Model development method and device, equipment, storage medium and product

Abstract

The invention discloses a model development method and device, equipment, a storage medium and a product, and relates to the technical field of model development. According to one or more technical schemes provided by the method, data source information is obtained according to a pre-selected data import mode; performing step arrangement based on the data source information and a pre-configured scene toolbox to generate a risk control rule model; and in response to a user operation instruction, carrying out model trial calculation on the risk control rule model, and carrying out development online, a set of comprehensive modeling workflow standardization framework formed through the scheme covers each link from real-time source data integration to model construction, model debugging trial calculation and online evaluation to final production deployment online, and the operation efficiency is improved. Therefore, the risk control model construction efficiency is improved, and the modeling data resource overhead is reduced.

Description

Technical Field

[0001] The present application relates to the field of model development technology, and in particular to a model development method, device, equipment, storage medium and product. Background Art

[0002] In the current data-driven business environment, risk control model building technology has become a key tool for finance, credit and other industries that require sophisticated risk management. Despite the continuous advancement of related technologies, there are still significant barriers to entry for non-data science experts in this field. Existing risk control model building methods usually require users to have a deep understanding of structured data and master specific programming languages ​​​​(such as SQL), which is a major obstacle for users without an IT background.

[0003] In addition, the existing modeling process is quite cumbersome, covering multiple steps from data collection to model implementation. Each step consumes a lot of time and effort, and the complexity of the process also increases the risk of errors. More importantly, the existing technology relies heavily on feedback from actual operation in model debugging, resulting in potential problems that can only be discovered after the model is online and combined with production data. This lagging feedback mechanism not only increases the need for repeated adjustments, but also prolongs the project cycle and increases communication costs.

[0004] Another major problem with existing risk control models is the poor interpretability of the models. Many models lack a clear explanation mechanism, making the logical structure within the model difficult to understand, which not only affects the user experience and trust, but also reduces the user's enthusiasm for participation. Finally, due to the lack of a standardized model building process, each new model almost needs to start from scratch, and it is impossible to effectively reuse existing model features, which significantly reduces the overall modeling efficiency and productivity. These problems have led to reduced efficiency and waste of resources in the establishment of risk control models to varying degrees.

[0005] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0006] The main purpose of this application is to provide a model development method, device, equipment, storage medium and computer program product, aiming to solve technical problems corresponding to the background technology.

[0007] To achieve the above objectives, the present application proposes a model development method, which comprises:

[0008] Obtain data source information according to the pre-selected data import mode;

[0009] Arrange steps based on the data source information and the pre-configured scenario toolbox to generate a risk control rule model;

[0010] In response to user operation instructions, the risk control rule model is tested and developed and launched.

[0011] In one embodiment, the data import mode includes a database direct connection mode or an external import mode, and the step of obtaining data source information according to the preselected data import mode includes:

[0012] According to the database direct connection mode, obtaining target table information in the target database, wherein the target table information includes target table structure information and target table data information; or,

[0013] According to the external import mode, obtaining an external data file;

[0014] The target table information and / or target field information in the external data file is parsed.

[0015] In one embodiment, the step of arranging steps based on the data source information and the pre-configured scenario toolbox to generate a risk control rule model further includes:

[0016] Determine the corresponding target risk control rule scenario based on the pre-acquired risk control demand information;

[0017] Determine a target scenario matching toolbox in the scenario toolbox according to the target risk control rule scenario;

[0018] The target scenario matching toolbox is used to write screening conditions and filter data for the data source information to generate risk control steps;

[0019] A risk control rule model is obtained by combining the risk control steps.

[0020] In one embodiment, before the step of combining the risk control steps to obtain the risk control rule model, the step further includes:

[0021] In response to the step measurement and evaluation instruction, determine the corresponding target risk control step;

[0022] Execute the step script corresponding to the target risk control step to obtain the step script execution result;

[0023] Display the intermediate result table data corresponding to the target risk control step based on the step script execution result;

[0024] In response to the step reset instruction, the step is executed: the screening conditions are written and the data is screened for the data source information through the target scenario matching toolbox to generate a risk control step.

[0025] In one embodiment, the steps of performing model calculation on the risk control rule model in response to the user operation instruction and developing and launching the model include:

[0026] In response to a user operation instruction, a model script corresponding to the risk control rule model is obtained, wherein the model script includes a plurality of step scripts;

[0027] Execute the model script to obtain a first trial calculation record result;

[0028] Adjust the risk control rule model based on the risk control demand information to obtain a risk control rule adjustment model;

[0029] Obtaining and executing an adjustment model script corresponding to the risk control rule adjustment model to obtain a second trial calculation record result;

[0030] According to the first trial calculation record results and the second trial calculation record results, a target risk control rule model is selected for development and launch.

