Risk decision-making method and system, electronic equipment and storage medium
Through a configured risk decision system, including feature derivative platform and model center subsystem, the problems of high cost and low efficiency of existing system construction are solved, and the accuracy and efficiency of credit risk control decisions are achieved quickly responding to market changes and improving the accuracy and efficiency of credit risk control decisions.
Patent Information
- Application Number
- CN202310546538.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-16
- Publication Date
- 2024-11-22
AI Technical Summary
The existing risk decision system with characteristic derivative functions is costly and inefficient during the construction and maintenance process, making it difficult to respond quickly to market changes. The timeliness of credit approval is improved while the risks also increase, and more efficient automated approval processes are required.
A risk decision-making system including feature derivative platform, model center subsystem, business product management subsystem and risk report generation subsystem was designed. The risk control model was quickly built through configuration methods, and users could choose and configure the risk control model to realize The variable library settings for dataset association and risk control models are set, risk reports are generated, and optimized through the model optimization subsystem.
It realizes rapid risk decision-making response capabilities for different business products, reduces the need to repeatedly build risk control systems, improves the accuracy and efficiency of risk control decisions, supports the full life cycle management of credit risk control business in the entire industry, and improves R&D Efficiency and market response speed.
Smart Images

Figure CN119006137A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of computer networks, and in particular to a risk decision-making system with a feature derivation function. In addition, the present application also relates to a related method, an electronic device, and a computer-readable storage medium. Background Art
[0002] With the rapid development of Internet financial services, fraud and crimes against financial institutions are increasing, and they are diversified and technological, which poses a great risk to financial institutions. At the same time, the policies of national regulatory authorities clearly stipulate that banks must have independent and autonomous full-process risk control management capabilities when conducting Internet financial services. Therefore, while developing Internet financial services, it is necessary to establish a supporting full-process risk control system. Financial institutions need to make credit decisions faster and improve the timeliness of credit approval. However, as the timeliness of approval increases, the risk of credit approval also increases, and a more complete automated approval process is needed. Therefore, it is necessary to build a risk decision-making system with feature derivative functions, improve the credit review system, conduct customer credit review efficiently and accurately, and improve the ability to assess and control credit risks. However, the existing risk decision-making systems with feature derivative functions are often fixed model systems, and different systems need to be built for different business products, which are costly and inefficient.
[0003] The content of this background technology description is only for facilitating understanding of the relevant technology in this field and is not regarded as an admission of the prior art. Summary of the invention
[0004] Therefore, the present invention intends to provide a risk decision-making system and method with feature derivation functions, which can quickly realize the construction of risk control models for business products in a configurable manner, improve the risk control decision-making capabilities of business products, improve R&D efficiency, and be able to quickly respond to market changes.
[0005] Specifically, the present invention provides a risk decision-making system with a feature derivation function, including a feature derivation platform, a model center subsystem, a business product management subsystem and a risk report generation subsystem, wherein:
[0006] The feature derivation platform is configured to extract basic variables based on incoming data, third-party data, and internal data, generate derived variables based on the basic variables, and generate data sets based on the basic variables and the derived variables;
[0007] The model center subsystem is configured to configure the risk control model according to the user's selection;
[0008] The business product management subsystem is configured to create a business product, associate the business product with a corresponding data set in the feature derivation platform according to user selection, create a risk control model associated with the business product, set the data set corresponding to the business product to a variable library of the risk control model, call the model center subsystem according to user selection, configure the risk control model associated with the business product, and enable the business product to run;
[0009] The risk report generating subsystem is configured to generate a risk report according to the operation result of the business product.
[0010] In some embodiments of the present invention, the business product management subsystem is further configured to perform simulation testing on the risk control model and generate simulation test results.
[0011] In some embodiments of the present invention, the incoming data includes transaction messages, and the feature derivation platform is specifically configured to automatically parse the transaction messages to obtain basic variables in the transaction messages; based on the basic variables, derived variables are generated according to derived variable configuration rules, wherein the transaction messages include personal second-generation credit reports, corporate credit reports, and third-party reports.
[0012] In some embodiments of the present invention, the business product management subsystem is configured to export the business product according to user selection, and export the data set and risk control model associated with the business product.
[0013] In some embodiments of the present invention, the business product management subsystem is configured to import external business products according to user selection, and import data sets and risk control models associated with the external business products.
[0014] Preferably, in one embodiment of the present invention, when the model center subsystem configures the risk control model associated with the business product, the model center subsystem is specifically configured to configure the risk control model using a policy management component, and the policy management component includes at least one of a decision set management component, a decision table management component, a decision tree management component, a scoring card management component, a machine learning model management component, and a decision flow management component.
[0015] In some embodiments of the present invention, the risk report includes transaction results, transaction input data, third-party data, internal data, and hit rules.
