Bank customer risk dynamic early warning prevention and control method, device and equipment

By constructing a transaction analysis model and correcting the customer risk evaluation model, the problem of poor model adaptability in the bank customer risk warning method is solved, the accuracy and timeliness of early warning are improved, and dynamic management of bank customer risks is realized.

CN118967290APending Publication Date: 2024-11-15CHINA CITIC BANK CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411026704.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-30
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The bank customer risk warning method has poor model adaptability and low warning accuracy and timeliness, and cannot be adjusted in time according to changes in the financial environment and customer needs.

Method used

By building a transaction analysis model, including early warning model and customer risk assessment model, data pull and risk score are carried out, early warning tasks and account risk disposal results are generated, and the customer risk assessment model is corrected based on the results.

Benefits of technology

It improves model adaptability, improves the accuracy and timeliness of risk warnings, and ensures the real-time and accuracy of risk management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118967290A_ABST
    Figure CN118967290A_ABST
Patent Text Reader

Abstract

The invention discloses a bank customer risk dynamic early warning prevention and control method, device and equipment, and relates to the related field of financial risk management, and the method comprises the steps: constructing a transaction analysis model; performing data pulling on the risk control platform by adopting an early warning model to obtain customer risk data of the risk-hit customer; synchronizing the customer risk data to a customer risk evaluation model for risk scoring to obtain a customer risk score; the risk control platform receives the customer risk score fed back by the transaction analysis model, and generates an early warning task and an account risk disposal result according to the customer risk score; and the transaction analysis model receives the early warning task and the account risk disposal result and corrects the customer risk evaluation model according to the early warning task and the account risk disposal result. The technical problems that an existing bank customer risk early warning method is poor in model adaptability and low in early warning accuracy and timeliness are solved, and the technical effects that the model adaptability is improved, and the risk early warning accuracy and timeliness are improved are achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of financial risk management, and in particular to a method, device and equipment for dynamic early warning and prevention of bank customer risks. Background Art

[0002] With the rapid development of banking business and the continuous advancement of financial technology, the scale and complexity of bank customer transaction data have increased dramatically, which has brought huge challenges to the traditional risk management model, especially in terms of customer risk warning and prevention. At present, banks often use static data and fixed risk scoring models for customer risk warning. Due to the lag in data update, the model cannot be adjusted in time according to changes in the financial environment and customer needs, resulting in reduced warning accuracy.

[0003] Among the current related technologies, bank customer risk early warning methods have technical problems such as poor model adaptability, low early warning accuracy and timeliness. Summary of the invention

[0004] This application provides a method, device and equipment for dynamic early warning and prevention of bank customer risks, adopts technical means such as building a transaction analysis model, pulling data, performing risk scoring, generating early warning tasks and account risk disposal results, and correcting the customer risk assessment model, thereby achieving the technical effect of improving the adaptability of the model and improving the accuracy and timeliness of risk warnings.

[0005] This application provides a bank customer risk dynamic early warning and prevention method, including:

[0006] Construct a transaction analysis model, wherein the transaction analysis model includes an early warning model and a customer risk assessment model; use the early warning model to pull data from the risk control platform to obtain customer risk data of risky customers; synchronize the customer risk data to the customer risk assessment model for risk scoring to obtain a customer risk score; the risk control platform receives the customer risk score fed back by the transaction analysis model, and generates an early warning task and an account risk disposal result based on the customer risk score; the transaction analysis model receives and modifies the customer risk assessment model based on the early warning task and the account risk disposal result.

[0007] The present application also provides a bank customer risk dynamic early warning and prevention device, comprising:

[0008] A transaction analysis model construction module, the transaction analysis model construction module is used to construct a transaction analysis model, wherein the transaction analysis model includes an early warning model and a customer risk assessment model; a customer risk data acquisition module, the customer risk data acquisition module is used to use the early warning model to pull data from the risk control platform to obtain customer risk data of risky customers; a risk scoring module, the risk scoring module is used to synchronize the customer risk data to the customer risk assessment model for risk scoring to obtain a customer risk score; a risk control platform processing module, the risk control platform processing module is used for the risk control platform to receive the customer risk score fed back by the transaction analysis model, and generate an early warning task and an account risk disposal result according to the customer risk score; a customer risk assessment model correction module, the customer risk assessment model correction module is used for the transaction analysis model to receive and correct the customer risk assessment model according to the early warning task and the account risk disposal result.

