature quality of each feature parameter in the feature data set meets the feature quality threshold. If the feature quality does not reach the feature threshold, feature data that does not reach the feature quality threshold is obtained.
[0140] S216: Obtain corresponding feature weights in the feature engineering model based on feature data that does not reach the feature quality threshold.
[0141] In this embodiment, the higher the feature quality of the customized data, the higher the feature weight corresponding to it.
[0142] S217: Adjust the corresponding feature weights in the feature engineering model to generate an adjusted feature engineering model.
[0143] Specifically, the feature weight corresponding to the feature data that does not reach the feature quality threshold is reduced, and the data dimension reduction operation is re-performed on the feature data that does not reach the feature quality threshold.
[0144] It can be seen from the above embodiments that by configuring the model monitoring probe to monitor the feature quality of the feature data output by the feature engineering model, if the feature quality does not reach the feature quality threshold, the corresponding feature weight in the feature engineering model is adjusted, thereby avoiding the problem of reduced feature output quality effect caused by the decline in feature quality.
[0145] Figure 4 A schematic diagram of the structure of the vehicle insurance data processing device provided in this application, such as Figure 4 As shown, the auto insurance data processing device 40 provided in this embodiment includes: a first acquisition module 401, a judgment module 402, a cleaning module 403, a conversion module 404, an encoding module 405, a screening module 406, a dimensionality reduction module 407, a first generation module 408, a second generation module 409, a training module 410, an extraction module 411 and an output module 412.
[0146] The first acquisition module 401 is used to acquire the vehicle insurance data in response to the operation of the staff inputting the vehicle insurance data.
[0147] The judgment module 402 is used to judge the availability of the vehicle insurance data according to the characteristic dimensions of the vehicle insurance data.
[0148] The cleaning module 403 is used to clean the vehicle insurance data if the availability of the vehicle insurance data meets a preset availability limit, so as to generate cleaned vehicle insurance data.
[0149] The conversion module 404 is used to perform data conversion on the cleaned auto insurance data according to the data format of the feature engineering model to generate converted auto insurance data.
[0150] The encoding module 405 is used to perform data encoding on the converted vehicle insurance data to generate training data.
[0151] The screening module 406 is used to screen the training data according to the relevance principle to obtain screened training data.
[0152] The dimension reduction module 407 is used to perform dimension reduction processing on the screened training data to obtain dimension-reduced training data.
[0153] The first generating module 408 is used to verify the dimension-reduced training data according to the chi-square test model to generate a first verification result.
[0154] The second generating module 409 is used to verify the dimension-reduced training data according to the information gain model to generate a second verification result.
[0155] The training module 410 is used to train the feature engineering model according to the reduced-dimensional training data to generate a trained feature engineering model if the first verification result and the second verification result meet the preset limit.
[0156] The extraction module 411 is used to extract features of the data to be evaluated according to the trained feature engineering model in response to the operation of the staff inputting the data to be evaluated, so as to generate a feature data set.
[0157] The output module 412 is used to input the feature data set into the risk assessment model to output the risk assessment information of the insured.
[0158] In a possible implementation, the dimension reduction module 407 includes:
[0159] The acquisition unit 4071 is used to acquire the feature dimension of the filtered training data.
[0160] The dimension reduction unit 4072 is used to perform dimension reduction processing on the feature dimensions of the filtered training data according to the dimension reduction model to obtain the training data after dimension reduction.
[0161] In a possible implementation, the dimension reduction model in the dimension reduction unit 4072 is:
[0162]
[0163]
[0164] In the formula, represents the variance of the training data after dimensionality reduction; X and Y represent feature dimension data; and Represents the average value of feature dimension data; Represents the variance of feature dimension X; It represents the variance of feature dimension Y; Mean(X) represents the mean of feature dimension data X.
[0165] In a possible implementation, the chi-square test model in the first generation module 408 is:
[0166]
[0167] Where X represents the chi-square statistic; A represents the training data after dimensionality reduction; T represents the training data before dimensionality reduction.