[0016] In some embodiments of the present invention, the system further includes a model optimization subsystem, which is configured to optimize the risk control model according to an optimization mode, wherein the optimization mode includes at least one of an AB testing mode, a champion challenger mode, and a rule backtracking mode.
[0017] In some embodiments of the present invention, a risk decision method is also provided, which is applied to a risk decision system with a feature derivation function, wherein the risk decision system with a feature derivation function includes a feature derivation platform, a model center subsystem, a business product management subsystem, and a risk report generation subsystem, and the method includes:
[0018] The feature derivation platform extracts basic variables based on incoming data, third-party data, and internal data, generates derived variables based on the basic variables, and generates data sets based on the basic variables and the derived variables;
[0019] The model center subsystem configures the risk control model according to the user's selection;
[0020] The business product management subsystem creates a business product, associates the business product with the corresponding data set in the feature derivation platform according to the user's selection, creates a risk control model associated with the business product, sets the data set corresponding to the business product to the variable library of the risk control model, calls the model center subsystem according to the user's selection, configures the risk control model associated with the business product, and enables the business product to run;
[0021] The risk report generating subsystem generates a risk report according to the operation result of the business product.
[0022] In some embodiments of the present invention, the risk decision system with feature derivation function further includes a model optimization subsystem, and the method further includes:
[0023] The model optimization subsystem optimizes the risk control model according to an optimization mode, and the optimization mode includes at least one of an AB testing mode, a champion challenger mode, and a rule backtracking mode.
[0024] In some embodiments of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, wherein when the computer program is executed by a processor, the risk decision-making method of any embodiment of the present invention is implemented.
[0025] In some embodiments of the present invention, an electronic device is provided, comprising: a processor and a memory storing a computer program, wherein the processor is configured to execute the risk decision method of any embodiment of the present invention when running the computer program.
[0026] The embodiment of the present invention proposes a risk decision system and method with a feature derivation function. A business product is established in the system, and the application data and the risk control model are associated. By selecting the corresponding risk control model, the configuration is performed, and the application data is associated with the risk control model. The business product is started to run, and a risk report is generated according to the operation result of the business product. A configurable risk decision system with a feature derivation function is realized, which improves the risk decision response capability for different business products, can quickly realize risk decisions for different business products, and reduces the need for repeated construction of risk control systems. Basic variables are extracted by applying the application data, and based on the derivation rules, life has derived variables, enriching the data type of the risk control model decision, and the decision result of the risk control model is more accurate. The decision system in the embodiment of the present invention can support the credit risk control business of the entire industry, and is locally deployed to realize the unified integration, operation and management of rules, data and models, provide a friendly operation page, and provide risk control strategy configuration operations for risk control personnel. The system can interact with the business system through an interface to provide decision support for the credit approval system. Through the integration and application of internal and external credit data, flexible setting of scenario-based risk control strategies and combination of various auxiliary technologies, we will comprehensively improve the risk control decision-making capabilities of the entire life cycle of credit business in the fields of consumer finance, credit, credit cards, inclusive finance, corporate finance, etc., and at the same time achieve the accumulation, reuse and management of the bank's credit decision-making capabilities, continuously improve demand research and development efficiency, and respond quickly to market changes.
[0027] Other optional features and technical effects of the embodiments of the present invention are partially described below, and partially can be understood by reading this document. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. The elements shown are not limited to the proportions shown in the accompanying drawings. The same or similar reference numerals in the accompanying drawings represent the same or similar elements, wherein:
[0029] Figure 1 The system structure diagram of the risk decision system with feature derivation function is shown;
[0030] Figure 2 A schematic diagram of the user operation flow of a risk decision system with a feature derivation function is shown;
[0031] Figure 3 A flow chart of the risk decision-making method is shown;
[0032] Figure 4 Another schematic diagram of the risk decision-making method is shown;
[0033] Figure 5 An exemplary structural diagram of an electronic device capable of implementing the method according to an embodiment of the present invention is shown. DETAILED DESCRIPTION
[0034] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific implementation methods and drawings. Here, the exemplary implementation methods of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.
[0035] In the embodiment of the present invention, "KeepaLive" is a mechanism that can detect dead connections; "nginx" is a high-performance HTTP and reverse proxy web server; "Vue" is a front-end framework;
[0036] "ES6" is a front-end technology; "SASS" is a cascading style sheet language; "Echarts" is a data visualization chart library; "Element-UI" is a user interaction component library; "Restful" is a design style and development method for network applications; "Kafka" is an open source stream processing platform; "RocketMQ" is a messaging system; "MySQL" is a database; "Oracle" is a relational database; "SQLServer" is a database management system; "PostgreSQL" is a relational database management system; "NoSQL" is a non-relational database; "Redis" is a key-value storage system; "MongoDB" is a database based on distributed file storage; "Jackrabbit" is a document-oriented database; "rpc" is a remote procedure call technology; "Future" is a model for asynchronous task processing.