[0031] In one embodiment, the step of selecting a target risk control rule model for development and launch according to the first trial calculation record result and the second trial calculation record result includes:

[0032] Determine the risk control rule model corresponding to the result that meets the preset online condition in the first trial calculation record result and the second trial calculation record result as the target online model;

[0033] Generate an evaluation report corresponding to the target online model based on the target online model and the model evaluation template;

[0034] In response to the approval information of the evaluation trial calculation report, configuring a mapping relationship between the target online model and the corresponding production environment;

[0035] The scheduling job configuration information corresponding to the target online model is set, wherein the scheduling job configuration information includes at least one of a scheduling job name, a number of job failure retries, a retry time interval, and job-related event trigger parameters.

[0036] In addition, to achieve the above objectives, the present application also proposes a model development device, the model development device comprising:

[0037] An acquisition module, used to acquire data source information according to a pre-selected data import mode;

[0038] A generation module, used to arrange steps based on the data source information and a pre-configured scenario toolbox to generate a risk control rule model;

[0039] The online module responds to the user's operation instructions, performs model trial calculation on the risk control rule model, and develops and launches it.

[0040] In addition, to achieve the above objectives, the present application also proposes a model development device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the model development method described above.

[0041] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the model development method described above are implemented.

[0042] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the model development method described above are implemented.

[0043] One or more technical solutions proposed in this application obtain data source information according to a pre-selected data import mode; arrange steps based on the data source information and a pre-configured scenario toolbox to generate a risk control rule model; respond to user operation instructions, perform model trial calculations on the risk control rule model, and develop and launch it. A comprehensive modeling workflow standardization framework formed by the above solutions covers all aspects from real-time source data integration, model construction, model debugging and trial calculations, and online evaluation to final production deployment and launch, thereby improving the efficiency of risk control model construction and reducing modeling data resource overhead. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0045] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0046] Figure 1 A schematic diagram of a process flow provided for the first embodiment of the model development method of this application;

[0047] Figure 2 A flow chart of the second embodiment of the model development method of this application;

[0048] Figure 3 A brief flowchart of the model development method provided in Example 2 of the present application;

[0049] Figure 4 This is a schematic diagram of the module structure of the model development device of the embodiment of the present application;

[0050] Figure 5 A schematic diagram of the device structure of the hardware operating environment involved in the model development method in the embodiment of the present application.

[0051] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0052] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0053] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0054] The main solution of the embodiment of the present application is to propose a method for generating a risk control model in a visual low-threshold modeling method, aiming to simplify the construction process of the risk control model, reduce the technical threshold of modeling, and improve the modeling efficiency. The solution optimizes the speed and convenience of data acquisition by building a systematic and user-friendly risk control model generation process, supporting real-time database synchronization and external data import. In the design of the modeling process, each stage is equipped with detailed guidance and step-by-step instructions, so that users can easily build models through an intuitive drag-and-drop interface without paying attention to the complex underlying technical logic, and debug and execute each step independently. In addition, the solution also provides feature reuse strategies and model templates, reducing duplication of work, so that modelers can focus more on core innovation links. The built-in debugging and effect evaluation tools allow users to check the calculation results of the measurement model in real time during the modeling process, ensuring that the decision of each step is based on accurate data feedback. Finally, the detailed trial calculation evaluation report helps users to deeply understand the performance of different model versions, so as to select the model version with the best performance in actual applications for deployment, and realize the standardization and automation of the whole process from demand analysis to final production deployment, which significantly improves the efficiency and quality of risk control model construction.

[0055] Although the existing technology for building data analysis models has been improved day by day, for non-data experts, this field is still full of ups and downs. These challenges are mainly reflected in the following aspects:

[0056] (1) Technical barriers: Building an effective rule model usually relies on a deep understanding of structured data and proficiency in specific programming languages ​​(such as SQL). For users who are not from IT professions, this technical barrier greatly limits their participation.

[0057] (2) Complicated process: The entire life cycle of building an effective rule model, from data collection to model implementation, involves multiple steps, each of which consumes a lot of time and energy. At the same time, the complexity of the process also increases the possibility of errors in the process.

[0058] (3) High trial and error costs: Model debugging often relies on feedback from actual operation, which means that potential problems can only be discovered after the model is launched and combined with production data. This "afterthought" approach may lead to repeated adjustments, prolonging the project cycle and increasing communication costs. The inability to predict the effect evaluation analysis in advance also increases the uncertainty of modeling.

[0059] (4) Poor interpretability: Many models lack a clear explanation mechanism, and the logical structure within the model is often obscure and difficult to understand. This not only affects the user experience and trust, but also reduces the user's enthusiasm for participation.

