Business adjustment method and device, equipment and storage medium thereof
By organizing the historical business data of target users in a time series and applying risk prediction models, risk values are identified and adjusted, solving the problem of low risk prediction accuracy in auto insurance underwriting scenarios and achieving more intelligent and accurate risk avoidance.
Patent Information
- Application Number
- CN202510833550.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-19
- Publication Date
- 2025-11-04
AI Technical Summary
In the context of auto insurance underwriting in financial applications, existing large-scale pre-trained models are prone to overlooking a few types of fraud claims when data is imbalanced, resulting in low accuracy in risk prediction and hindering business risk avoidance and adjustment.
By acquiring historical business data from target users, processing it in a time series, and inputting it into a pre-trained business risk prediction model, the relationship between predicted risk values and preset risk thresholds is identified, and business adjustment measures are selected based on this relationship.
It improves the intelligence and accuracy of business risk prediction, enabling timely business adjustments for risky users and mitigating business risks.
Smart Images
Figure CN120894142A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of artificial intelligence, and is applied to a scenario of adjusting a business contract or a business agreement, and relates to a business adjustment method and device, equipment and a storage medium thereof. BACKGROUND
[0002] In the financial application industry, many risk control models have begun to use deep learning methods (such as large-scale pre-training models, deep neural networks, etc.) for risk assessment and fraud detection. These models can generally capture complex nonlinear relationships through big data training, thereby effectively identifying abnormal behavior and potential risks.
[0003] Although the application of large models performs well in prediction accuracy, in the context of car insurance underwriting in financial applications, there is still a serious problem of data imbalance due to the lack of claim data and fraud data. The above-mentioned large models tend to make biased predictions for the majority class (normal claim data) and ignore the minority class (fraud claim data), and cannot capture long-term potential risk signals in prediction, resulting in low accuracy of risk prediction, which is not conducive to related parties to avoid business risks and adjust business. SUMMARY
[0004] The purpose of the embodiments of the present application is to provide a business adjustment method, device, equipment and storage medium thereof to solve the problem of low accuracy of risk prediction when predicting business risks in related financial application scenarios, which is not conducive to related parties to avoid business risks and adjust business.
[0005] In a first aspect, the embodiments of the present application provide a business adjustment method, which adopts the technical solution as follows:
[0006] A business adjustment method includes the following steps:
[0007] Obtaining historical business data of a target user;
[0008] Time-series sorting the historical business data to obtain time-series historical business data;
[0009] Inputting the time-series historical business data into a pre-trained business risk prediction model;
[0010] Obtaining a predicted risk value output by the pre-trained business risk prediction model;
[0011] By comparison, identifying the relationship between the predicted risk value and a preset risk threshold;
[0012] According to the relationship, filtering different business adjustment measures to adjust the business of the target user.
[0013] In a second aspect, the embodiments of the present application further provide a service adjustment apparatus, which adopts the technical scheme as follows:
[0014] The service adjustment apparatus comprises:
[0015] a historical service data acquisition module, configured to acquire historical service data of a target user;
[0016] a time-series arrangement module, configured to time-series arrange the historical service data to acquire time-series historical service data;
[0017] a risk prediction model input module, configured to input the time-series historical service data into a pre-trained service risk prediction model;
[0018] a predicted risk value acquisition module, configured to acquire a predicted risk value output by the pre-trained service risk prediction model;
[0019] a comparison and identification module, configured to identify a relationship between the predicted risk value and a preset risk threshold through comparison;
[0020] a service adjustment module, configured to filter different service adjustment measures to adjust the service of the target user according to the relationship.
[0021] In a third aspect, the embodiments of the present application further provide a computer device, which adopts the technical scheme as follows:
[0022] The computer device comprises a memory and a processor, the memory stores computer readable instructions, and the processor executes the computer readable instructions to realize the steps of the service adjustment method.
[0023] In a fourth aspect, the embodiments of the present application further provide a computer readable storage medium, which adopts the technical scheme as follows:
[0024] The computer readable storage medium stores computer readable instructions, and the computer readable instructions are executed by a processor to realize the steps of the service adjustment method.