[0037] like Figure 1 As shown, the embodiment of the present invention provides a risk decision system 100 with a feature derivation function, including a feature derivation platform 110, a model center subsystem 120, a business product management subsystem 130 and a risk report generation subsystem 140, wherein:
[0038] The feature derivation platform 110 is configured to extract basic variables according to incoming data, third-party data, and internal data, generate derived variables according to the basic variables, and generate data sets according to the basic variables and the derived variables;
[0039] The model center subsystem 120 is configured to configure the risk control model according to the user selection;
[0040] The business product management subsystem 130 is configured to create a business product, associate the business product with a corresponding data set in the feature derivation platform according to user selection, create a risk control model associated with the business product, set the data set corresponding to the business product to the variable library of the risk control model, call the model center subsystem 120 according to user selection, configure the risk control model associated with the business product, and enable the business product to run;
[0041] The risk report generating subsystem 140 is configured to generate a risk report according to the operation result of the business product.
[0042] The system in the embodiment of the present invention can create business products, associate data sets with risk control models, load data sets into the variable library of the risk control model, enable business product operation after the user configures the risk control model through a visual interface, and generate risk reports based on the operation structure of the business product, thereby improving the efficiency of building a risk control system and being able to quickly respond to different business products. In the embodiment of the present invention, the business products may be credit products, credit card products, etc.
[0043] The system in some embodiments of the present invention can support the entire life cycle of consumer finance, credit, credit card, inclusive finance, corporate products, etc., and has good support for pre-loan, mid-loan and post-loan strategies and models. At the same time, the supported business products are also diverse. The system is designed from two dimensions: life cycle and business product type, and can cover the deployment of most business product strategies and models. By flexibly configuring the relevant rules in the decision engine, risk management can be performed throughout the life cycle of credit products. Before the loan, application anti-fraud, equipment anti-fraud, credit assessment, product pricing, real-time credit decision-making, etc. can be realized; during the loan, dynamic early warning of existing customers can be realized; after the loan, collection strategy formulation can be realized, etc.
[0044] In some embodiments of the present invention, the business product management subsystem 130 is further configured to perform simulation testing on the risk control model and generate simulation test results, so as to quickly verify the effectiveness of the risk control model and improve the building efficiency.
[0045] In some embodiments of the present invention, the incoming data includes transaction messages, and the feature derivation platform 110 is specifically configured to automatically parse the transaction messages to obtain basic variables in the transaction messages; based on the basic variables, derived variables are generated according to derived variable configuration rules, wherein the transaction messages include personal second-generation credit reports, corporate credit reports, and third-party reports.
[0046] By analyzing and deriving variables in the second-generation PBOC credit report, the risk department's use of PBOC variables is met. Provide online management of feature variables to achieve real-time reception, reading, calculation and service. Including unified management of feature variables, variable formulation, derivative logic configuration, real-time reception, reading and processing of variables, etc., more efficiently adapt to the implementation of the second-generation credit report, and provide solid basic data services for decision engines and model training platforms.
[0047] In some embodiments of the present invention, the feature derivation process includes:
[0048] )Analysts translate the variables required by the risk control model into specific logic based on the requirements provided by business personnel.
[0049] ) Derived variable configuration. Click the "Add a new plan" button to enter the variable plan adding page. The new variable plan is divided into 4 steps, namely 4 pages, namely: basic information, message import, basic variable modification, and feature variable configuration. On the basic variable page, fill in the basic information of the plan, including the plan name, data source, etc. The message import page implements the import of sample messages corresponding to the plan, preparing for the subsequent parsing of basic variables and configuration of variables. The basic variable modification page is used to view the basic variables after the sample message is parsed, and supports page modification. The derived variable configuration page configures derived variables based on the basic message, and supports testing, startup and shutdown, and version management functions.
[0050] ) Variables are published. After the derived variable configuration is completed, it is enabled on the variable configuration page. After the scheme is enabled, it means that the scheme and variables are in the published state. Users can import message data in real time on the variable import page and view the derived variable results on the page.
[0051] ) Feature report. The user can configure the corresponding interface through the interface management page, and the third-party system can push the message to the feature derivation platform 110 through the interface. The feature derivation platform 110 performs real-time calculations, and the synchronization interface can return the derived variable settlement results in real time to form a data set.
[0052] The feature derivation platform 110 in the embodiment of the present invention has built-in commonly used feature derivation logic and more than 16,000 built-in second-generation credit derivative variables for individuals / enterprises. The top 20 variables with the highest effectiveness can be listed in combination with the scenarios. The built-in derivative variable template system has more than 80 built-in functions to meet the derivation requirements and supports uploading custom functions to meet user requirements to the greatest extent. The derivative platform supports developers to develop custom functions using Java code and import them into the system through page import. When used during the configuration of derivative variables, development efficiency can be improved. It supports multiple ways to configure feature variables, including page configuration expression methods, SQL statement editing, JS code, Python code, etc.