[0009] The present application also provides an electronic device, including:

[0010] The memory is used to store executable instructions; the processor is used to implement a dynamic early warning and prevention method for bank customer risks when executing the executable instructions stored in the memory.

[0011] A method, device and equipment for dynamic early warning and control of bank customer risks proposed in this application first constructs a transaction analysis model, wherein the transaction analysis model includes an early warning model and a customer risk assessment model, and then uses the early warning model to pull data from the risk control platform to obtain customer risk data of risky customers, and then synchronizes the customer risk data to the customer risk assessment model for risk scoring to obtain a customer risk score. The risk control platform then receives the customer risk score fed back by the transaction analysis model, and generates early warning tasks and account risk disposal results based on the customer risk score. Finally, the transaction analysis model receives and corrects the customer risk assessment model based on the early warning tasks and account risk disposal results, thereby achieving the technical effect of improving the adaptability of the model and improving the accuracy and timeliness of risk warnings. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the accompanying drawings of the embodiment of the present invention will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the device according to the embodiment of the present application. It should be understood that the previous or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0013] Figure 1A flow chart of a method for dynamic early warning and prevention of bank customer risks provided in an embodiment of the present application.

[0014] Figure 2 A schematic diagram of the structure of a bank customer risk dynamic early warning and control device provided in an embodiment of the present application.

[0015] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0016] Explanation of the accompanying drawings: transaction analysis model building module 10, customer risk data acquisition module 20, risk scoring module 30, risk control platform processing module 40, customer risk assessment model correction module 50. DETAILED DESCRIPTION

[0017] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0018] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0019] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, device, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0020] The present application embodiment provides a bank customer risk dynamic early warning and prevention method, such as Figure 1 As shown, the method includes:

[0021] Step S100, constructing a transaction analysis model, wherein the transaction analysis model includes an early warning model and a customer risk assessment model. Specifically, the transaction analysis model is a mathematical model for analyzing and predicting customer transaction risks, collecting historical transaction data, personal information, credit records, etc. of customers from the banking system, cleaning, converting and standardizing the collected data to ensure data quality, extracting key features from the pre-processed data, such as transaction amount, transaction frequency, transaction location, etc., and using machine learning or statistical methods to train the early warning model and customer risk assessment model, wherein the early warning model is a model specifically used to identify customers or transactions that may be risky; the customer risk assessment model is a model used to quantitatively evaluate the overall risk of customers.

[0022] In a possible implementation, step S100 further includes step S110, obtaining risk control rules according to business demand configuration, and constructing and generating the early warning model according to the risk control rules. Specifically, clarify business needs, such as identifying high-risk transactions, fraudulent behaviors, etc., formulate corresponding risk control rules according to business needs, such as transaction amount thresholds, transaction frequency limits, etc., convert risk control rules into a form understandable to mathematical models, such as logical expressions, conditional statements, etc., and use tools such as machine learning or rule engines to construct an early warning model based on the converted risk control rules. Step S120, the interactive big data platform obtains the first calculation factor set, the first calculation factor weight set and the unit customer calculation rule, and constructs a unit customer linear scoring model based on the first calculation factor set, the first calculation factor weight set and the unit customer calculation rule. Specifically, the method interacts with the big data platform to obtain the calculation factor set, weight set and calculation rules related to the unit customer, wherein the unit customer refers to a customer in the form of a unit (such as an enterprise or an institution), and uses linear methods such as linear regression and logistic regression to combine the obtained calculation factor (indicator or variable for assessing customer risk) set, weight (relative importance or weight of each calculation factor in the assessment) set and calculation rules to construct a unit customer linear scoring model, wherein the linear scoring model is a scoring model constructed based on a linear method, and the risk score is calculated by weighted summation and other methods. Step S130, interacting with the big data platform to obtain a second calculation factor set, a second calculation factor weight set and a personal customer calculation rule, and constructing a personal customer linear scoring model based on the second calculation factor set, the second calculation factor weight set and the personal customer calculation rule. Specifically, it is similar to step S120, but is aimed at personal customers, that is, customers in the form of individuals. Step S140, interacting with the big data platform to obtain a unit customer nonlinear scoring model and a personal customer nonlinear scoring model. Specifically, interact with the big data platform to directly obtain the trained nonlinear scoring models for corporate customers and individual customers, that is, the scoring models built based on nonlinear methods (such as decision trees, neural networks, etc.), which can capture more complex nonlinear relationships. Step S150, integrate the corporate customer linear scoring model, the individual customer linear scoring model, the corporate customer nonlinear scoring model and the individual customer nonlinear scoring model to obtain the customer risk assessment model. Specifically, the multiple scoring models constructed or obtained in steps S120, S130 and S140 are integrated to form a comprehensive customer risk assessment model. This implementation method integrates different scoring models (including linear and nonlinear models) and comprehensively considers multiple factors and indicators, thereby achieving the technical effect of improving the accuracy and comprehensiveness of customer risk assessment.