[0060] (5) Insufficient reusability: The lack of a standardized model building process means that each new model must start from scratch and existing model features cannot be effectively reused, which significantly reduces overall modeling efficiency and productivity.

[0061] The present application provides a solution, which obtains data source information according to a pre-selected data import mode; arranges steps based on the data source information and a pre-configured scenario toolbox to generate a risk control rule model; responds to user operation instructions, performs model trial calculation on the risk control rule model, and develops and launches it. A comprehensive modeling workflow standardization framework formed by the above solution covers all aspects from real-time source data integration, model construction, model debugging trial calculation, and online evaluation to final production deployment and launch, thereby improving the efficiency of risk control model construction and reducing modeling data resource overhead.

[0062] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a model development system, etc. The following takes the model development system as an example to illustrate this embodiment and the following embodiments.

[0063] Based on this, the present application embodiment provides a model development method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the model development method of the present application.

[0064] In this embodiment, the model development method includes steps S1000 to S4000:

[0065] Step S1000: obtaining data source information according to a pre-selected data import mode;

[0066] It should be noted that in order to effectively improve the efficiency of risk control model construction and reduce the modeling data resource overhead, this embodiment proposes a method for generating a risk control model in a visual low-threshold modeling manner, and is committed to building a systematic and user-friendly risk control model generation process. The core of the solution is to simplify the process, ensure standardized modeling operations and intuitive visual interfaces, and promote the reuse of model features. While lowering the modeling threshold, it also improves the overall modeling efficiency. At the data acquisition level, this solution supports real-time database synchronization and external data import functions to optimize the speed and convenience of data acquisition. In the design of the modeling process, each stage is equipped with detailed guidance and step-by-step instructions. With the help of an intuitive drag-and-drop interface, users can easily build models without paying attention to the complex underlying technical logic, and independently debug and execute each step. This design takes into account business experts with non-technical backgrounds, so that they can easily get started without in-depth technical details, thereby greatly shortening the conversion cycle from model demand concepts to practical applications and accelerating the implementation of business insights.

[0067] The step of obtaining data source information is the basis of the entire risk control model building process. In this step, the system provides two flexible data import modes, namely database direct connection mode and external import mode, to meet the data acquisition needs of different users. In database direct connection mode, the system will directly obtain the required table structure and data information from the target database, which can ensure the real-time and accuracy of the data. In external import mode, the system allows users to upload external data files such as Excel tables, and parse the target table information and field information contained in these files, making the data import process more flexible and convenient. In addition, it should be noted that the purpose of this step is to ensure that accurate and complete data source information can be accessed during the model development process, and to provide data support for subsequent model construction and analysis.

[0068] For example, in a specific implementation, the system will first prompt the user to select a data import mode. If the user selects the database direct connection mode, the system will ask for database connection information and automatically synchronize the relevant data in the target database. If the user selects the external import mode, an Excel file can be uploaded, and the system will parse the file and extract the required data information to prepare for the next step of model development.

[0069] Step S2000: Arrange steps based on the data source information and the pre-configured scenario toolbox to generate a risk control rule model;

[0070] It should be noted that the step of arranging steps based on data source information and scenario toolboxes is the core link of risk control rule model generation. In this step, the system first determines the corresponding target risk control rule scenario based on the pre-acquired risk control demand information, and then selects the corresponding toolbox from the scenario toolbox for data screening and condition writing according to the target risk control rule scenario. In this process, the system provides a wealth of toolboxes, such as data extraction, data association, data merging, etc., to cope with different data processing needs. Through these toolboxes, users can easily filter and process data source information to generate risk control steps that meet risk control needs. In addition, it should be noted that users can also combine and adjust the generated risk control steps to form a complete risk control rule model. The purpose of this step is to enable users to build personalized risk control models in an intuitive and flexible way according to specific risk control needs.

[0071] For example, in a specific implementation, the system will first prompt the user to enter risk control demand information, and then recommend corresponding risk control rule scenarios based on this information. After the user selects a scenario, the system will display a list of toolboxes that match it. The user can select the required toolbox, such as a data extraction toolbox, to filter and process the data source. The system will generate risk control steps based on the user's screening conditions, and allow the user to combine and adjust these steps to eventually form a complete risk control rule model.

[0072] Step S3000: In response to user operation instructions, the risk control rule model is tested and developed and launched.