[0025] Compared with the prior art, the embodiments of the present application have the following beneficial effects:
[0026] The business adjustment method described in the application, by acquiring the historical business data of the target user; the historical business data is time-sequenced and sorted to obtain time-sequenced historical business data; the time-sequenced historical business data is input into the pre-trained business risk prediction model; the prediction risk value output by the pre-trained business risk prediction model is obtained; by comparison, the relationship between the prediction risk value and the preset risk threshold is identified; according to the relationship, different business adjustment measures are selected to adjust the business of the target user. The business adjustment method is applied to the financial business risk prediction scene, such as investment risk prediction, financial risk prediction, and car insurance underwriting risk prediction, etc. By using the artificial intelligence prediction model, and combining the historical business data of the user to predict the business risk value, the business risk value is predicted more intelligently and accurately, which is also convenient for related parties to avoid certain business risks and timely adjust the business for risk users. BRIEF DESCRIPTION OF DRAWINGS
[0027] In order to more clearly illustrate the scheme in the application, the drawings needed in the description of the embodiments of the application will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the application, and those skilled in the art can also obtain other drawings according to these drawings without creating labor.
[0028] Figure 1 is an exemplary system architecture diagram to which the application can be applied;
[0029] Figure 2 is a flowchart of an embodiment of a business adjustment method according to the application;
[0030] Figure 3 is a flowchart of a specific embodiment of the pre-training of the business risk prediction model in the business adjustment method described in the application;
[0031] Figure 4 is a flowchart of a specific embodiment of the expansion of the historical risk business data in the business adjustment method described in the application;
[0032] Figure 5 is Figure 4 is a flowchart of a specific embodiment of the step 402 shown in the figure;
[0033] Figure 6 is a flowchart of a specific embodiment of the termination control and expansion learning training in the business adjustment method described in the application;
[0034] Figure 7 is a structural schematic diagram of an embodiment of a business adjustment device according to the application;
[0035] Figure 8 is a structural schematic diagram of one embodiment of the computer device according to the present application. DETAILED DESCRIPTION
[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting of the application; the use herein of terms such as "comprise" and "have" and any variations thereof are intended to cover a non-exclusive inclusion; the use herein of terms such as "first", "second" and the like are intended to distinguish between similar objects unless the context indicates otherwise.
[0037] Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase that the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily mutually exclusive of one another. It is expressly understood that the embodiments described herein are merely examples from a
[0038] In order to make the technical personnel in the art better understand the scheme of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings.
[0039] As shown in Figure 1 The system architecture 100 can include a terminal device 101, a network 102 and a server 103, and the terminal device 101 can be a notebook computer 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is a medium for providing a communication link between the terminal device 101 and the server 103. The network 102 can include various connection types, such as wired, wireless communication links or optical fiber cables, etc.
[0040] A user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.
[0041] The terminal device 101 can be various electronic devices with a display screen and supporting web browsing, in addition to the notebook computer 1011, the tablet computer 1012 or the mobile phone 1013, the terminal device 101 can also be an electronic book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 player (Moving Picture Experts Group Audio Layer IV), a laptop computer and a desktop computer, etc.
[0042] The server 103 can be a server providing various services, for example, a background server providing support for a page displayed on the terminal device 101.
[0043] It should be noted that the business adjustment method provided by the embodiments of the present application is generally executed by a server, and accordingly, a business adjustment device is generally arranged in a server.
[0044] It should be understood that Figure 1 The number of terminal devices, networks and servers in
[0045] With reference to Figure 2 , a flow chart of one embodiment of a business adjustment method according to the present application is shown. The business adjustment method comprises the following steps:
[0046] Step 201, obtaining historical business data of a target user.
[0047] In the embodiment, the target user is, for example, a vehicle owner user, and the corresponding historical business data includes historical vehicle insurance purchase data and historical vehicle insurance claim data, etc.
[0048] By obtaining the historical business data of the target user, the historical business status of the target user is analyzed, specifically, by obtaining the historical vehicle insurance purchase data and the historical vehicle insurance claim data of the vehicle owner user, the vehicle insurance policy and the claim situation of the vehicle owner user are comprehensively analyzed, and the claim risk prediction of the underwriting institution is facilitated.
[0049] Step 202, time-series processing of the historical business data is performed to obtain time-series historical business data.
[0050] Specifically, before the step of performing the chronological arrangement on the historical business data and obtaining the chronological historical business data, the method further comprises: performing data cleaning, missing value supplementing and data format unification processing on the historical business data, so as to ensure the integrity of the relevant historical business data during the chronological arrangement.
[0051] By performing the chronological arrangement on the historical business data and obtaining the chronological historical business data, the subsequent business risk value prediction on the chronological dynamic trend can be facilitated, for example, the historical car insurance purchase data and the historical car insurance claim data of a car owner are arranged according to the insurance year and the claim year, so that when the car insurance claim risk of the car owner is predicted, the claim risk value can be predicted according to the specific changed insurance and claim data, and the accuracy of the final business risk value prediction is improved.