[0023] In a possible implementation, step S140 further includes step S141, constructing the nonlinear scoring model for the unit customer based on a black box algorithm, wherein the input of the nonlinear scoring model for the unit customer is the customer number, customer nature and model hit times, and the output is the customer risk score of the unit customer. Specifically, select a suitable nonlinear black box algorithm, such as a neural network, a random forest, a gradient boosting machine, etc., collect historical data, including customer number, customer nature (such as enterprise type, industry category, etc.), model hit times (i.e., the number of times previously marked as risk by other risk control models), etc. as input features, assign a risk score or risk level to each customer based on historical defaults, fraud records or other risk events, use the selected black box algorithm and the prepared data to train the model, optimize the model performance by adjusting the algorithm parameters, and use the validation set to verify the accuracy, stability and generalization ability of the model. Step S142, preset a first data update cycle, and execute the ORMM base data call with the first data update cycle as a constraint, and use the ORMM base data to execute the construction and training of the nonlinear scoring model for the unit customer. Specifically, according to business needs and the frequency of data changes, a suitable data update cycle (such as daily, weekly or monthly) is set. When the data update cycle arrives, the latest underlying data is called from ORMM (a specific data management system or data warehouse used to store and manage various types of enterprise data). These data include the latest customer information, transaction records, risk events, etc. The unit customer nonlinear scoring model is retrained using the newly acquired underlying data to ensure that the model can reflect the latest data changes and risk trends. Step S143, construct the individual customer nonlinear scoring model based on the black box algorithm, wherein the input of the individual customer nonlinear scoring model is the customer number, customer nature and model hit times, and the output is the customer risk score of the individual customer. Specifically, similar to step S141, it is aimed at individual customers. Step S144, preset a second data update cycle, and use the second data update cycle as a constraint to call the platform underlying data of the risk control platform, and use the platform underlying data to perform the construction and training of the individual customer nonlinear scoring model. Specifically, similar to step S142, different data update cycles are set for individual customers, and the latest underlying data, including individual customers' credit records, transaction histories, social data, etc., are called from the risk control platform. The newly acquired underlying data is used to retrain the individual customer's nonlinear scoring model to ensure that the model can reflect the individual customer's latest risk status. This implementation method ensures that the model can reflect the latest risk trends and data changes through regular data updates and model retraining, achieving the technical effect of maintaining the effectiveness and reliability of the nonlinear scoring model.

[0024] Step S200, using the early warning model to pull data from the risk control platform to obtain customer risk data of risky customers. Specifically, according to the needs of the early warning model, the conditions and scope of data pulling are set, and qualified data are pulled from the risk control platform, wherein the risk control platform is a system or platform for recording bank customer risk management information, and the pulled data is verified to ensure the accuracy and completeness of the data, and the customer risk data, that is, all data related to risky customers that may reflect their risk status, is obtained.

[0025] In a possible implementation, before using the early warning model to pull data from the risk control platform and obtain the customer risk data of the risky customers, step S200 further includes step S210, wherein the risk control platform stores multiple customer transaction analysis data of multiple customers; and step S220, calling the stored procedure script to perform daily batch updates on the multiple customer transaction analysis data. Specifically, the risk control platform stores multiple customer transaction analysis data of multiple customers, including customer transaction records, transaction amounts, transaction times, transaction types, etc. The stored procedure script is used to process the update logic of the customer transaction analysis data. Through a scheduled task or scheduling system, the execution of the stored procedure script is triggered on a daily basis. The stored procedure script updates the customer transaction analysis data in the risk control platform according to preset rules and logic, such as adding new data, modifying old data, or deleting invalid data. Step S230, configuring a scheduled scheduling job to perform a scoring job on the multiple customer transaction analysis data, and synchronizing the job results to the multiple customer transaction analysis data. Specifically, a set of scoring logic or algorithm is defined to evaluate the risk level in the customer transaction analysis data, and a scheduled task is configured using the scheduling system to specify the execution time and frequency of the scoring job. When the scheduled task is triggered, the scoring job starts to execute, the customer transaction analysis data is traversed, and the risk score is calculated for each customer or transaction according to the scoring logic. The calculated risk score is synchronized back to the customer transaction analysis data table and stored as an attribute or field of the customer or transaction. Step S240, the early warning model traverses the multiple customer transaction analysis data of the risk control platform based on the risk control rules to pull data and obtain customer risk data of risky customers. Specifically, the early warning model traverses and filters the customer transaction analysis data in the risk control platform according to the risk control rules. During the traversal process, it identifies customers or transactions that meet the risk control rules, that is, the risk conditions. For the identified customers or transactions that meet the risk conditions, its detailed customer risk data is further pulled, such as risk scores, transaction details, and related information. This implementation method ensures the timeliness and accuracy of customer transaction analysis data through daily batch updates and scheduled scoring operations, achieving the technical effect of keeping the data on the risk control platform up to date and accurate.