[0073] It should be noted that the step of performing model trial calculation on the risk control rule model in response to the user's operation instructions and developing and launching it is the last link in the model development process. In this step, the system will obtain the model script corresponding to the risk control rule model according to the user's operation instructions, and save these scripts in the preset database for execution, so as to obtain the trial calculation record results. In this process, the system allows the user to adjust the risk control rule model based on the risk control demand information, and re-execute the adjusted model script to obtain new trial calculation record results. The system will select the risk control rule model that meets the preset online conditions for development and launch based on these trial calculation record results. In addition, it should be noted that before the model goes online, the system will also require the user to generate an evaluation trial calculation report based on the target online model and the model evaluation template, and configure the mapping relationship between the model and the production environment and the scheduling job configuration information according to the approval results of the report to ensure the smooth launch and operation of the model.

[0074] For example, in a specific implementation, a user can select an already constructed risk control rule model through the system interface and initiate a trial calculation operation. The system will automatically obtain the scripts of the model, execute these scripts, and display the trial calculation results. If the user is not satisfied with the trial calculation results, the model can be adjusted and then the trial calculation can be performed again. When the trial calculation results meet expectations, the user can initiate an online operation. The system will guide the user to generate an evaluation trial calculation report, and configure the mapping relationship between the model and the production environment and the scheduling job information, such as the job name, number of retries, and retry interval, based on the approval results of the report. After completing these configurations, the system will update the model status to online, and execute the model script regularly according to the scheduling job configuration, thereby realizing the automated operation of the model.

[0075] In a feasible implementation manner, the data import mode includes a database direct connection mode or an external import mode, and step S1000 may include steps S1100 to S1300:

[0076] Step S1100: acquiring target table information in a target database according to the database direct connection mode, wherein the target table information includes target table structure information and target table data information; or,

[0077] Step S1200: acquiring an external data file according to the external import mode;

[0078] Step S1300: parsing target table information and / or target field information in the external data file.

[0079] It should be noted that data import is the first step in the model development method, and its purpose is to provide the necessary data foundation for the construction of the risk control model. In this embodiment, the system provides flexible data import modes, including database direct connection mode and external import mode, to adapt to different data acquisition needs. The database direct connection mode allows the system to directly obtain table structure and data information from the target database. The advantage of this method is that it can ensure the real-time and accuracy of the data, and is suitable for scenarios with high requirements for data real-time. Relatively speaking, the external import mode provides a kind of flexibility, allowing users to upload external data files, such as Excel tables, and parse the required data information from them, which is suitable for situations where data sources are scattered or data in a specific format is required. Through these two modes, the system can flexibly obtain and integrate the required data source information, laying a solid foundation for the subsequent construction of the risk control model.

[0080] In addition, it should be noted that after obtaining the data source information, the system will arrange the steps according to the pre-configured scenario toolbox to generate a risk control rule model. This step involves in-depth analysis and processing of the data source information, including the writing of screening conditions and data screening, to generate risk control steps that meet risk control requirements. In this way, the system can simplify the complex data processing and model building process, so that non-professional IT users can easily participate in the construction of risk control models.

[0081] For example, in a feasible implementation, the system first prompts the user to select a data import mode. If the user selects the database direct connection mode, the system will require the input of database connection information and automatically synchronize the relevant data in the target database. If the user selects the external import mode, an Excel file can be uploaded, and the system will parse the file and extract the required data information. Subsequently, the system will determine the corresponding risk control rule scenario based on the risk control demand information input by the user, and select a suitable toolbox from the scenario toolbox for data processing. Users can use these toolboxes to filter and process data by dragging and dropping, generate risk control steps, and finally combine them into a risk control rule model. During the model building process, users can test each step at any time, view and analyze the calculation results to verify the effect of the model. After a series of trial calculations and adjustments, users can select a trial calculation version with the best effect for development and launch. Before going online, the system will require users to generate an evaluation trial calculation report, and configure the mapping relationship between the model and the production environment and the scheduling job information according to the approval results of the report, so as to realize the automatic operation of the model. This process not only improves the efficiency of risk control model construction, but also reduces the technical threshold, making the construction of risk control models more systematic and user-friendly.

[0082] In a feasible implementation manner, step S2000 may include steps S2100 to S2400:

[0083] Step S2100: determining a corresponding target risk control rule scenario based on the pre-acquired risk control demand information;

[0084] Step S2200: determining a target scenario matching toolbox in the scenario toolbox according to the target risk control rule scenario;

[0085] Step S2300: Using the target scenario matching toolbox, the data source information is subjected to screening condition writing and data screening to generate risk control steps;

[0086] Step S2400: A risk control rule model is obtained by combining the risk control steps.

[0087] It should be noted that in this embodiment, first, the system will determine the corresponding target risk control rule scenario based on the pre-acquired risk control demand information. This process involves an in-depth understanding of business needs in order to accurately define the problems that the model needs to solve. Subsequently, the system will select a suitable target scenario matching toolbox from the scenario toolbox based on the target risk control rule scenario. The scenario toolbox is a set of predefined tools for specific risk control scenarios that can help users quickly write screening conditions and filter data for data source information to generate risk control steps. The purpose of this step is to convert risk control requirements into specific data processing and analysis steps, thereby building a risk control model that meets specific needs.