[0052] Step 203: inputting the chronological historical business data into the pre-trained business risk prediction model.
[0053] Step 204: obtaining the predicted risk value output by the pre-trained business risk prediction model.
[0054] By inputting the chronological historical business data into the pre-trained business risk prediction model and obtaining the predicted risk value output by the pre-trained business risk prediction model, the prediction is more intelligent and automatic compared with the previous artificial experience prediction, and the resource consumption during the human prediction is saved.
[0055] Step 205: identifying the relationship between the predicted risk value and the preset risk threshold value by comparison.
[0056] Specifically, for example, the preset risk threshold value is a fixed numerical value, the predicted risk value and the preset risk threshold value are input into a preset comparison function as comparison parameters, and the comparison function is executed to determine the size relationship between the predicted risk value and the preset risk threshold value.
[0057] For another example, the preset risk threshold value can also be replaced by a hierarchical interval value to determine the hierarchical interval to which the predicted risk value belongs, and the preset risk threshold value can also be replaced by a risk level, each risk level corresponding to a different risk value interval range, and the risk level to which the predicted risk value belongs is determined.
[0058] Step 206: according to the relationship, screening different business adjustment measures to adjust the business of the target user.
[0059] Specifically, if the predicted risk value exceeds the preset risk threshold, a business plan that increases the target user's liability is screened out for business adjustment; if the predicted risk value is lower than the preset risk threshold, a business plan that reduces the target user's liability is screened out for business adjustment.
[0060] Correspondingly, the business plan that increases the target user's liability, for example, when it is predicted that the claim risk value of the car owner user is high, a car insurance premium plan that increases the car insurance premium is screened out; and correspondingly, the business plan that reduces the target user's liability, for example, when it is predicted that the claim risk value of the car owner user is low, a car insurance premium plan that reduces the car insurance premium is screened out.
[0061] The business adjustment method in this embodiment can also be applied to other financial business risk prediction scenarios, such as investment risk prediction, financial risk prediction, etc. At this time, the historical business data is the historical investment or financial data of the investment user. By using the artificial intelligence prediction model and combining the historical business data of the user to predict the business risk value, the business risk value is predicted more intelligently and accurately, and it is also convenient for related parties to avoid certain business risks.
[0062] In this embodiment, the historical business data of the target user is obtained; the historical business data is time-sequenced and sorted to obtain time-sequenced historical business data; the time-sequenced historical business data is input into the pre-trained business risk prediction model; the predicted risk value output by the pre-trained business risk prediction model is obtained; by comparison, the relationship between the predicted risk value and the preset risk threshold is identified; according to the relationship, different business adjustment measures are screened out for business adjustment of the target user. The business adjustment method is applied to the financial business risk prediction scenario, such as investment risk prediction, financial risk prediction, and car insurance underwriting risk prediction, etc. By using the artificial intelligence prediction model and combining the historical business data of the user to predict the business risk value, the business risk value is predicted more intelligently and accurately, and it is also convenient for related parties to avoid certain business risks and timely adjust the business for risk users.
[0063] Continuing to refer to Figure 3 In some optional implementations, before step 203, the method further includes a step of pre-training the business risk prediction model, Figure 3 is a flowchart of a specific embodiment of the business adjustment method described in the present application for pre-training the business risk prediction model, including the following steps:
[0064] Step 301, obtaining batch time-sequenced historical risk business data;
[0065] In this embodiment, the historical risk business data includes past claim data for risk assessment during vehicle insurance underwriting and fraud data for fraud insurance, etc.
[0066] In step 302, the number of data pieces of the batch time-series historical risk business data is counted.
[0067] By obtaining the batch time-series historical risk business data, the number of data pieces of the batch time-series historical risk business data is counted, so as to count whether the training data for the business risk prediction model is sufficient, to strictly ensure that the business risk prediction model is trained by using sufficient training data, and to improve the prediction accuracy of the business risk prediction model.
[0068] In step 303, if the number of data pieces exceeds a preset data piece threshold, a first training data set is generated from the batch time-series historical risk business data.
[0069] In step 304, the first training data set is input into a to-be-trained time-series neural network, and business risk features are extracted.
[0070] In this embodiment, the to-be-trained time-series neural network includes an LSTM time-series neural network. The LSTM time-series neural network can help capture long-term dependencies in time-series data, and can also use the gating mechanism (forget gate, input gate, and output gate) of the LSTM time-series neural network to allow the network to filter important data and discard unimportant data according to the business prediction needs, so that the model training and actual prediction pay more attention to important risk prediction features, better time-series modeling of historical data, extraction of potential long-term risk signals, and improvement of prediction accuracy of business risk.