[0026] In a possible implementation, step S230 further includes step S231, configuring batch timed scoring to perform scoring operations on the multiple customer transaction analysis data to obtain multiple initial scores. Specifically, a scheduled task is configured using the scheduling system, and the scheduled task will be automatically triggered at a specific time every day to perform a batch scoring operation. When the scheduled task is triggered, the batch scoring operation begins to execute, traversing multiple customer transaction analysis data stored in the risk control platform, and calculating an initial risk score for each customer or transaction according to the scoring logic. Step S232, the risk control platform receives the customer risk score fed back by the transaction analysis model, and supplements and updates the multiple initial scores according to the customer number of the risky customer. Specifically, the risk control platform receives feedback from the transaction analysis model, including the risk score for the risky customers. The risk control platform matches the customer number in the feedback with multiple initial scores stored in the database to find the corresponding customer record. Once the corresponding customer record is found, the risk control platform uses the customer risk score fed back by the transaction analysis model to update the initial score, such as replacing the initial score with a more accurate customer risk score, or merging or weighting the customer risk score with the initial score in some form. This implementation method ensures that on the basis of batch scoring, the risk control platform can receive feedback from the transaction analysis model, thereby supplementing and updating the initial score, achieving the technical effect of improving the accuracy and reliability of risk scoring.

[0027] In a possible implementation, step S200 further includes step S250, presetting the observation period time span. Specifically, according to the purpose of observation, a reasonable time span is set as the observation period, that is, a specific time period for observing and evaluating the customer's transaction behavior or risk status, such as a fixed time period (such as 30 days, 90 days, etc.), or a dynamic time period (such as the most recent complete financial quarter, the most recent natural month, etc.). Step S260, the transaction analysis model interacts with the risk control platform with the observation period time span as a constraint to obtain the observation period data. Specifically, according to the observation period time span, the transaction analysis model sends a data request to the risk control platform, requesting to obtain the customer transaction analysis data generated during the observation period. After receiving the data request, the risk control platform screens out the corresponding customer transaction analysis data from the database according to the time span constraint in the request, and returns it to the transaction analysis model. Step S270, the observation period data is used to evaluate and update the early warning model. Specifically, the early warning model is evaluated using observation period data, and the performance of the early warning model during the observation period is analyzed. Based on the evaluation results, the early warning model is updated and adjusted as necessary to improve the model's predictive ability and accuracy. This implementation method ensures that the early warning model always works based on the latest and representative data by presetting the observation period time span and using the data within the time span to evaluate and update the early warning model, achieving the technical effect of maintaining the timeliness and accuracy of the early warning model and improving the efficiency and accuracy of risk identification and early warning.

[0028] Step S300: synchronize the customer risk data to the customer risk assessment model for risk scoring to obtain a customer risk score. Specifically, the pulled customer risk data is preprocessed and converted as necessary, and the prepared data is input into the customer risk assessment model for risk scoring to output the risk score of each customer.