[0088] In addition, it should be noted that the generated risk control steps will be combined to form a complete risk control rule model. In this process, the system may use functions such as data extraction, data association, data merging, grouping and aggregation, sequence number generation, data update, etc. These functions are provided to users through a visual interface, allowing users to arrange and combine these steps through simple drag and drop operations without having to deeply understand the underlying technical logic. In addition, the system also provides model script preview and measurement evaluation functions, allowing users to verify the model's calculation results in real time and adjust the model accordingly to ensure the accuracy and effectiveness of the model.

[0089] For example, in a feasible implementation, the system first collects risk control demand information through interaction with the user, and determines the risk control rule scenarios that need attention. Then, the system displays a series of toolboxes that match the scenario, and the user can select the appropriate toolbox for data processing according to the needs. In this embodiment, the user selects a toolbox for credit risk assessment, and uses the data screening and condition writing functions therein to extract key financial indicators and credit records from the data source and generate a series of risk control steps. These steps are then combined into a complete risk control model, and the user adjusts and optimizes the model through the system's visual interface. In the model trial calculation stage, the system saves and executes the model script and generates trial calculation record results. The user further adjusts the model based on these results and performs another trial calculation. Finally, based on the trial calculation results and model evaluation report, the user selects a model version with the best performance for development and launch. Before launching, the system guides the user to complete the mapping relationship configuration between the target launch model and the production environment, as well as the scheduling job configuration, to ensure that the model can run smoothly in the production environment. This process not only improves the efficiency of risk control model construction, but also lowers the technical threshold, making the construction of risk control models more systematic and user-friendly.

[0090] This embodiment provides a model development method, which obtains data source information according to a pre-selected data import mode; arranges steps based on the data source information and a pre-configured scenario toolbox to generate a risk control rule model; responds to user operation instructions, performs model trial calculation on the risk control rule model, and develops and launches it. A comprehensive modeling workflow standardization framework formed by the above scheme covers all aspects from real-time source data integration, model construction, model debugging trial calculation, and online evaluation to final production deployment and launch, thereby improving the efficiency of risk control model construction and reducing modeling data resource overhead.

[0091] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 2 The model development method further includes steps A1000 to A4000 before step S2400:

[0092] Step A1000: In response to the step measurement and evaluation instruction, determine the corresponding target risk control step;

[0093] Step A2000: Execute the step script corresponding to the target risk control step to obtain the step script execution result;

[0094] Step A3000: displaying the intermediate result table data corresponding to the target risk control step based on the step script execution result;

[0095] Step A4000: In response to the step reset instruction, execute the step: write the screening conditions and screen the data source information through the target scenario matching toolbox to generate the risk control step.

[0096] It should be noted that in order to ensure the accuracy and effectiveness of each step in the risk control model, the system will first evaluate the individual risk control steps before building the complete model. The purpose of this evaluation instruction is to verify whether each step can independently complete its predetermined data processing and analysis tasks. The system will identify the risk control steps that need to be evaluated and execute the corresponding scripts one by one to collect the execution data and results of each step. This measurement and evaluation of individual steps helps developers accurately identify problems and optimize them in a targeted manner.

[0097] The role of the execution subject system in this link is to automatically execute the user's measurement and evaluation instructions, identify the risk control steps that need to be measured, and run the corresponding scripts. In this way, the system can collect detailed execution results of each risk control step, including data processing and analysis output. These results are crucial for evaluating the performance of the model because they provide specific information about the performance of the model when processing actual data. If it is found that a certain step fails to produce the expected output, the user can adjust the parameters or logic of the step in time, and then re-execute the measurement and evaluation until the step can be accurately executed. Based on the data in the intermediate result table, if the user finds that any step fails to achieve the expected effect, the corresponding step can be reset. The system will respond to the step reset instruction and reuse the target scenario matching toolbox to re-write the screening conditions and filter the data.

[0098] In addition, it should be noted that the goal of this process is to optimize the performance of the risk control model through careful evaluation and adjustment of the risk control steps. In this way, the system can ensure that the risk control model can accurately identify risks when processing actual data, thereby providing support for decision-making.