[0168] In a possible implementation, the information gain model in the second generation module 409 is:
[0169] g(D,A)=H(D)-H(D|A)
[0170]
[0171] In the formula, g(D,A) represents information gain; D represents the training data after dimensionality reduction; A represents the application scenario of auto insurance data; H(D|A) represents the information entropy of the training data after dimensionality reduction in application scenario A; H(D) represents the information entropy of the training data after dimensionality reduction; C k represents the characteristic parameters in the training data after dimensionality reduction; W i represents the weight; i and k represent the number of features of the training data after dimensionality reduction.
[0172] In a possible implementation manner, the vehicle insurance data processing device 40 further includes:
[0173] The configuration module 413 is used to configure the model monitoring probe.
[0174] The verification module 414 is used to verify the feature data set output by the feature engineering model through the model monitoring probe.
[0175] The second acquisition module 415 is configured to acquire feature data that does not reach the feature quality threshold if the feature quality in the feature data set does not reach the preset feature quality threshold.
[0176] The third acquisition module 416 is used to acquire the corresponding feature weight in the feature engineering model according to the feature data that does not reach the feature quality threshold.
[0177] The adjustment module 417 is used to adjust the corresponding feature weights in the feature engineering model to generate an adjusted feature engineering model.
[0178] The vehicle insurance data processing device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and this embodiment will not be described in detail here.
[0179] Figure 5 This is a schematic diagram of the structure of the vehicle insurance data processing device provided in this application. Figure 5 As shown, the vehicle insurance data processing device 50 provided in this embodiment includes: at least one processor 501 and a memory 502. Optionally, the device 50 also includes a communication component 503. The processor 501, the memory 502 and the communication component 503 are connected via a bus 504.
[0180] In the specific implementation process, at least one processor 501 executes the computer execution instructions stored in the memory 502, so that at least one processor 501 executes the above-mentioned vehicle insurance data processing method.
[0181] The specific implementation process of the processor 501 can be found in the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.
[0182] In the above embodiments, it should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the invention may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.
[0183] The memory may include a high-speed memory (Random Access Memory, RAM), and may also include a non-volatile memory (Non-volatile Memory, NVM), such as at least one disk memory.
[0184] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of the present application is not limited to only one bus or one type of bus.
[0185] The present application also provides a computer program product, including a computer program, which implements the above method when executed by a processor.
[0186] The present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the above method is implemented.
[0187] The above-mentioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.
[0188] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC). Of course, the processor and the readable storage medium can also exist in the device as discrete components.
[0189] The division of units is only a logical function division, and there may be other divisions in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.
[0190] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0191] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0192] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.
[0193] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.
[0194] Finally, it should be noted that those skilled in the art will readily conceive of other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary technical means in the art not disclosed by the present invention, are not limited to the precise structure described above and shown in the drawings, and may be modified and changed in various ways without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.
Claims
1. A method for processing auto insurance data, characterized in that: Applicable to computer equipment, including: In response to the operation of the staff inputting the vehicle insurance data, obtaining the vehicle insurance data; Determining the availability of the auto insurance data according to the characteristic dimensions of the auto insurance data; If the availability of the vehicle insurance data meets the preset availability limit, the vehicle insurance data is cleansed to generate cleaned vehicle insurance data; According to the data format of the feature engineering model, data conversion is performed on the cleaned auto insurance data to generate converted auto insurance data; Performing data encoding on the converted auto insurance data to generate training data; Screening the training data according to a relevance principle to obtain screened training data; Performing dimensionality reduction processing on the screened training data to obtain dimensionality-reduced training data; Verifying the reduced-dimensional training data according to a chi-square test model to generate a first verification result; Verifying the reduced-dimensional training data according to the information gain model to generate a second verification result; If the first verification result and the second verification result meet the preset limit, training the feature engineering model according to the dimensionally reduced training data to generate a trained feature engineering model; In response to the operation of inputting the data to be evaluated by the staff, extracting features of the data to be evaluated according to the trained feature engineering model to generate a feature data set; The feature data set is input into a risk assessment model to output risk assessment information of the policyholder.