[0029] In a possible implementation, step S300 further includes step S310, locating a target linear scoring model and a target nonlinear scoring model in the customer risk assessment model according to the customer nature of the risk-hitting customer. Specifically, the customer nature information of the risk-hitting customer is obtained from the customer risk data, such as corporate customers, individual customers, etc., and based on the customer nature information, the corresponding target linear scoring model and target nonlinear scoring model are searched in the customer risk assessment model library, wherein the customer risk assessment model library pre-contains risk scoring models constructed for different customer natures, and once a matching model is found in the model library, these models are determined as target scoring models. Step S320, synchronizing the customer risk data to the target linear scoring model and the target nonlinear scoring model to obtain a linear risk score and a nonlinear risk score. Specifically, synchronizing the customer risk data to the target linear scoring model and the target nonlinear scoring model, processing and analyzing the customer risk data through the target scoring model, and calculating the corresponding linear risk score and nonlinear risk score. Step S330, the linear risk score and the nonlinear risk score constitute the customer risk score. Specifically, the linear risk score and the nonlinear risk score are combined. The combination method can be a simple weighted average or a complex calculation based on a certain strategy or algorithm. The combined score is the customer's final risk score, which is used to reflect the customer's overall risk level. This implementation method comprehensively and accurately assesses the customer's risk level by combining the linear scoring model and the nonlinear scoring model. The technical effect of obtaining a comprehensive and detailed customer risk score is achieved.

[0030] Step S400, the risk control platform receives the customer risk score fed back by the transaction analysis model, and generates a warning task and account risk disposal results according to the customer risk score. Specifically, the risk control platform receives the customer risk score fed back by the transaction analysis model, determines whether a warning task needs to be generated according to the customer risk score, generates corresponding warning tasks (indicating prompt information of customers or transactions that may be at risk) for risky customers that need attention, and performs further risk disposal on the accounts of risky customers according to the warning tasks, such as freezing the accounts, contacting the customers, etc., and records the account risk disposal results (result information on the actual status of the account risk, the effectiveness of the disposal measures, etc.).

[0031] Step S500, the transaction analysis model receives and modifies the customer risk assessment model according to the warning task and the account risk disposal result. Specifically, the transaction analysis model receives the warning task and the account risk disposal result sent by the risk control platform, evaluates the customer risk assessment model, and modifies and optimizes the customer risk assessment model according to the warning task and the account risk disposal result to improve the prediction performance. The embodiment of the present application adopts technical means such as constructing a transaction analysis model, pulling data, performing risk scoring, generating warning tasks and account risk disposal results, and modifying the customer risk assessment model, so as to achieve the technical effect of improving the adaptability of the model and improving the accuracy and timeliness of risk warning.

[0032] In a possible implementation, step S500 further includes step S510, presetting a model revision cycle, and accumulating the warning tasks and account risk disposal results with the model revision cycle as a constraint. Specifically, based on business needs, model stability, data update frequency and other factors, determine a suitable model revision cycle, that is, a time interval for regularly evaluating and adjusting the customer risk assessment model. During the model revision cycle, all warning tasks are accumulated. At the same time, during the model revision cycle, account risk disposal results related to these warning tasks are accumulated, including actual risk conditions, effectiveness of disposal measures, whether losses have occurred, etc. Step S520, perform periodic revisions of the customer risk assessment model based on the accumulated results of the warning tasks and account risk disposal results. Specifically, based on the accumulated early warning tasks and account risk disposal results, the performance of the customer risk assessment model during the model revision cycle is evaluated, including the calculation of indicators such as accuracy, recall rate, and F1 value. For problems found in the performance evaluation, such as false positives and false negatives, in-depth analysis is conducted to determine the causes and sources of the problems. According to the results of the problem diagnosis, the customer risk assessment model is adjusted, which may include adjusting model parameters, updating feature selection, improving algorithms, etc. The adjusted customer risk assessment model is verified using historical data or new test data to ensure that the performance of the customer risk assessment model has been improved. If the verification results show that the performance of the customer risk assessment model has been improved, the customer risk assessment model is updated and used for subsequent risk assessments. This implementation method can timely discover problems and deficiencies in the customer risk assessment model by presetting the model revision cycle and regularly accumulating early warning tasks and account risk disposal results, and periodically revise the customer risk assessment model based on these accumulated results, thereby achieving the technical effect of continuously improving the accuracy and stability of the customer risk assessment model.

[0033] In the above, refer to Figure 1 A bank customer risk dynamic early warning and control method according to an embodiment of the present invention is described in detail. Figure 2A bank customer risk dynamic early warning and control device according to an embodiment of the present invention is described.

[0034] A bank customer risk dynamic early warning and control device according to an embodiment of the present invention is used to solve the technical problems of poor model adaptability, low early warning accuracy and timeliness in existing bank customer risk early warning methods, so as to achieve the technical effect of improving model adaptability, improving the accuracy and timeliness of risk early warning. A bank customer risk dynamic early warning and control device includes: a transaction analysis model construction module 10, a customer risk data acquisition module 20, a risk scoring module 30, a risk control platform processing module 40, and a customer risk evaluation model correction module 50.