[0099] For example, in a feasible implementation, the user designs a risk control model through the visual interface of the system, and hopes to evaluate specific risk control steps in the model before putting the entire model into production. The user first issues a measurement and evaluation instruction, and the system immediately determines the risk control steps that the user wants to evaluate. These steps may involve specific data screening, feature calculation, or rule application, etc. Next, the system executes the scripts of these risk control steps and generates detailed execution results. The user can view these results and analyze whether the output of each step meets expectations, whether the data processing is correct, and whether it can effectively identify risks. Based on these evaluation results, the user can make necessary adjustments to the risk control steps, such as modifying the screening conditions, adjusting the feature calculation method, or optimizing the rule logic. After a series of measurements and adjustments, the user finally obtains a well-performing risk control model and decides to put it into production. In the model trial calculation stage, the system saves and executes the complete model script and generates trial calculation record results. Based on these results and the evaluation trial calculation report generated based on the model evaluation template, the user finally selects a model version that meets the preset online conditions for development and launch. Before going online, the system also guides users to complete the mapping relationship configuration between the target online model and the production environment, as well as the scheduling job configuration, to ensure that the model can run stably in the production environment. This process not only improves the quality and reliability of the risk control model, but also makes the model development and online process more efficient and controllable.

[0100] In a feasible implementation manner, step S3000 may include steps S3100 to S3500:

[0101] Step S3100: In response to a user operation instruction, a model script corresponding to the risk control rule model is obtained, where the model script includes a plurality of step scripts;

[0102] Step S3200: executing the model script to obtain a first trial calculation record result;

[0103] Step S3300: adjusting the risk control rule model based on the risk control demand information to obtain a risk control rule adjustment model;

[0104] Step S3400: obtaining and executing an adjustment model script corresponding to the risk control rule adjustment model to obtain a second trial calculation record result;

[0105] Step S3500: Select a target risk control rule model for development and launch based on the first trial calculation record results and the second trial calculation record results.

[0106] It should be noted that the model trial calculation and development and launch process are key links to ensure the accuracy and practicality of the risk control model. This process begins when the user issues an operation instruction through the visual interface, and the system then obtains the model script associated with the risk control rule model. The model script is a series of predefined calculation and logical judgment instructions that form the basis for the operation of the model. The purpose of obtaining these scripts is to execute the model on actual data to verify the logic and output of the model.

[0107] In addition, it should be noted that the system saves these model scripts to a preset database and executes them to generate the first round of trial calculation record results. These results provide the preliminary performance of the model when processing actual data and are an important basis for evaluating model performance. Based on these preliminary results, if it is found that the model deviates from the risk control requirements, the user can adjust the risk control rule model and generate an adjusted model. Subsequently, the system obtains the script for adjusting the model and executes it to obtain the second round of trial calculation record results. This second round of trial calculation results reflects the effect of the model adjustment and provides users with a basis for further optimizing the model.

[0108] After at least two rounds of trial calculations, users can select the risk control rule model with the best performance for development and launch based on the results of two or more trial calculations. The criteria for selecting the model are based on the preset launch conditions, which may include the accuracy, stability, and response time of the model. The model that meets these conditions will be determined as the launch model, and this model will be mapped to the production environment and the corresponding scheduling job configuration information will be set to ensure that the model can run stably in the production environment.

[0109] For example, in a feasible implementation, the user issues a trial calculation instruction to the risk control model through the system interface. The system automatically extracts the corresponding model scripts based on the risk control rule model defined by the user and saves these scripts in the database. After executing these scripts, the system records the results of the first round of trial calculations, which reflect the performance of the model under the current configuration. If necessary, the user can fine-tune the model based on the feedback from the first round of trial calculations, and the system then executes the adjusted model script and records the results of the second round of trial calculations. By comparing the results of the two rounds of trial calculations, the user selects the optimal model version for online launch. The system then guides the user to complete the configuration of the online model and the production environment, including setting the job name, retry strategy, and trigger parameters, to ensure the smooth execution and monitoring of the model in the production environment. This process not only improves the quality and reliability of the model, but also ensures that the model can meet actual business needs.

[0110] In a feasible implementation manner, step S3500 may include steps S3510 to S3540:

[0111] Step S3510: Determine the risk control rule model corresponding to the result that meets the preset online condition in the first trial calculation record result and the second trial calculation record result as the target online model;

[0112] Step S3520: generating an evaluation report corresponding to the target online model based on the target online model and the model evaluation template;

[0113] Step S3530: In response to the approval information of the evaluation trial calculation report, configuring a mapping relationship between the target online model and the corresponding production environment;

[0114] Step S3540: Setting the scheduling job configuration information corresponding to the target online model, wherein the scheduling job configuration information includes at least one of the scheduling job name, the number of job failure retries, the retry time interval, and job-related event trigger parameters.

[0115] It should be noted that the development and launch process is a series of operations to deploy the risk control model from the development environment to the production environment. The purpose of this process is to ensure that the risk control model can run stably in actual business and achieve the expected results. The system first compares the first trial calculation record results with the second trial calculation record results, and selects the risk control rule model that meets the preset launch conditions as the target launch model. The preset launch conditions may include the accuracy, stability and performance indicators of the model, which are key factors in measuring whether the model is suitable for launch.