[0071] In step 305, the time-series neural network is trained according to the business risk features, and a pre-trained business risk prediction model is obtained.
[0072] Specifically, by training the business risk prediction model, the business risk prediction model can identify the business risk data contained in the to-be-predicted historical business data according to the learned business risk features when obtaining the to-be-predicted historical business data, and can calculate a total business risk value according to the business risk feature weights corresponding to different business risk data, so as to obtain a final business risk.
[0073] With reference to Figure 4 In some optional implementations, after step 302, the method further includes a step of expanding the historical risk business data, Figure 4is a flowchart of a specific embodiment of the business adjustment method described in the present application for expanding historical risk business data, including the following steps:
[0074] Step 401, if the number of data does not exceed the preset data number threshold, a second training data set is generated from the batch time-series historical risk business data;
[0075] Specifically, for example: in the car insurance underwriting scenario, the claim data or fraud data, the data volume is still not enough, and only the existing related data is used for business risk prediction model training, which is easy to lead to the prediction accuracy of the trained business risk prediction model is low.
[0076] Step 402, input the second training data set as the discriminant information in the discriminator into the preset adversarial learning training network, and perform adversarial generation training to train the generator corresponding to the discriminator;
[0077] Specifically, when the number of historical risk business data does not exceed the preset data number threshold, the GAN adversarial generation network is used to generate adversarial generation for the historical risk business data in the second training data set, and the second training data set is expanded according to the adversarial generation result. The GAN adversarial generation network can generate high-quality simulation data, which can alleviate the problem of data scarcity in the car insurance field, thereby improving the training effect of the business risk prediction model.
[0078] Step 403, continue to obtain the latest generated simulation data until the simulation data generated by the generator meets the preset condition;
[0079] Step 404, add the simulation data to the second training data set in an incremental manner.
[0080] By adversarial generation of historical risk business data in the second training data set, and expanding the second training data set according to the adversarial generation result, sufficient training data is ensured for subsequent business risk prediction model training, and the prediction accuracy of the business risk prediction model is improved.
[0081] Continue to refer to Figure 5 , Figure 5 is Figure 4 a flowchart of a specific embodiment of step 402, including the following steps:
[0082] Step 501, data feature extraction is performed on all time-series historical risk business data in the second training data set;
[0083] In this embodiment, data feature extraction is performed on historical risk business data to facilitate subsequent identification of business risk features.
[0084] Step 502, according to the data feature extraction result and the preset risk feature label, screening out common business risk features from the data feature extraction result;
[0085] Step 503, generating business data continuously by simulating through the common business risk features, to obtain simulation data;
[0086] Step 504, continuously inputting the latest simulation data into the discriminator to perform true data and simulation data binary classification judgment, to obtain a binary classification judgment result;
[0087] Step 505, counting the binary classification judgment results for all simulation data, and calculating a misclassification probability according to the statistical results, wherein the misclassification probability refers to the probability of dividing simulation data into true data;
[0088] Specifically, as the number of adversarial training of the generator increases, the generator will gradually be trained to maturity, and the misclassification probability will become larger and larger. Moreover, as the generated simulation data gradually increases, the misclassification probability will gradually tend to a stable value. Assuming that the number of adversarial generation is large, the misclassification probability will approach 100% as the generator is gradually trained to completion.
[0089] In this embodiment, the step of counting the binary classification judgment results for all simulation data and calculating the misclassification probability according to the statistical results specifically includes:
[0090] According to a preset probability formula:
[0091]
[0092] Calculate the misclassification probability, wherein P t represents the misclassification probability, m t represents the number of data classified as true data, m s represents the number of data classified as simulation data.
[0093] Step 506, until the misclassification probability is in a stable value state within a preset target test time period, the generator training is completed.
[0094] Specifically, the longer the preset target test time period is, the closer the generator's generation result is to the true data.
[0095] In this embodiment, the preset condition includes that the misclassification probability is in a stable value state within a preset target test time period.
[0096] Specifically, the step of continuously obtaining the latest generated simulation data until the simulation data generated by the generator meets the preset condition comprises: determining whether the misclassification probability is always in a stable value state within a preset target test time period; if the misclassification probability is not always in a stable value state within the preset target test time period, continuing to perform the adversarial generation training on the generator; and if the misclassification probability is always in a stable value state within the preset target test time period, continuously obtaining the latest generated simulation data.