[0035] The transaction analysis model construction module 10 is used to construct a transaction analysis model, wherein the transaction analysis model includes an early warning model and a customer risk assessment model; the customer risk data acquisition module 20 is used to use the early warning model to pull data from the risk control platform to obtain customer risk data of risky customers; the risk scoring module 30 is used to synchronize the customer risk data to the customer risk assessment model for risk scoring to obtain a customer risk score; the risk control platform processing module 40 is used for the risk control platform to receive the customer risk score fed back by the transaction analysis model, and to generate an early warning task and an account risk disposal result based on the customer risk score; the customer risk assessment model correction module 50 is used for the transaction analysis model to receive and correct the customer risk assessment model based on the early warning task and the account risk disposal result.

[0036] Next, the specific configuration of the transaction analysis model building module 10 will be described in detail. As described above, a transaction analysis model is constructed, wherein the transaction analysis model includes an early warning model and a customer risk assessment model, and the transaction analysis model construction module 10 may further include: an early warning model generation unit is used to obtain risk control rules according to business demand configuration, and to construct and generate the early warning model according to the risk control rules; a unit customer linear scoring model generation unit is used to interact with the big data platform, obtain a first calculation factor set, a first calculation factor weight set and a unit customer calculation rule, and to construct and generate a unit customer linear scoring model according to the first calculation factor set, the first calculation factor weight set and the unit customer calculation rule; a personal customer linear scoring model generation unit is used to interact with the big data platform, obtain a second calculation factor set, a second calculation factor weight set and a personal customer calculation rule, and to construct and generate a personal customer linear scoring model according to the second calculation factor set, the second calculation factor weight set and the personal customer calculation rule; a customer non-linear scoring model acquisition unit is used to interact with the big data platform to obtain a unit customer non-linear scoring model and a personal customer non-linear scoring model; a customer risk assessment model acquisition unit is used to integrate the unit customer linear scoring model, the personal customer linear scoring model, the unit customer non-linear scoring model and the personal customer non-linear scoring model to obtain the customer risk assessment model.

[0037] The nonlinear scoring model for unit customers and the nonlinear scoring model for individual customers are obtained by interacting with the big data platform. The customer nonlinear scoring model acquisition unit may further include: a unit nonlinear scoring model construction subunit for constructing the unit customer nonlinear scoring model based on a black box algorithm, wherein the input of the unit customer nonlinear scoring model is the customer number, the customer nature and the number of model hits, and the output is the customer risk score of the unit customer; a unit customer nonlinear scoring model training subunit for presetting a first data update cycle, and executing the ORMM base data call with the first data update cycle as a constraint, and using the ORMM base data to execute the construction and training of the unit customer nonlinear scoring model; a personal customer nonlinear scoring model construction subunit for constructing the personal customer nonlinear scoring model based on a black box algorithm, wherein the input of the personal customer nonlinear scoring model is the customer number, the customer nature and the number of model hits, and the output is the customer risk score of the individual customer; a personal customer nonlinear scoring model training subunit for presetting a second data update cycle, and executing the platform base data call with the second data update cycle as a constraint, and using the platform base data to execute the construction and training of the personal customer nonlinear scoring model.

[0038] The specific configuration of the risk scoring module 30 will be described in detail below. As described above, the customer risk data is synchronized to the customer risk assessment model for risk scoring to obtain a customer risk score. The risk scoring module 30 may further include: a target scoring model positioning unit for positioning a target linear scoring model and a target nonlinear scoring model in the customer risk assessment model according to the customer properties of the risk-hitting customer; a risk score acquisition unit for synchronizing the customer risk data to the target linear scoring model and the target nonlinear scoring model to obtain a linear risk score and a nonlinear risk score; and a customer risk score acquisition unit for forming the customer risk score from the linear risk score and the nonlinear risk score.

[0039] The specific configuration of the customer risk assessment model correction module 50 will be described in detail below. As described above, the transaction analysis model receives and corrects the customer risk assessment model according to the early warning task and the account risk disposal result. The customer risk assessment model correction module 50 may further include: a model correction cycle preset unit for presetting the model correction cycle and accumulating the early warning task and the account risk disposal result with the model correction cycle as a constraint; a periodic correction unit for performing periodic correction of the customer risk assessment model according to the accumulated result of the early warning task and the account risk disposal result.