[0053] For example, extract basic variables from the message data, such as the number of overdue days for loan repayments in a repayment sequence of a certain loan product {x1, x2, ... xn}, and generate derived variables {X1, X2, X3} based on derivative rules, such as the rule of obtaining the sum, the rule of obtaining the number of data, and the rule of obtaining the data variance. Generate a data set DataX = {x1, x2, ... xn, X1, X2, X3} based on the basic variables and the derived variables. After the business product management subsystem 130 creates the loan product, it is associated with the corresponding data set DataX and the risk control model M. The risk control model M is configured by calling the model center subsystem 120.
[0054] In some embodiments of the present invention, when the model center subsystem 120 configures the risk control model associated with the business product, the model center subsystem 120 is specifically configured to configure the risk control model using a policy management component, and the policy management component includes at least one of a decision set management component, a decision table management component, a decision tree management component, a scoring card management component, a machine learning model management component, and a decision flow management component.
[0055] In some embodiments of the present invention, when configuring through a decision set management component, complex rule customization can be completed by clicking the mouse; when configuring through a decision table management component, a specific complex type of rule is defined in the form of a decision table, and the condition or action configuration can be completed by clicking the mouse, which is familiar to business personnel; when configuring through a decision tree management component, business rules are presented in a visual, tree-like structure; when configuring through a scorecard management component, a graphical scorecard editor is used to maximize the fit with the scorecard in the business system; when configuring through a machine learning model management component, complex machine learning model file deployment is supported; when configuring through a decision flow management component, the orchestration and execution of several complex business rules is realized, and decision involvement is realized based on the browser, and decision flow involvement is completed by dragging the mouse. Rule sets, decision tables, scorecards, and decision trees can be bound to process nodes to realize rule orchestration, and routing judgment, serial, parallel and other industry-type process nodes are provided in the process to maximize the rule orchestration of various complex businesses.
[0056] Preferably, in some embodiments of the present invention, in order to handle concurrent product creation tasks, the business product management subsystem 130 also has an import and export function, and the business product management subsystem 130 is configured to export the business product according to the user's selection, and export the data set and risk control model associated with the business product. The business product management subsystem 130 can import the created business product into other systems for execution. After testing and verification in this system, it can be conveniently imported into the production environment for execution, which improves the efficiency of deployment.
[0057] In some embodiments of the present invention, the business product management subsystem 130 is configured to import external business products according to user selection, and import data sets and risk control models associated with the external business products.
[0058] In some embodiments of the present invention, the risk report generation subsystem 140 can automatically generate a risk report based on the operating results of the business product. The risk report includes the results of the transaction, transaction input data, third-party data, internal data and hit rules, etc., and can also generate corresponding charts, PDF files, etc.
[0059] In some embodiments of the present invention, multiple methods are provided for model optimization. The system also includes a model optimization subsystem 150. The model optimization subsystem 150 is configured to optimize the risk control model according to an optimization mode, and the optimization mode includes at least one of the following optimization modes.
[0060] 1) AB testing mode, the optimization process is: Model A is the model used by the current business product, Model B is a new model, and a transaction data is executed by both Model A and Model B at the same time, which will not affect the normal business. The execution results of Model A and Model B can be compared to find a better model.
[0061] 2) Champion-Challenger Mode: The optimization process is as follows: the champion is the model used by the current business product, and the challenger is a new model. Multiple challengers can be configured, and then an execution ratio can be set. For each input, the model to be executed can be calculated based on the designed ratio. Only one model will be executed at the same time for each input.
[0062] 3) Rule backtracking mode, the optimization process is: run the new model with the previous input data.
[0063] The system in the embodiment of the present invention can support the creation of consumer finance, credit, credit card, inclusive finance, corporate products and other products throughout the entire life cycle, and can well support pre-loan, mid-loan and post-loan strategies and models. At the same time, the supported business products are also diverse. The system is designed from two dimensions: life cycle and business product type, and can cover the deployment of most business product strategies and models. By flexibly configuring the relevant rules in the decision engine, risk management can be performed throughout the life cycle of credit products. Before the loan, application anti-fraud, equipment anti-fraud, credit assessment, product pricing, real-time credit decision-making, etc. can be realized; during the loan, dynamic early warning of existing customers can be realized; after the loan, collection strategy formulation can be realized, etc.
[0064] The system in the embodiment of the present invention supports the flexible deployment, update, maintenance and management of various scoring models and strategies, reduces the uncontrollable human factors in the manual credit process, gradually improves the standardization and differentiation management capabilities of risk models and strategies before, during and after the loan business, and improves the automation of model and strategy operation. Detailed function description is as follows:
[0065] 1) Support risk strategy model management, including risk strategy model definition, maintenance, and release, etc., and can iterate the strategy model and complete the release in real time.
[0066] 2) Supports a variety of policy management components, such as rules / rule sets, decision tables, decision trees, scoring models, data calculations, decision flows, etc., and supports corresponding management of validity periods, policy priorities, etc.