2. The method according to claim 1, characterized in that The performing dimensionality reduction processing on the screened training data to obtain dimensionality-reduced training data includes: Obtaining feature dimensions of the filtered training data; The feature dimensions of the screened training data are reduced in dimension according to the dimensionality reduction model to obtain training data after dimensionality reduction.
3. The method according to claim 2, characterized in that The dimension reduction model is: In the formula, represents the variance of the training data after dimensionality reduction; X and Y represent feature dimension data; and Represents the average value of feature dimension data; Represents the variance of feature dimension X; It represents the variance of feature dimension Y; Mean(X) represents the mean of feature dimension data X.
4. The method according to claim 1, characterized in that The chi-square test model is: Where X represents the chi-square statistic; A represents the training data after dimensionality reduction; T represents the training data before dimensionality reduction.
5. The method according to claim 1, characterized in that The information gain model is: g(D,A)=H(D)-H(D|A) In the formula, g(D,A) represents information gain; D represents the training data after dimensionality reduction; A represents the application scenario of auto insurance data; H(D|A) represents the information entropy of the training data after dimensionality reduction in application scenario A; H(D) represents the information entropy of the training data after dimensionality reduction; C k represents the characteristic parameters in the training data after dimensionality reduction; W i represents the weight; i and k represent the number of features of the training data after dimensionality reduction.
6. The method according to any one of claims 1 to 5, characterized in that: If the first verification result and the second verification result meet the preset limit, training the feature engineering model according to the dimension-reduced training data to generate a trained feature engineering model, further comprising: Configure model monitoring probes; Verifying the feature data set output by the feature engineering model through the model monitoring probe; If the feature quality in the feature data set does not reach a preset feature quality threshold, acquiring feature data that does not reach the feature quality threshold; Acquire a corresponding feature weight in the feature engineering model according to the feature data that does not reach the feature quality threshold; The corresponding feature weights in the feature engineering model are adjusted to generate an adjusted feature engineering model.
7. A vehicle insurance data processing device, characterized in that: Applicable to computer equipment, including: A first acquisition module, configured to acquire the vehicle insurance data in response to the operation of inputting the vehicle insurance data by the staff; A judgment module, used to judge the availability of the vehicle insurance data according to the characteristic dimensions of the vehicle insurance data; A cleaning module, configured to clean the vehicle insurance data if the availability of the vehicle insurance data meets a preset availability limit, so as to generate cleaned vehicle insurance data; A conversion module, used for performing data conversion on the cleaned auto insurance data according to the data format of the feature engineering model to generate converted auto insurance data; An encoding module, used for performing data encoding on the converted auto insurance data to generate training data; A screening module, used for screening the training data according to the relevance principle to obtain screened training data; A dimension reduction module, used for performing dimension reduction processing on the screened training data to obtain dimension-reduced training data; A first generating module, used to verify the reduced-dimensional training data according to a chi-square test model to generate a first verification result; A second generating module is used to verify the dimension-reduced training data according to the information gain model to generate a second verification result; A training module, configured to train a feature engineering model according to the reduced-dimensional training data to generate a trained feature engineering model if the first verification result and the second verification result meet a preset limit; An extraction module, configured to extract features of the data to be evaluated according to the trained feature engineering model in response to the operation of the staff inputting the data to be evaluated, so as to generate a feature data set; The output module is used to input the feature data set into the risk assessment model to output the risk assessment information of the insured.
8. A vehicle insurance data processing device, characterized in that: include: at least one processor and memory; The memory stores computer-executable instructions; The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the vehicle insurance data processing method as described in any one of claims 1 to 6.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the vehicle insurance data processing method according to any one of claims 1 to 6.
10. A computer program product, characterized in that It comprises a computer program, which, when executed by a processor, implements the vehicle insurance data processing method as claimed in any one of claims 1 to 6.