[0097] With reference to the foregoing Figure 6 In some optional implementations, after step 404, the method further comprises the steps of terminating the expansion and learning training after the expansion. Figure 6 FIG. 4 is a flowchart of a specific embodiment of the method for adjusting a business according to the present application, which comprises the following steps of:
[0098] Step 601: iteratively counting the number of data in the second training data set;
[0099] Step 602: terminating the generation of the generator until the number of data in the second training data set exceeds the data quantity threshold;
[0100] Step 603: updating the second training data set as the first training data set;
[0101] Step 604: inputting the first training data set into the time-series neural network to be trained to extract business risk features;
[0102] Step 605: learning and training the time-series neural network according to the business risk features to obtain the pre-trained business risk prediction model.
[0103] By terminating the expansion and learning training after the expansion, it is strictly ensured that the business risk prediction model is trained by using sufficient training data, and the prediction accuracy of the business risk prediction model is improved.
[0104] The method for adjusting a business according to the present application can be applied to a financial business risk prediction scenario, such as investment risk prediction, financial risk prediction, and car insurance underwriting risk prediction, etc. By using an artificial intelligence prediction model and combining historical business data of a user to predict a business risk value, the business risk value is predicted more intelligently and accurately, and it is also convenient for related parties to avoid certain business risks.
[0105] In the embodiment, the historical service data of the target user is acquired; the historical service data is sequentially arranged to obtain the sequentially arranged historical service data; the sequentially arranged historical service data is input into the pre-trained service risk prediction model; a predicted risk value output by the pre-trained service risk prediction model is acquired; by comparison, a relationship between the predicted risk value and a preset risk threshold is identified; and different service adjustment measures are screened according to the relationship to perform service adjustment on the target user. The service adjustment method is applied to a financial service risk prediction scene, such as investment risk prediction, financial risk prediction, and vehicle insurance underwriting risk prediction, etc. By using an artificial intelligence prediction model and combining historical service data of a user to predict a service risk value, the service risk value is predicted more intelligently and accurately, and it is convenient for related parties to avoid certain service risks and timely adjust services for risk users.
[0106] The embodiments of the present application can acquire and process related data based on artificial intelligence technology. Artificial intelligence (AI) is the use of digital computers or computer-controlled machines to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.
[0107] The basic technology of artificial intelligence generally includes technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction system, mechatronics, etc. The software technology of artificial intelligence mainly includes computer vision technology, robot technology, biometric technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc.
[0108] In the embodiment, the historical service data of the target user is acquired; the historical service data is sequentially arranged to obtain the sequentially arranged historical service data; the sequentially arranged historical service data is input into the pre-trained service risk prediction model; a predicted risk value output by the pre-trained service risk prediction model is acquired; by comparison, a relationship between the predicted risk value and a preset risk threshold is identified; and different service adjustment measures are screened according to the relationship to perform service adjustment on the target user. The service adjustment method is applied to a financial service risk prediction scene, such as investment risk prediction, financial risk prediction, and vehicle insurance underwriting risk prediction, etc. By using an artificial intelligence prediction model and combining historical service data of a user to predict a service risk value, the service risk value is predicted more intelligently and accurately, and it is convenient for related parties to avoid certain service risks and timely adjust services for risk users.
[0109] Further reference is made to Figure 7, as an implementation of the method shown in the above Figure 2 The application provides an embodiment of a service adjustment device, which corresponds to the method embodiment shown in the above Figure 2 The device can be applied to various electronic devices.
[0110] As shown in the above Figure 7 The service adjustment device 700 comprises a historical service data acquisition module 701, a time-series arrangement module 702, a risk prediction model input module 703, a predicted risk value acquisition module 704, a comparison and identification module 705 and a service adjustment module 706.
[0111] The historical service data acquisition module 701 is configured to acquire historical service data of a target user.
[0112] The time-series arrangement module 702 is configured to arrange the historical service data in time series to acquire time-series historical service data.
[0113] The risk prediction model input module 703 is configured to input the time-series historical service data into a pre-trained service risk prediction model.
[0114] The predicted risk value acquisition module 704 is configured to acquire a predicted risk value output by the pre-trained service risk prediction model.
[0115] The comparison and identification module 705 is configured to identify a relationship between the predicted risk value and a preset risk threshold through comparison.
[0116] The service adjustment module 706 is configured to screen different service adjustment measures to adjust the service of the target user according to the relationship.