[0067] ① Rules: Organize decision logic according to logical expression statements, such as "if...then..."
[0068] ②Rule set: A set of rules organized based on unified input parameters and processing logic scope.
[0069] ③Decision tree: Use tree branch graphics to calculate: Process and calculate through input variables or application form data elements to obtain the parameters required for subsequent rule judgment or risk calculation, and can be used in other decision components.
[0070] ④Decision table: Use tabular graphics to organize decision logic.
[0071] ⑤ Decision flow: Arrange multiple decision components in a certain way to realize the overall decision logic.
[0072] ⑥ Scoring model: supports the scoring card model, evaluates and quantifies credit risk according to the scoring card method and obtains specific scores.
[0073] ⑦Champion / Challenger Mode: You can formulate a champion / challenger implementation mode and specify multiple challenger strategies. The system will conduct statistical evaluation on the execution efficiency and operation effect of multiple challenger strategies and upgrade the optimal challenger strategy to the champion strategy.
[0074] ⑧Supports graphical operation and has a beautiful UI interface. Users can build models by dragging and dropping. Model creation can be completed without programming experience. It supports organizing decision logic by dragging and dropping to select data sources.
[0075] 3) Supports algorithm selection and decision flow configuration. You can configure parameters, filter variables, perform simple data processing, view the algorithm execution process, etc. on the interface.
[0076] 4) Support flexible and complex data structures, including business object models based on XML schemas or business object models expressed using hierarchical nested Java objects. For example, the complex hierarchical data structure of the PBOC credit report can be implemented so that policy managers can create derived variables based on the detailed data of the PBOC credit report.
[0077] 5) Supports permission management, and can flexibly formulate permission management strategies to limit the scope of rules that roles can access. Permission control can be performed based on the rule base directory / package. A user can be granted multiple roles.
[0078] 6) Support change approval management. The system supports approval management of changed policy models. The changed policy model can only be released after approval by relevant authorized users.
[0079] 7) Supports auditing of operation logs of all rule bases and various detailed queries, including which user logged into the rule base at what time and what operations were performed.
[0080] Preferably, the system in the embodiment of the present invention includes:
[0081] 1. Feature derivation platform: The source of the variables used in the model may come from the input of some other business systems, or query the interface provided by the third party and query some databases. This subsystem provides unified integration and management for these data, and can manage the field verification rules, set the input fields, and maintain the data products. The feature derivation platform in the embodiment of the present invention can perform regular expression verification configuration on the input fields, perform unified and centralized maintenance of the input fields, and maintain internal and external data products in a unified and centralized manner. It supports multiple modes and is configured through XML files to reduce the later code development.
[0082] 2. Business product management subsystem: Different business products are created for different business scenarios. The business products are associated with the input, internal and external data, and machine learning models used in the scenario, and a model project is automatically created to associate with the business product. The data selected in the business product is automatically created in the model's variable library, which facilitates the direct use of variables for modeling; business products can automatically generate API files to provide external services and provide simulation testing functions; business products provide import and export functions, and when exporting, the input, internal and external, machine learning models and associated models selected by the business product will be exported. Therefore, when the production environment is launched, you only need to export the business products tested in the test environment and then import them into the production environment for direct use.
[0083] The business product management subsystem in the embodiment of the present invention associates the business product with the risk control model and data set (data product) used, can perform maintenance and testing, and can port the business product to other environments. It can also enable or disable the business product and has a security control mechanism.
[0084] 3. Model center subsystem: Provides page clicks to configure models, supports multiple models, provides decision flow to orchestrate other models into a process, and provides simulation test functions to verify the accuracy of the model. When the model center subsystem configures the risk control model associated with the business product, the model center subsystem is specifically configured to use a policy management component to configure the risk control model, and the policy management component includes at least one of a decision set management component, a decision table management component, a decision tree management component, a scorecard management component, a machine learning model management component, and a decision flow management component.
[0085] In the embodiment of the present invention, when configuring through the decision set management component, complex rule customization can be completed by clicking the mouse; when configuring through the decision table management component, specific complex type rules are defined in the form of a decision table, and the condition or action configuration can be completed by clicking the mouse, which is familiar to business personnel; when configuring through the decision tree management component, business rules are presented in a visual, tree-like structure; when configuring through the score card management component, a graphical score card editor is used to maximize the fit with the score card in the business system; when configuring through the machine learning model management component, complex machine learning model file deployment is supported; when configuring through the decision flow management component, the orchestration and execution of several complex business rules are realized, and decision involvement is realized based on the browser, and decision flow involvement is completed by dragging the mouse. Rule sets, decision tables, score cards, and decision trees can be bound to process nodes to realize rule orchestration. Routing judgment, serial, parallel and other industry type process nodes are provided in the process to maximize the rule orchestration of various complex businesses.
