Training method and device of business handling evaluation model, equipment and storage medium
By constructing a business processing evaluation model, using machine learning to train historical data to determine feature importance and perform feature fusion, the problem of banks struggling to assess whether customers are suitable for wealth management services has been solved, achieving higher prediction accuracy and personalized services.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- INDUSTRIAL AND COMMERCIAL BANK OF CHINA
- Filing Date
- 2026-02-05
- Publication Date
- 2026-05-08
AI Technical Summary
Banks struggle to effectively assess whether customers are suitable for wealth management services, resulting in an inability to provide personalized financial services and impacting customer loyalty and experience.
By constructing a business processing evaluation model, using machine learning models to train historical business data, determining feature importance, and performing feature fusion, the model's prediction accuracy and analytical capabilities are improved.
It improves the adaptability of banking services to customers, provides personalized and high-quality services, and enhances customer loyalty and experience.
Smart Images
Figure CN121998750A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of computer technology and can be applied to the field of financial technology. Specifically, it relates to a training method, apparatus, device, and storage medium for a business processing evaluation model. Background Technology
[0002] Currently, most of the existing customer base of banks has wealth management needs. However, determining whether customers with wealth management needs are truly suitable for wealth management services, providing them with better wealth management services, and offering them matching wealth management products and services can not only significantly improve customer loyalty but also optimize customer experience and deepen customer relationships. This has become a core competitive advantage for commercial banks.
[0003] Therefore, there is an urgent need for an evaluation method for banking business processing. Summary of the Invention
[0004] This application provides a training method, apparatus, equipment, and storage medium for a business processing evaluation model to improve the adaptability of banking business processing to customers, provide customers with better services, and increase user stickiness.
[0005] According to one aspect of this application, a method for training a business processing evaluation model is provided, the method comprising: Obtain historical transaction data for the target banking business, and construct business sample data based on the historical transaction data; wherein, the business sample data includes real transaction tags; Based on the business sample data, the pre-built machine learning model is trained with the first model until the training end condition of the first model is met, and a candidate business processing evaluation model is obtained. Using the candidate business processing evaluation model, business processing prediction is performed on different combinations of business dimension features in the historical business processing data to obtain candidate predicted business processing labels, and the feature importance of different business dimension features in the historical business processing data is determined based on the predicted business processing labels. Based on the feature importance and the historical business processing data, the candidate business processing evaluation model is trained as a second model until the training end condition of the second model is met, thus obtaining the target business processing evaluation model.
[0006] According to another aspect of this application, a training apparatus for a business processing evaluation model is provided, the apparatus comprising: The sample generation module is used to acquire historical business processing data of the target banking business and construct business sample data based on the historical business processing data; wherein, the business sample data includes real business processing tags; The first training module is used to train a pre-built machine learning model based on the business sample data until the first model training termination condition is met, thereby obtaining a candidate business processing evaluation model. The feature importance module is used to use the candidate business processing evaluation model to predict the business processing of different business dimension feature combinations in the historical business processing data, obtain candidate predicted business processing labels, and determine the feature importance of different business dimension features in the historical business processing data based on the predicted business processing labels. The second training module is used to train the candidate business processing evaluation model based on the feature importance and the historical business processing data, until the training end condition of the second model is met, and the target business processing evaluation model is obtained.
[0007] According to another aspect of this application, an electronic device is provided, the electronic device comprising: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the training method of any of the business processing evaluation models provided in the embodiments of this application.
[0008] According to another aspect of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements a training method for any of the business processing evaluation models provided in the embodiments of this application.
[0009] According to another aspect of this application, a computer program product is provided, including a computer program that, when executed by a processor, implements a training method for any of the business processing evaluation models provided in the embodiments of this application.
[0010] This application trains a first model on a machine learning model, and then fuses the feature importance corresponding to each business dimension feature in the determined historical business processing data with their respective business dimension features. Based on the new features after fusion, a second model is trained on the machine learning model trained by the first model. This improves the ability of the machine learning model trained by the second model to analyze each extracted business dimension feature, thereby improving the model performance and prediction accuracy. Attached Figure Description
[0011] Figure 1 This is a flowchart of a training method for a business processing evaluation model provided in Embodiment 1 of this application; Figure 2This is a flowchart of a training method for a business processing evaluation model according to Embodiment 2 of this application; Figure 3 This is a schematic diagram of the structure of a training device for a business processing evaluation model according to Embodiment 3 of this application; Figure 4 This is a schematic diagram of the structure of an electronic device that implements the training method of the business processing evaluation model in Embodiment 4 of this application. Detailed Implementation
[0012] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.