[0117] The application acquires historical service data of a target user, arranges the historical service data in time series to acquire time-series historical service data, inputs the time-series historical service data into a pre-trained service risk prediction model, acquires a predicted risk value output by the pre-trained service risk prediction model, identifies a relationship between the predicted risk value and a preset risk threshold through comparison, and screens different service adjustment measures to adjust the service of the target user according to the relationship. The service adjustment method is applied to a financial service risk prediction scene, for example, investment risk prediction, financial risk prediction and vehicle insurance underwriting risk prediction, and the service risk value is predicted in a more intelligent and accurate manner by using an artificial intelligence prediction model and combining historical service data of a user, so that the related parties can avoid certain service risks and timely adjust the service for a risk user.
[0118] In the embodiment, the business adjustment device 700 further comprises a historical risk business data acquisition module, a historical risk business data statistical module, a first branch processing module, a business risk feature extraction module and a business risk prediction model training module. Among them:
[0119] The historical risk business data acquisition module is used to acquire batch time-series historical risk business data.
[0120] The historical risk business data statistical module is used to count the number of data of the batch time-series historical risk business data.
[0121] The first training data set generation module is used to generate a first training data set from the batch time-series historical risk business data if the number of data exceeds a preset data number threshold.
[0122] The business risk feature extraction module is used to input the first training data set into a to-be-trained time-series neural network to extract business risk features.
[0123] The business risk prediction model training module is used to learn and train the time-series neural network according to the business risk features to obtain the pre-trained business risk prediction model.
[0124] In the embodiment, the business adjustment device 700 further comprises a second training data set generation module, an adversarial generation training module, a simulation data acquisition module and a simulation data adding module. Among them:
[0125] The second training data set generation module is used to generate a second training data set from the batch time-series historical risk business data if the number of data does not exceed a preset data number threshold.
[0126] The adversarial generation training module is used to input the second training data set as discrimination information in a discriminator into a preset adversarial learning training network to perform adversarial generation training and train a generator corresponding to the discriminator.
[0127] The simulation data acquisition module is used to continuously acquire newly generated simulation data until the simulation data generated by the generator meets a preset condition.
[0128] The simulation data adding module is used to add the simulation data to the second training data set in a piece-by-piece incremental adding manner.
[0129] In the embodiment, the adversarial generation training module comprises a data feature extraction unit, a common business risk feature screening unit, a data simulation generation unit, a binary classification judgment unit, a misclassification probability calculation unit and a generator state determination unit. Among them:
[0130] a data feature extraction unit configured to perform data feature extraction on all time-series historical risk business data in the second training data set;
[0131] a common business risk feature screening unit configured to screen common business risk features from the data feature extraction results according to the data feature extraction results and preset risk feature labels;
[0132] a data simulation generation unit configured to continuously simulate and generate business data based on the common business risk features to obtain simulation data;
[0133] a binary classification judgment unit configured to continuously input the latest simulation data into the discriminator to perform binary classification judgment on the real data and the simulation data to obtain a binary classification judgment result;
[0134] a misclassification probability calculation unit configured to count the binary classification judgment results for all simulation data and calculate a misclassification probability according to the counting results, wherein the misclassification probability refers to a probability of dividing the simulation data into real data;
[0135] a generator state determination unit configured to determine that the generator training is completed when the misclassification probability is in a stable value state within a preset target test time period.
[0136] In this embodiment, the adversarial generation training module further includes a stable value state unit, a first branch processing unit and a second branch processing unit. Wherein:
[0137] the stable value state unit is configured to determine whether the misclassification probability is in a stable value state within a preset target test time period;
[0138] the first branch processing unit is configured to continue adversarial generation training of the generator if the misclassification probability is not in a stable value state within a preset target test time period;
[0139] the second branch processing unit is configured to continuously obtain the latest simulation data generated if the misclassification probability is in a stable value state within a preset target test time period.
[0140] In this embodiment, the business adjustment device 700 further includes a data number iteration statistical module, a generator generation termination module and a training data set update module. Wherein:
[0141] the data number iteration statistical module is configured to iteratively count the number of data in the second training data set;
[0142] A generator generation termination module is used to terminate the generator's generation until the number of data entries in the second training dataset exceeds the data entry threshold.
[0143] The training dataset update module is used to update the second training dataset to the first training dataset.
[0144] For the first training dataset obtained from the update, the business risk feature extraction module is used to input the first training dataset into the temporal neural network to be trained to extract business risk features; and the business risk prediction model training module is used to learn and train the temporal neural network based on the business risk features to obtain the pre-trained business risk prediction model.
[0145] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing related hardware through computer-readable instructions. These computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, optical disk, or read-only memory (ROM), or random access memory (RAM).