[0086] 4. Risk report generation subsystem: For each transaction, it provides risk report query transaction details data, including transaction results, transaction entry data, queried third-party data and internal data, details of hit rules, etc., and can generate PDF files for download.
[0087] 5. Model optimization subsystem: provides multiple ways to optimize the model.
[0088] AB testing: Model A is the model used by the current business product, and model B is a new model. A transaction data is executed using both models A and B at the same time, which will not affect normal business. The results of the execution of models A and B can be compared to find a better model.
[0089] Champion Challenger: The Champion is the model used by the current business product, and the Challenger is a new model. Multiple Challengers can be configured, and then an execution ratio can be set. When an input is received, the model to be executed will be calculated based on the designed ratio. Only one model will be executed at a time for an input.
[0090] Rule backtracking: Use the previous input data to run the new model.
[0091] Preferably, the system in the embodiment of the present invention includes three parts: a load layer, a service layer, and a middleware.
[0092] In the load balancing layer, KeepaLive and nginx are used for service distribution to ensure high availability of the application.
[0093] The service layer adopts the mainstream MVC architecture and efficient data caching tools, and cooperates with front-end technologies such as Vue, ES6, SASS, Echarts, Element-UI, etc. for WEB display, and adopts SAAS development method to provide Restful service interface.
[0094] The message queue uses Kafka or RocketMQ to implement asynchronous transactions. It can be selected according to the actual situation of the customer to ensure the business decoupling of the decision engine and the business system and the stability and reliability of the business.
[0095] The relational database decision engine is compatible with MySQL and Oracle by default, and also supports adaptation to SQLServer and PostgreSQL.
[0096] NoSQL uses Redis, which can be deployed in cluster / sentinel / master-slave / standalone mode according to business needs, providing high-speed cache for incoming transactions to ensure fast response of transactions. At the same time, MongoDB and Jackrabbit are used for file storage.
[0097] The internal service uses a custom RPC remote procedure call protocol, which improves the transmission efficiency between services, reduces service performance consumption, and facilitates service management.
[0098] External services use a custom http calling protocol, mainly providing external heterogeneous environments, browser interface calls, app interface calls, third-party interface calls, etc.
[0099] Asynchronous task processing uses the Future mode to execute model confrontation and model backtracking strategy tasks, obtain report difference comparison results, and adjust rules to optimize the model.
[0100] like Figure 2 As shown, from the perspective of business personnel, when using the system in the embodiment of the present invention, the operation steps are as follows:
[0101] S210: In the feature derivation platform, the required input fields, internal data and external data are configured into the system.
[0102] S220: Create a business product and select the required incoming, internal, and external data.
[0103] S230: Click Model Configuration on the created business product to enter the menu of the model center, configure the model by clicking on the page, create a new policy package in the policy package in the model management, reference the configured model file in the policy package, and verify the accuracy of the model through the simulation test in the policy package.
[0104] S240: Select the policy package created in the model management on the business product, enable the business product, and call the business product through the API file of the business product.
[0105] S250: View transaction details in the risk report.
[0106] The embodiment of the present invention proposes a risk decision system with a feature derivation function, establishes a business product in the system, and associates the application data and the risk control model. By selecting the corresponding risk control model, the configuration is performed, and the application data is associated with the risk control model, the business product is started, and a risk report is generated according to the operation result of the business product, so as to realize a configurable risk decision system with a feature derivation function, improve the risk decision response capability for different business products, and can quickly realize risk decisions for different business products, reducing the need for repeated construction of risk control systems. Basic variables are extracted by applying the application data, and variables are derived based on the derivation rules, which enriches the data type of the risk control model decision, and the decision result of the risk control model is more accurate. The decision system in the embodiment of the present invention can support the credit risk control business of the whole industry, and be deployed locally to realize the unified integration, operation and management of rules, data and models, provide a friendly operation page, and provide risk control strategy configuration operation for risk control personnel. The system can interact with the business system through an interface to provide decision support for the credit approval system. Through the integration and application of internal and external credit data, flexible setting of scenario-based risk control strategies and combination of various auxiliary technologies, we will comprehensively improve the risk control decision-making capabilities of the entire life cycle of credit business in the fields of consumer finance, credit, credit cards, inclusive finance, corporate finance, etc., and at the same time achieve the accumulation, reuse and management of the bank's credit decision-making capabilities, continuously improve demand research and development efficiency, and respond quickly to market changes.
[0107] like Figure 3As shown, the embodiment of the present invention also provides a risk decision method, which is applied to a risk decision system with a feature derivation function. The risk decision system with a feature derivation function includes a feature derivation platform, a model center subsystem, a business product management subsystem, and a risk report generation subsystem. The method includes:
[0108] S310: The feature derivation platform extracts basic variables according to the incoming data, third-party data, and internal data, generates derived variables according to the basic variables, and generates a data set according to the basic variables and the derived variables;
[0109] S320: The model center subsystem configures the risk control model according to the user selection;
[0110] S330: the business product management subsystem creates a business product, associates the business product with a corresponding data set in the feature derivation platform according to user selection, creates a risk control model associated with the business product, sets the data set corresponding to the business product to a variable library of the risk control model, calls the model center subsystem according to user selection, configures the risk control model associated with the business product, and enables the business product to run;
[0111] S340: The risk report generating subsystem generates a risk report according to the operation result of the business product.