[0040] The specific configuration of the customer risk data acquisition module 20 will be described in detail below. As described above, before using the early warning model to pull data from the risk control platform to obtain the customer risk data of the risk-hitting customers, the customer risk data acquisition module 20 may further include: a daily batch update unit for storing multiple customer transaction analysis data of multiple customers on the risk control platform, calling a stored procedure script to perform daily batch updates on the multiple customer transaction analysis data; a scoring job execution unit for configuring a timed scheduling job to perform a scoring job on the multiple customer transaction analysis data, and synchronizing the job results to the multiple customer transaction analysis data; a data pulling unit for the early warning model to traverse the multiple customer transaction analysis data of the risk control platform based on the risk control rules to pull data and obtain the customer risk data of the risk-hitting customers.

[0041] Among them, a scheduled job is configured to execute a scoring job based on the multiple customer transaction analysis data, and the job results are synchronized to the multiple customer transaction analysis data. The scoring job execution unit may further include: an initial scoring acquisition subunit is used to configure batch scheduled scoring to execute a scoring job on the multiple customer transaction analysis data to obtain multiple initial scores; a supplementary update subunit is used for the risk control platform to receive the customer risk score fed back by the transaction analysis model, and to supplement and update the multiple initial scores according to the customer number of the risky customer.

[0042] Among them, the customer risk data acquisition module 20 may further include: an observation period time span preset unit for presetting the observation period time span; an observation period data acquisition unit for the transaction analysis model to interact with the risk control platform with the observation period time span as a constraint to obtain observation period data; and an early warning model evaluation update unit for using the observation period data to perform evaluation updates on the early warning model.

[0043] A bank customer risk dynamic early warning and control device provided in an embodiment of the present invention can execute a bank customer risk dynamic early warning and control method provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0044] Although the present application makes various references to certain modules in the device according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0045] Based on the above embodiments, the present application also provides an electronic device, Figure 3 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, showing a block diagram of an exemplary electronic device suitable for implementing an implementation of the present invention. Figure 3 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention. The electronic device is in the form of a general computing device, and its components may include but are not limited to an input device 301, a processor 302, a memory 303, and an output device 304. Among them, the processor 302 may be one or more; the memory 303 may include a computer-readable medium and at least one program product, and the program product has a set (at least one) of program modules, which are configured to perform the functions of each embodiment of the present application.

[0046] The memory 303 shown in the embodiment of the present invention may adopt any combination of one or more computer-readable media; the computer-readable storage medium may be, but is not limited to, an infrared, semiconductor system, device or component, or any combination of the above, for storing software programs, computer executable programs and modules, such as the program instructions / modules corresponding to a method for dynamic early warning and prevention of bank customer risks in an embodiment of the present invention. The processor 302 executes various functional applications and data processing of the computer device by running the software programs, instructions and modules stored in the memory 303, thereby realizing the above-mentioned method for dynamic early warning and prevention of bank customer risks.

[0047] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for dynamic early warning and prevention of bank customer risk, characterized in that: The method comprises: Constructing a transaction analysis model, wherein the transaction analysis model includes an early warning model and a customer risk assessment model; Use the early warning model to pull data from the risk control platform to obtain customer risk data of risky customers; Synchronizing the customer risk data to the customer risk assessment model for risk scoring to obtain a customer risk score; The risk control platform receives the customer risk score fed back by the transaction analysis model, and generates a warning task and an account risk disposal result according to the customer risk score; The transaction analysis model receives and modifies the customer risk assessment model according to the early warning task and the account risk disposal result.

2. A bank customer risk dynamic early warning and prevention method as claimed in claim 1, characterized in that: Constructing a transaction analysis model, wherein the transaction analysis model includes an early warning model and a customer risk assessment model, and the method includes: Obtain risk control rules according to business requirements and configure and generate the early warning model based on the risk control rules; The interactive big data platform obtains a first calculation factor set, a first calculation factor weight set, and a unit customer calculation rule, and constructs a unit customer linear scoring model based on the first calculation factor set, the first calculation factor weight set, and the unit customer calculation rule; Interacting with the big data platform, obtaining a second calculation factor set, a second calculation factor weight set and a personal customer calculation rule, and constructing a personal customer linear scoring model based on the second calculation factor set, the second calculation factor weight set and the personal customer calculation rule; Interacting with the big data platform to obtain a unit customer nonlinear scoring model and an individual customer nonlinear scoring model; The customer risk assessment model is obtained by integrating the unit customer linear scoring model, the individual customer linear scoring model, the unit customer non-linear scoring model and the individual customer non-linear scoring model.