[0116] In addition, it should be noted that the system generates an evaluation report based on the target online model and the model evaluation template. The report records the performance and effect of the model in detail, providing an important basis for the approval of the model. After the evaluation report is approved, the system will configure the mapping relationship between the target online model and the production environment to ensure that the model can correctly obtain data and execute in the production environment. This process may involve mapping of data sources, configuration of interfaces for external services, etc., to ensure that the model can run smoothly in the production environment.

[0117] Next, the system sets the scheduling job configuration information corresponding to the target online model, including the job name, number of failed retries, retry interval, trigger parameters, etc. These configuration information ensures the scheduled execution and exception handling of the model in the production environment, thereby ensuring the continuity and stability of the model operation.

[0118] For example, in a feasible implementation, after the user issues an online instruction through the system interface, the system automatically compares the results of the two rounds of trial calculations and selects the model that meets the preset conditions as the target online model. Then, the system uses the evaluation template to generate a detailed evaluation report, which contains the performance data and business indicators of the model for review by the approval personnel. Once approved, the system will configure the production environment, including data source mapping and interface configuration, to ensure that the model can run seamlessly in the production environment. Finally, the system sets up the model's scheduling job, including the job name, failure retry strategy, and trigger time, to ensure that the model can be automatically executed as planned and retried when an exception occurs. Through this series of operations, the risk control model is successfully deployed to the production environment to provide support for business decisions.

[0119] For example, to help understand the implementation process of the model training method obtained by combining this embodiment with the above-mentioned embodiment 1, please refer to Figure 3 , Figure 3 A brief flowchart of a model training method is provided, specifically:

[0120] In this embodiment, the entire process of model development mainly includes four modules, namely, a data import module, a model development module, a model verification module and a model production module.

[0121] The data import module provides two ways to obtain data source information. One is to directly connect to the trial calculation database and directly synchronize the table structure, table data and other related information in the database according to business needs. The other is to import through external data files (such as Excel tables). After the import is successful, you can still maintain and adjust the table, fields and other related information.

[0122] The model development module mainly builds rule models through step arrangement. In the step arrangement, each step provides a rich toolbox for dealing with different rule scenarios, including data extraction, data association, data merging, grouping and aggregation, sequence number generation, data update and other toolboxes. Each toolbox generates a complete step after selecting a table, selecting a field and writing a screening condition. This process can be completed without paying attention to the complex underlying technical logic.

[0123] After the step is created, you can view the generated script statements through the model script preview in the interface. You can select any complete step that has been generated to measure and evaluate it. The system will execute the script corresponding to this step separately. When the step status changes from waiting for execution to completed, it means that the step is executed successfully. You can view the intermediate result table data of the current step through the model result data preview. For multiple steps in the orchestration, the dependencies between the steps can be indicated by connecting lines. If all the steps of the current rule model have been built, the model script generated by the entire rule model will be executed in the database through full execution. After successful execution, you can select any step to view the measurement results of each step, just like single-step execution.

[0124] The model verification module includes the trial calculation and approval of the model. For a model with complete steps, by adding a trial calculation, the model script is saved and executed in the database. During this period, you can still return to the model development module to adjust the steps of the rule model, and then continue to generate new trial calculation records and results. The results of multiple trial calculation records can be compared, and finally the trial calculation version with the best effect is selected as the online model version to be put into production. Before going into production, users can fill in relevant content based on the customizable evaluation template to generate an evaluation trial calculation report for the model. The trial calculation report measures the effectiveness of the model from various dimensions, which is convenient for the production auditor to conduct model evaluation and review, thereby ensuring the quality of each online model.

[0125] The model production module can put the approved trial calculation version into production. This step mainly configures the mapping relationship between the trial calculation and the production environment. After the configuration is completed, after confirming the final production script, the scheduling job is configured, mainly including the job name of the scheduling job, the number of retries for job failure, the time between retries, the notification user, the trigger time of the job and other parameters. After the scheduling job configuration is completed, the status of the entire model will become online. The scheduling service will trigger the execution of the corresponding production script at a fixed time according to the configured parameters, so that the entire model is put into production and launched.

[0126] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the model development method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0127] Please refer to Figure 4 The present application also provides a model development device, the model development device comprising:

[0128] An acquisition module 41 is used to acquire data source information according to a pre-selected data import mode;

[0129] A generation module 42, configured to perform step arrangement based on the data source information and a pre-configured scenario toolbox to generate a risk control rule model;

[0130] The online module 43 performs model calculation on the risk control rule model in response to user operation instructions, and develops and launches it.