[0112] In some embodiments of the present invention, the risk decision system with feature derivation function also includes a model optimization subsystem, such as Figure 4 As shown, the method includes:
[0113] S410: The feature derivation platform extracts basic variables according to the incoming data, third-party data, and internal data, generates derived variables according to the basic variables, and generates a data set according to the basic variables and the derived variables;
[0114] S420: The model center subsystem configures the risk control model according to the user selection;
[0115] S430: the business product management subsystem creates a business product, associates the business product with a corresponding data set in the feature derivation platform according to user selection, creates a risk control model associated with the business product, sets the data set corresponding to the business product to a variable library of the risk control model, calls the model center subsystem according to user selection, configures the risk control model associated with the business product, and enables the business product to run;
[0116] S440: The risk report generating subsystem generates a risk report according to the operation result of the business product.
[0117] S450: The model optimization subsystem optimizes the risk control model according to an optimization mode, where the optimization mode includes at least one of an AB testing mode, a champion challenger mode, and a rule backtracking mode.
[0118] The operating steps and effects of each subsystem in the method of the embodiment of the present invention refer to the operating steps and effects of the corresponding subsystem in the above system, and will not be described in detail here.
[0119] In an embodiment of the present invention, an electronic device is provided, including: a processor and a memory storing a computer program, wherein the processor is configured to execute any risk decision-making method of the embodiment of the present invention when running the computer program.
[0120] Figure 5 A schematic diagram of an electronic device 500 that can implement a method or implement an embodiment of the present invention is shown, and in some embodiments, more or fewer electronic devices than shown may be included. In some embodiments, a single or multiple electronic devices may be used for implementation. In some embodiments, a cloud or distributed electronic device may be used for implementation.
[0121] like Figure 5 As shown, the electronic device 500 includes a processor 501, which can perform various appropriate operations and processes according to the programs and / or data stored in the read-only memory (ROM) 502 or the programs and / or data loaded from the storage part 508 to the random access memory (RAM) 503. Processor 501 can be a multi-core processor, or it can include multiple processors. In some embodiments, processor 501 can include a general main processor and one or more special coprocessors, such as a graphics processing unit (GPU), a neural network processor (NPU), a digital signal processor (DSP), etc. In RAM503, various programs and data required for the operation of electronic device 500 are also stored. Processor 501, ROM502 and RAM503 are connected to each other via bus 504. Input / output (I / O) interface 505 is also connected to bus 504.
[0122] The processor and the memory are used together to execute the program stored in the memory. When the program is executed by the computer, the steps or functions of the risk decision-making method described in the above embodiments can be implemented.
[0123] The following components are connected to the I / O interface 505: an input section 506 including a keyboard, a mouse, a touch screen, etc.; an output section 507 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, a modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as needed. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as needed, so that a computer program read therefrom is installed into the storage section 508 as needed. Figure 5 Only some components are schematically shown in the figure, which does not mean that the computer system 500 only includes Figure 5 Components shown.
[0124] In some embodiments, the electronic device 1020 refers to a mobile terminal, including a mobile phone, a car terminal, a smart TV, etc. Taking a mobile phone as an example, the electronic device 500 also includes a display screen with a touch function, an external speaker, a gyroscope, a camera, a 4G / 5G antenna and other device modules.
[0125] The systems, devices, modules or units described in the above embodiments may be implemented by a computer or its associated components. The computer may be, for example, a mobile terminal, a smart phone, a personal computer, a laptop computer, a vehicle-mounted human-computer interaction device, a personal digital assistant, a media player, a navigation device, a game console, a tablet computer, a wearable device, a smart TV, an Internet of Things system, a smart home, an industrial computer, a server or a combination thereof.
[0126] Although not shown, in an embodiment of the present invention, a storage medium is provided, wherein the storage medium stores a computer program, and the computer program is configured to execute any risk decision-making method of the embodiments of the present invention when executed.
[0127] Storage media in embodiments of the present invention include permanent and non-permanent, removable and non-removable items that can be used to store information by any method or technology. Examples of storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.
[0128] The methods, programs, systems, devices, etc. of the embodiments of the present invention may be executed or implemented in a single or multiple networked computers, or may be practiced in a distributed computing environment. In the embodiments of this specification, in these distributed computing environments, tasks may be performed by remote processing devices connected via a communication network.
[0129] Those skilled in the art should understand that the embodiments of the present specification may be provided as methods, systems or computer program products. Therefore, those skilled in the art may imagine that the functional modules / units or controllers and related method steps described in the above embodiments may be implemented in software, hardware or a combination of software / hardware.