[0146] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.
[0147] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 8 , Figure 8 This is a basic structural block diagram of the computer device in this embodiment.
[0148] The computer device 8 includes a memory 8a, a processor 8b, and a network interface 8c that are interconnected via a system bus. It should be noted that... Figure 8The computer device 8 is shown to have a component storage 8a, a processor 8b, and a network interface 8c, but it should be understood that not all of the illustrated components are required to be implemented, and more or fewer components can be implemented instead. As understood by one skilled in the art, the computer device herein is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, a microprocessor, an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital signal processor (DSP), an embedded device, etc.
[0149] The computer device can be a desktop computer, a notebook computer, a palm computer, a cloud server, or the like. The computer device can interact with a user through a keyboard, a mouse, a remote controller, a touchpad, a voice control device, or the like.
[0150] The storage 8a includes at least one type of readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the storage 8a can be an internal storage unit of the computer device 8, such as a hard disk or a memory of the computer device 8. In other embodiments, the storage 8a can also be an external storage device of the computer device 8, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. The storage 8a can also include both an internal storage unit and an external storage device of the computer device 8. In this embodiment, the storage 8a is generally used to store an operating system and various application software installed in the computer device 8, such as computer readable instructions of a service adjustment method, etc. In addition, the storage 8a can also be used to temporarily store various data that have been output or will be output.
[0151] The processor 8b may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 8b is generally used to control the overall operation of the computer device 8. In the present embodiment, the processor 8b is configured to execute computer readable instructions stored in the memory 8a or process data, such as computer readable instructions for implementing the business adjustment method.
[0152] The network interface 8c may include a wireless network interface or a wired network interface, and is generally used to establish a communication connection between the computer device 8 and other electronic devices.
[0153] The computer device according to the present embodiment belongs to the field of artificial intelligence technology and is applied to a scenario of adjusting a business contract or a business agreement. The present application obtains historical business data of a target user, time-series processes the historical business data to obtain time-series historical business data, inputs the time-series historical business data into a pre-trained business risk prediction model, obtains a predicted risk value output by the pre-trained business risk prediction model, identifies a relationship between the predicted risk value and a preset risk threshold through comparison, and selects different business adjustment measures to adjust the business of the target user according to the relationship. The business adjustment method is applied to a financial business risk prediction scenario, such as investment risk prediction, financial risk prediction, and vehicle insurance underwriting risk prediction. The business risk value is predicted by using an artificial intelligence prediction model and combining historical business data of a user, which is more intelligent and accurate, and is also convenient for related parties to avoid certain business risks and timely adjust the business of a risk user.
[0154] The present application also provides another embodiment, i.e., a computer readable storage medium storing computer readable instructions, which can be executed by a processor to enable the processor to perform the steps of a business adjustment method as described above.
[0155] The computer readable storage medium provided in the embodiment belongs to the technical field of artificial intelligence, and is applied to a scenario of adjusting a business contract or a business agreement. The application obtains historical business data of a target user, sequentially processes the historical business data to obtain sequentially processed historical business data, inputs the sequentially processed historical business data into a pre-trained business risk prediction model, obtains a predicted risk value output by the pre-trained business risk prediction model, identifies a relationship between the predicted risk value and a preset risk threshold through comparison, and adjusts the business of the target user according to different business adjustment measures according to the relationship. The business adjustment method is applied to a financial business risk prediction scenario, such as investment risk prediction, financial risk prediction, and vehicle insurance underwriting risk prediction. The business risk value is predicted in a more intelligent and accurate manner by using an artificial intelligence prediction model and combining historical business data of a user, which is convenient for related parties to avoid certain business risks and timely adjust the business of a risk user.
[0156] From the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment method can be realized by means of software and a general hardware platform, of course, it can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a plurality of instructions for making a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device) execute the methods described in the various embodiments of the present application.
[0157] Obviously, the above-described embodiments are only some of the embodiments of the present application, not all the embodiments, and the preferred embodiments of the present application are given in the drawings, but do not limit the patent scope of the present application. The present application can be realized in many different forms, and conversely, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing specific embodiments, or make equivalent replacements to some technical features. Any equivalent structure made by using the contents of the specification and drawings, directly or indirectly applied to other related technical fields, is also within the scope of the patent protection of the present application. The non-company software tools or components appearing in the embodiments of the present application are only examples for introduction, not representing actual use.