[0131] The model development device provided by the present application adopts the model development method in the above embodiment to solve the technical problems of model development. Compared with the prior art, the beneficial effects of the model development device provided by the present application are the same as the beneficial effects of the model development method provided by the above embodiment, and the other technical features in the model development device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0132] The present application provides a model development device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the model development method in the above-mentioned embodiment one.

[0133] Reference below Figure 5 , which shows a schematic diagram of the structure of a model development device suitable for implementing the embodiment of the present application. The model development device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The model development device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0134] like Figure 5As shown, the model development device may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the model development device are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the model development device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a model development device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0135] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0136] The model development device provided in the present application adopts the model development method in the above-mentioned embodiment. Compared with the prior art, the beneficial effects of the model development device provided in the present application are the same as the beneficial effects of the model development method provided in the above-mentioned embodiment, and other technical features in the model development device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0137] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0138] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0139] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the model development method in the above-mentioned embodiment.

[0140] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0141] The computer-readable storage medium may be included in the model development device; or may exist independently without being assembled into the model development device.

[0142] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0143] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0144] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0145] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned model development method, and can solve the technical problems of model development. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the model development method provided in the above-mentioned embodiment, and will not be repeated here.

[0146] The present application also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the steps of the model development method as described above are implemented.

[0147] The computer program product provided in this application can solve the technical problems of model development. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the model development method provided in the above embodiment, which will not be repeated here.

[0148] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A model development method, characterized in that: The method comprises: Obtain data source information according to the pre-selected data import mode; Arrange steps based on the data source information and the pre-configured scenario toolbox to generate a risk control rule model; In response to user operation instructions, the risk control rule model is tested and developed and launched.

2. The method according to claim 1, characterized in that The data import mode includes a database direct connection mode or an external import mode, and the step of obtaining data source information according to the pre-selected data import mode includes: According to the database direct connection mode, obtaining target table information in the target database, wherein the target table information includes target table structure information and target table data information; or, According to the external import mode, obtaining an external data file; The target table information and / or target field information in the external data file is parsed.

3. The method according to claim 1, characterized in that The step of arranging steps based on the data source information and the pre-configured scenario toolbox to generate a risk control rule model also includes: Determine the corresponding target risk control rule scenario based on the pre-acquired risk control demand information; Determine a target scenario matching toolbox in the scenario toolbox according to the target risk control rule scenario; The target scenario matching toolbox is used to write screening conditions and filter data for the data source information to generate risk control steps; A risk control rule model is obtained by combining the risk control steps.

4. The method according to claim 3, characterized in that Before the step of combining the risk control steps to obtain the risk control rule model, the following steps are also included: In response to the step measurement and evaluation instruction, determine the corresponding target risk control step; Execute the step script corresponding to the target risk control step to obtain the step script execution result; Display the intermediate result table data corresponding to the target risk control step based on the step script execution result; In response to the step reset instruction, the step is executed: the screening conditions are written and the data is screened for the data source information through the target scenario matching toolbox to generate a risk control step.

5. The method according to claim 3, characterized in that The step of performing a model trial calculation on the risk control rule model in response to the user operation instruction and developing and launching the model comprises: In response to a user operation instruction, a model script corresponding to the risk control rule model is obtained, wherein the model script includes a plurality of step scripts; Execute the model script to obtain a first trial calculation record result; Adjust the risk control rule model based on the risk control demand information to obtain a risk control rule adjustment model; Obtaining and executing an adjustment model script corresponding to the risk control rule adjustment model to obtain a second trial calculation record result; According to the first trial calculation record results and the second trial calculation record results, a target risk control rule model is selected for development and launch.

6. The method according to claim 5, characterized in that The step of selecting a target risk control rule model for development and launch according to the first trial calculation record result and the second trial calculation record result comprises: Determine the risk control rule model corresponding to the result that meets the preset online condition in the first trial calculation record result and the second trial calculation record result as the target online model; Generate an evaluation report corresponding to the target online model based on the target online model and the model evaluation template; In response to the approval information of the evaluation trial calculation report, configuring a mapping relationship between the target online model and the corresponding production environment; The scheduling job configuration information corresponding to the target online model is set, wherein the scheduling job configuration information includes at least one of a scheduling job name, a number of job failure retries, a retry time interval, and job-related event trigger parameters.

7. A model development device, characterized in that: The device comprises: An acquisition module, used to acquire data source information according to a pre-selected data import mode; A generation module, used to arrange steps based on the data source information and a pre-configured scenario toolbox to generate a risk control rule model; The online module responds to the user's operation instructions, performs model trial calculation on the risk control rule model, and develops and launches it.

8. A model development device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the model development method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the model development method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the model development method according to any one of claims 1 to 6 are implemented.

Images