[0013] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0014] Furthermore, it should be noted that the historical business processing data and other related data involved in the technical solution of this application are information and data authorized by the user or fully authorized by all parties. The collection, storage, use, processing, transmission, provision, and disclosure of this data all comply with relevant laws and regulations and do not violate public order and good morals. Corresponding operation entry points are provided for users to choose to authorize or refuse. Banks will specifically set terms regarding the use of business processing data when users conduct target banking business. In accordance with the requirements of banking regulatory authorities, these terms will be placed in a prominent position in the agreement, using plain language to explain the consequences of authorization. The user's signature is considered as completion of authorization and confirmation of the informed content. For example, in online business application scenarios such as bank clients and online banking, users will be informed through pop-ups, separate checkboxes, etc. When a user submits a business application online, the system will display a prompt page for authorization of business processing data, detailing that their business processing data and other business information will be collected for the target banking business. Only after the user actively clicks "agree," "confirm," or checks the authorization box and completes SMS verification will the bank initiate the information collection process.
[0015] Example 1 Figure 1 This is a flowchart illustrating a training method for a business processing evaluation model according to Embodiment 1 of this application. This embodiment is applicable to situations where, when a customer is processing a target banking transaction, the evaluation determines whether to recommend that the customer proceed with the target banking transaction. This evaluation can be performed by a training device for the business processing evaluation model, which can be implemented in hardware and / or software and can be configured in a computer device, such as a server. Figure 1 As shown, the method includes: S110. Obtain historical business processing data of the target bank's business, and construct business sample data based on the historical business processing data.
[0016] The business sample data may include labels from actual business transactions. The target banking business refers to the services that customers at the bank need to conduct.
[0017] In this embodiment of the invention, historical business processing data can refer to customer business data of the target banking business processed in the past, such as business data of different business dimensions such as the number of business transactions, the amount of business transactions, and business transaction information.
[0018] The "Real Business Transaction" tag can be used to characterize the historical processing status of target banking transactions.
[0019] Optionally, the sample dataset consisting of business sample data can be represented in the following data format: ; Where S is the sample dataset, Let i be the business data vector describing historical business processing data. for The corresponding real business processing label, where N is the number of samples.
[0020] It should be noted that, in this embodiment of the invention, the real business processing label can include two processing statuses: processed and not processed. For example, 1 can represent processed and 0 can represent not processed.
[0021] Optionally, after obtaining the sample dataset, the historical business processing data in the sample dataset can be standardized to eliminate the influence of different data dimensions and improve data quality.
[0022] S120. Based on business sample data, train the first model on the pre-built machine learning model until the training end condition of the first model is met, and obtain the candidate business processing evaluation model.
[0023] In this embodiment of the invention, the machine learning model can be adaptively configured according to those skilled in the art. For example, the machine learning model may include a classification prediction model with multiple decision trees. It should be noted that the termination condition for the first model training can be adaptively configured according to those skilled in the art.
[0024] Optionally, a machine learning model can be represented by the following formula: ; in, For the predicted business processing label corresponding to the i-th business sample data, Let K be the predicted score of the i-th business sample data in the k-th decision tree, where K is the number of decision trees in the machine learning model.
[0025] Optionally, before training the first model on the pre-built machine learning model based on business sample data, the method further includes: using a preset model parameter optimization algorithm to optimize the metadata model structure parameters in the machine learning model.
[0026] Specifically, by using a pre-defined model parameter optimization algorithm, the metadata and structural parameters of a machine learning model, such as the number of weak evaluators, tree depth, and model learning rate, can be optimized to obtain the optimal model parameters and improve the training quality of the model. It should be noted that the model parameter optimization algorithm can be adapted to suit the needs of those skilled in the art; for example, a sparrow search algorithm that integrates chaotic mapping and flight strategies can be used.
[0027] S130. Using a candidate business processing evaluation model, business processing prediction is performed on different combinations of business dimension features in historical business processing data to obtain candidate predicted business processing labels, and feature importance of different business dimension features in historical business processing data is determined based on the predicted business processing labels.
[0028] Among them, the business dimension feature combination can refer to the feature combination of different feature dimensions formed by the candidate business processing evaluation model after extracting features from the business sample data.
[0029] Feature importance can be used to characterize the contribution of business dimension features to the prediction results.
[0030] S140. Based on feature importance and historical business processing data, train the second model for the candidate business processing evaluation model until the training end condition of the second model is met, and obtain the target business processing evaluation model.