[0130] Unless explicitly stated, the actions or steps of the methods, programs, and embodiments of the present invention do not have to be performed in a specific order and can still achieve the desired results. In some implementations, multitasking and parallel processing are also possible or may be advantageous.
[0131] In this article, multiple embodiments of the present invention are described, but for the sake of brevity, the description of each embodiment is not exhaustive, and the same or similar features or parts between the embodiments may be omitted. In this article, "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" are meant to be applicable to at least one embodiment or example according to the present invention, but not all embodiments. The above terms do not necessarily mean to refer to the same embodiment or example. In the absence of contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples.
[0132] The exemplary systems and methods of the present invention have been specifically shown and described with reference to the above embodiments, which are merely examples of the best modes for implementing the present systems and methods. It will be appreciated by those skilled in the art that various changes may be made to the embodiments of the systems and methods described herein when implementing the present systems and / or methods without departing from the spirit and scope of the present invention as defined in the appended claims.
Claims
1. A risk decision system with feature derivation function, characterized in that: The risk decision-making system includes a feature derivation platform, a model center subsystem, a business product management subsystem and a risk report generation subsystem, wherein: The feature derivation platform is configured to extract basic variables based on incoming data, third-party data, and internal data, generate derived variables based on the basic variables, and generate data sets based on the basic variables and the derived variables; The model center subsystem is configured to configure the risk control model according to user selection; The business product management subsystem is configured to create a business product, associate the business product with a corresponding data set in the feature derivation platform according to user selection, create a risk control model associated with the business product, set the data set corresponding to the business product to a variable library of the risk control model, call the model center subsystem according to user selection, configure the risk control model associated with the business product, and enable the business product to run; The risk report generating subsystem is configured to generate a risk report according to the operation result of the business product.
2. The system according to claim 1, characterized in that The business product management subsystem is also configured to perform simulation testing on the risk control model and generate simulation test results.
3. The system according to claim 1, characterized in that The incoming data includes transaction messages, and the feature derivation platform is specifically configured to automatically parse the transaction messages to obtain basic variables in the transaction messages; based on the basic variables, derived variables are generated according to derived variable configuration rules, wherein the transaction messages include personal second-generation credit reports, corporate credit reports and third-party reports.
4. The system according to claim 1, characterized in that The business product management subsystem is configured to export the business product according to user selection, and export the data set and risk control model associated with the business product.
5. The system according to claim 1, characterized in that The business product management subsystem is configured to import external business products according to user selection, and import data sets and risk control models associated with the external business products.
6. The system according to claim 1, characterized in that When the model center subsystem configures the risk control model associated with the business product, the model center subsystem is specifically configured to configure the risk control model using a policy management component, and the policy management component includes at least one of a decision set management component, a decision table management component, a decision tree management component, a scoring card management component, a machine learning model management component, and a decision flow management component.
7. The system according to claim 1, characterized in that The risk report includes the transaction results, transaction input data, third-party data, internal data and hit rules.
8. The system according to claim 1, characterized in that The system also includes a model optimization subsystem, which is configured to optimize the risk control model according to an optimization mode, wherein the optimization mode includes at least one of an AB testing mode, a champion challenger mode, and a rule backtracking mode.
9. A risk decision-making method, characterized in that: Applied to a risk decision system with a feature derivation function, the risk decision system with a feature derivation function includes a feature derivation platform, a model center subsystem, a business product management subsystem and a risk report generation subsystem, the method includes: The feature derivation platform extracts basic variables based on incoming data, third-party data, and internal data, generates derived variables based on the basic variables, and generates data sets based on the basic variables and the derived variables; The model center subsystem configures the risk control model according to the user's selection; The business product management subsystem creates a business product, associates the business product with the corresponding data set in the feature derivation platform according to the user's selection, creates a risk control model associated with the business product, sets the data set corresponding to the business product to the variable library of the risk control model, calls the model center subsystem according to the user's selection, configures the risk control model associated with the business product, and enables the business product to run; The risk report generating subsystem generates a risk report according to the operation result of the business product.
10. The method according to claim 9, characterized in that The risk decision-making system with feature derivation function also includes a model optimization subsystem, and the method also includes: The model optimization subsystem optimizes the risk control model according to an optimization mode, and the optimization mode includes at least one of an AB testing mode, a champion challenger mode, and a rule backtracking mode.
11. A computer-readable storage medium having a computer program stored thereon, wherein: When the program is executed by a processor, the method according to any one of claims 9 to 10 is implemented.
12. An electronic device, characterized in that: include: A processor and a memory storing a computer program, wherein the processor is configured to execute the method according to any one of claims 9 to 10 when running the computer program.
Citation Information
Cited By
Risk feature management method and device in risk business, equipment and medium
CN120146995A
Financial risk intelligent risk control system based on ABTest technology
CN122312285A