Claims
1. A business adjustment method, characterized in that, Includes the following steps: Obtain historical business data of the target user; The historical business data is processed in a time-series manner to obtain time-series historical business data; The time-series historical business data is input into the pre-trained business risk prediction model; Obtain the predicted risk value output by the pre-trained business risk prediction model; By comparison, the relationship between the predicted risk value and the preset risk threshold is identified; Based on the aforementioned relationship, different business adjustment measures are selected to adjust the business for the target users.
2. The business adjustment method according to claim 1, characterized in that, Before performing the step of inputting the time-series historical business data into the pre-trained business risk prediction model, the method further includes: Acquire batch-time-series historical risk business data; Perform a count of the number of data entries on the batch-time-series historical risk business data; If the number of data entries exceeds a preset data entry threshold, then the first training dataset is generated using the batch-time-series historical risk business data. The first training dataset is input into the temporal neural network to be trained to extract business risk features; The temporal neural network is trained based on the business risk characteristics to obtain the pre-trained business risk prediction model.
3. The business adjustment method according to claim 2, characterized in that, After performing the step of counting the number of data entries in the batch-time-series historical risk business data, the method further includes: If the number of data entries does not exceed the preset data entry threshold, then a second training dataset is generated using the batch-time-series historical risk business data; The second training dataset is used as the discrimination information in the discriminator and input into the preset adversarial learning training network to perform adversarial generation training, thereby training the generator corresponding to the discriminator. The generator continues to acquire the latest generated simulation data until the simulation data it generates meets the preset conditions. The simulated data is added to the second training dataset incrementally, one record at a time.
4. The business adjustment method according to claim 3, characterized in that, The step of inputting the second training dataset as discrimination information in the discriminator into a preset adversarial learning training network for adversarial generation training, and training the generator corresponding to the discriminator, specifically includes: Data feature extraction is performed on all time-series historical risk business data in the second training dataset; Based on the data feature extraction results and preset risk feature labels, common business risk features are selected from the data feature extraction results; Simulated data is obtained by continuously simulating and generating business data based on the aforementioned common business risk characteristics; The latest simulated data is continuously input into the discriminator to perform binary classification judgment between real data and simulated data, and to obtain the binary classification judgment result. The binary classification results for all simulated data are statistically analyzed, and the misclassification probability is calculated based on the statistical results. The misclassification probability refers to the probability of classifying simulated data as real data. The generator training is complete when the misclassification probability remains stable within the preset target testing time period.
5. The business adjustment method according to claim 4, characterized in that, The step of statistically analyzing the binary classification results of all simulated data and calculating the misclassification probability based on the statistical results specifically includes: According to the preset probability formula: Calculate the misclassification probability, where P t Let m represent the misclassification probability. t m represents the number of data entries classified as real data. s This indicates the number of data entries that are classified as simulated data.
6. The business adjustment method according to claim 3, characterized in that, The preset conditions include ensuring that the misclassification probability remains stable within a preset target testing period. The step of continuously acquiring the latest generated simulation data until the simulated data generated by the generator meets the preset conditions specifically includes: Determine whether the misclassification probability remains stable within the preset target testing period; If the misclassification probability does not remain stable within the preset target testing time period, then the generator will continue to undergo adversarial generation training. If the misclassification probability remains stable within the preset target test period, the latest generated simulation data will be continuously acquired.
7. The business adjustment method according to claim 3, characterized in that, After performing the step of adding the simulated data to the second training dataset incrementally, the method further includes: Iteratively count the number of data entries in the second training dataset; The generator continues to generate data until the number of data entries in the second training dataset exceeds the data entry threshold. Update the second training dataset to the first training dataset; The first training dataset is input into the temporal neural network to be trained to extract business risk features; The temporal neural network is trained based on the business risk characteristics to obtain the pre-trained business risk prediction model.
8. A business adjustment device, characterized in that, The service adjustment device is used to implement the steps of the service adjustment method as described in any one of claims 1 to 7, and the service adjustment device includes: The historical business data acquisition module is used to acquire the historical business data of the target user. The time-series processing module is used to process the historical business data in a time-series manner to obtain time-series historical business data. The risk prediction model input module is used to input the time-series historical business data into the pre-trained business risk prediction model; The predicted risk value acquisition module is used to acquire the predicted risk value output by the pre-trained business risk prediction model. The comparison and identification module is used to identify the relationship between the predicted risk value and the preset risk threshold by comparison; The business adjustment module is used to select different business adjustment measures to adjust the business of the target user based on the relationship.
9. A computer device, characterized in that, The device includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the business adjustment method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the service adjustment method as described in any one of claims 1 to 7.