[0031] It should be noted that the training termination conditions for the second model can be adapted to the needs of those skilled in the art.
[0032] After obtaining the target business processing evaluation model, when a customer is preparing to conduct a target banking transaction, the model can prompt the customer to actively input relevant transaction processing data. Based on the customer's input data, the model can predict the target banking transaction that the customer will conduct, determine the customer's suitability for the target banking transaction, and provide personalized and high-quality services to the customer.
[0033] Optionally, in this embodiment of the invention, the target loss function in the first model training and the second model training can be determined based on the deviation between the real business processing label and the predicted business processing label, as well as the regularization parameter that characterizes the model structure of the machine learning model.
[0034] For example, the target loss function can be determined by the following formula: ; ; Where N is the sample size. Let K be the loss between the predicted business processing label and the actual business processing label corresponding to the i-th business sample data, and K be the number of decision trees in the machine learning model. This is the regularization term for the k-th decision tree, used to control the complexity of the model; is the penalty coefficient for the number of leaf nodes in the decision tree, T is the number of leaf nodes in the decision tree, and w is the weight (i.e., prediction score) of the leaf nodes in the decision tree. is the L2 regularization coefficient of the leaf node weights in the decision tree.
[0035] By training the machine learning model using the label loss between the predicted label and the true label, and the regularization terms of different decision trees, the model complexity can be effectively controlled to prevent overfitting and improve the model performance.
[0036] This application embodiment trains a machine learning model using a first model, and fuses the feature importance corresponding to each business dimension feature in the determined historical business processing data with their respective business dimension features. Based on the new features after fusion, a second model is trained on the machine learning model trained by the first model. This improves the analytical ability of the machine learning model after the second model training on each extracted business dimension feature, and improves the model performance and prediction accuracy.
[0037] Example 2 Figure 2 This is a flowchart of a training method for a business processing evaluation model according to Embodiment 2 of this application. Based on the technical solutions of the above embodiments, this embodiment further refines the process of "using a candidate business processing evaluation model to predict business processing for different combinations of business dimension features in historical business processing data, obtaining candidate predicted business processing labels, and determining the feature importance of different business dimension features in historical business processing data based on the predicted business processing labels." It should be noted that for parts not detailed in this embodiment, please refer to the relevant descriptions in other embodiments. Figure 2 As shown, the method includes: S210. Obtain historical business processing data of the target bank's business, and construct business sample data based on the historical business processing data.
[0038] S220. Based on business sample data, train the first model on the pre-built machine learning model until the training end condition of the first model is met, and obtain the candidate business processing evaluation model.
[0039] S230. Input historical business processing data into the candidate business processing evaluation model for feature extraction, generate a set of business dimension features, and at least one subset of business dimension features.
[0040] Among them, the combinations of business dimension features exist in different subsets of business dimension features; the set of business dimension features can refer to the combination of features that includes all business dimension features.
[0041] It should be noted that a subset of business dimension features can be a subset of the set of business dimension features, and different subsets of business dimension features have the same number of feature types; the difference in the number of feature types between a subset of business dimension features and the set of business dimension features is 1. For example, if the set of business dimension features includes 8 business dimension features, then the subset of business dimension features contains 7 business dimension features, and different subsets of business dimension features share 6 identical business dimension features.
[0042] S240. Perform business processing prediction on the business dimension feature set and the business dimension feature subset respectively to obtain the corresponding candidate predicted business processing labels.
[0043] S250. Determine the feature importance of each business dimension feature based on the business dimension feature set, the business dimension feature subset, and the candidate predicted business processing labels corresponding to each of the business dimension feature set and the business dimension feature subset.
[0044] Optionally, feature importance can be determined using the following formula: ; Where F is the set of business dimension features, S is a subset of business dimension features, and x refers to a feature dimension in S that does not exist in F. This refers to the predicted business processing label output by the candidate business processing evaluation model based on the set of business dimension features. It refers to the predicted business processing label output by the candidate business processing evaluation model based on a subset of business dimension features.
[0045] S260. Based on feature importance and historical business processing data, train the second model for the candidate business processing evaluation model until the training end condition of the second model is met, and obtain the target business processing evaluation model.
[0046] Optionally, in this embodiment of the invention, a second model training is performed on the candidate business processing evaluation model based on feature importance and historical business processing data, including: inputting the historical business processing data into the candidate business processing evaluation model; extracting features from the historical business processing data to generate a business dimension feature set; fusing each business dimension feature in the business dimension feature set with its corresponding feature importance to generate a target fusion feature set; using the target fusion feature set to predict business processing to obtain a target predicted business processing label; and training the candidate business processing evaluation model based on the target predicted business processing label and the actual business processing label.