3. A bank customer risk dynamic early warning and control method as claimed in claim 2, characterized in that: Interacting with the big data platform to obtain a unit customer nonlinear scoring model and an individual customer nonlinear scoring model, the method comprising: Constructing the nonlinear scoring model for unit customers based on a black box algorithm, wherein the input of the nonlinear scoring model for unit customers is the customer number, the customer nature and the number of model hits, and the output is the customer risk score of the unit customer; Preset a first data update cycle, and execute the ORMM base data call with the first data update cycle as a constraint, and use the ORMM base data to execute the construction and training of the unit customer nonlinear scoring model; Constructing the individual customer nonlinear scoring model based on a black box algorithm, wherein the input of the individual customer nonlinear scoring model is the customer number, customer nature and model hit count, and the output is the customer risk score of the individual customer; A second data update cycle is preset, and with the second data update cycle as a constraint, the risk control platform is called for platform base data, and the platform base data is used to execute the construction and training of the individual customer nonlinear scoring model.

4. A bank customer risk dynamic early warning and control method as claimed in claim 2, characterized in that: The customer risk data is synchronized to the customer risk assessment model for risk scoring to obtain a customer risk score, the method comprising: Positioning a target linear scoring model and a target nonlinear scoring model in the customer risk assessment model according to the customer properties of the risky customers; Synchronizing the customer risk data to the target linear scoring model and the target nonlinear scoring model to obtain a linear risk score and a nonlinear risk score; The linear risk score and the nonlinear risk score constitute the customer risk score.

5. A bank customer risk dynamic early warning and control method as claimed in claim 1, characterized in that: The transaction analysis model receives and modifies the customer risk assessment model according to the early warning task and the account risk disposal result, and the method includes: Preset a model revision cycle, and use the model revision cycle as a constraint to accumulate the warning tasks and account risk disposal results; The customer risk assessment model is periodically revised according to the accumulated results of the early warning tasks and account risk disposal results.

6. A bank customer risk dynamic early warning and control method as claimed in claim 2, characterized in that: The early warning model is used to pull data from the risk control platform to obtain customer risk data of risky customers. Previously, the method included: The risk control platform stores multiple customer transaction analysis data of multiple customers; Calling a stored procedure script to perform daily batch update on the plurality of customer transaction analysis data; Configure a timed scheduling job to perform a scoring job on the plurality of customer transaction analysis data, and synchronize the job results to the plurality of customer transaction analysis data; The early warning model traverses the multiple customer transaction analysis data of the risk control platform based on the risk control rules to pull data and obtain customer risk data of risky customers.

7. A bank customer risk dynamic early warning and control method as claimed in claim 6, characterized in that: Configuring a timed scheduling job to perform a scoring job based on the plurality of customer transaction analysis data, and synchronizing the job result to the plurality of customer transaction analysis data, the method comprising: Configure batch timed scoring to perform scoring operations on the plurality of customer transaction analysis data to obtain a plurality of initial scores; The risk control platform receives the customer risk score fed back by the transaction analysis model, and supplements and updates the multiple initial scores according to the customer numbers of the risky customers.

8. A bank customer risk dynamic early warning and control method as claimed in claim 6, characterized in that: The method comprises: Preset observation period time span; The transaction analysis model interacts with the risk control platform with the observation period time span as a constraint to obtain observation period data; The observation period data is used to evaluate and update the early warning model.

9. A bank customer risk dynamic early warning and control device, characterized in that: The device is used to implement a bank customer risk dynamic early warning and prevention method according to any one of claims 1 to 8, and the device comprises: A transaction analysis model construction module, wherein the transaction analysis model construction module is used to construct a transaction analysis model, wherein the transaction analysis model includes an early warning model and a customer risk assessment model; A customer risk data acquisition module, which is used to extract data from the risk control platform using an early warning model to obtain customer risk data of risky customers; A risk scoring module, which is used to synchronize the customer risk data to the customer risk assessment model for risk scoring to obtain a customer risk score; A risk control platform processing module, wherein the risk control platform receives the customer risk score fed back by the transaction analysis model, and generates an early warning task and an account risk disposal result according to the customer risk score; A customer risk assessment model correction module is used for the transaction analysis model to receive and correct the customer risk assessment model according to the early warning task and the account risk disposal result.

10. An electronic device, characterized in that: The electronic device comprises: A memory for storing executable instructions; The processor is used to implement a bank customer risk dynamic early warning and prevention method as described in any one of claims 1 to 8 when executing the executable instructions stored in the memory.