[0047] It should be noted that the candidate business processing evaluation model can extract features from the business data of each business dimension in the historical business processing data to generate business dimension feature data corresponding to each business dimension; the business dimension feature data corresponding to each business dimension is then fused with the feature importance of its respective business dimension, and a target fused feature set is generated based on the fused new business dimension feature data.
[0048] By fusing the feature importance of the business dimension with the features of the business dimension to generate new features, a secondary model training is performed on the candidate business processing evaluation model, which improves the model performance and prediction accuracy of the trained model.
[0049] This application embodiment improves the model's ability to analyze the importance of local features by performing importance analysis on the features of each business dimension.
[0050] Example 3 Figure 3 This is a schematic diagram of a training device for a business processing evaluation model according to Embodiment 3 of this application. It is applicable to situations where, when a customer is processing a target banking transaction, the device evaluates whether to recommend that the customer proceed with that transaction. The training device for this business processing evaluation model can be implemented in hardware and / or software and can be configured in a computer device, such as a server. Figure 3 As shown, the device includes: The sample generation module 310 is used to acquire historical business processing data of the target banking business and construct business sample data based on the historical business processing data; wherein, the business sample data includes real business processing tags; The first training module 320 is used to train a pre-built machine learning model based on the business sample data until the first model training termination condition is met, thereby obtaining a candidate business processing evaluation model. The feature importance module 330 is used to use the candidate business processing evaluation model to predict the business processing of different business dimension feature combinations in the historical business processing data, obtain candidate predicted business processing labels, and determine the feature importance of different business dimension features in the historical business processing data based on the predicted business processing labels. The second training module 340 is used to train the candidate business processing evaluation model based on the feature importance and the historical business processing data, until the training end condition of the second model is met, and thus obtain the target business processing evaluation model.
[0051] This application embodiment trains a machine learning model using a first model, and fuses the feature importance corresponding to each business dimension feature in the determined historical business processing data with their respective business dimension features. Based on the new features after fusion, a second model is trained on the machine learning model trained by the first model. This improves the analytical ability of the machine learning model after the second model training on each extracted business dimension feature, and improves the model performance and prediction accuracy.
[0052] Optionally, the feature importance module 330 includes: The first feature extraction unit is used to input the historical business processing data into the candidate business processing evaluation model for feature extraction, generate a business dimension feature set, and at least one business dimension feature subset; wherein, different business dimension feature subsets contain different combinations of business dimension features; the business dimension feature set refers to a feature combination that includes all business dimension features; The candidate prediction unit is used to predict business processing for the business dimension feature set and the business dimension feature subset respectively, and obtain the corresponding candidate predicted business processing labels. The feature importance unit is used to determine the feature importance of each business dimension feature based on the business dimension feature set, the business dimension feature subset, and the candidate predicted business processing labels corresponding to the business dimension feature set and the business dimension feature subset.
[0053] Optionally, the business dimension feature subset is a subset of the business dimension feature set, and different business dimension feature subsets have the same number of feature types; the difference in the number of feature types between the business dimension feature subset and the business dimension feature set is 1.
[0054] Optionally, the second training module 340 includes: The second feature extraction unit is used to input the historical business processing data into the candidate business processing evaluation model, extract features from the historical business processing data, and generate a business dimension feature set. The feature fusion unit is used to fuse each business dimension feature in the business dimension feature set with its corresponding feature importance to generate a target fused feature set. The second training unit is used to perform business processing prediction using the target fusion feature set, obtain target predicted business processing labels, and train the second model of the candidate business processing evaluation model based on the target predicted business processing labels and the real business processing labels.
[0055] Optionally, the device may also include: The parameter optimization module is used to optimize the metadata model structure parameters in the machine learning model using a preset model parameter optimization algorithm before training the first model on the pre-built machine learning model based on the business sample data.
[0056] Optionally, the target loss function in the training of the first model and the second model is determined based on the deviation between the real business processing label and the predicted business processing label, as well as the regularization parameter that characterizes the model structure of the machine learning model.
[0057] The training device for the business processing evaluation model provided in this application embodiment can execute the training method of the business processing evaluation model provided in any embodiment of this application, and has the corresponding functional modules and beneficial effects for executing the training method of each business processing evaluation model.
[0058] According to embodiments of this application, this application also provides an electronic device, a readable storage medium, and a computer program product.
[0059] Example 4 Figure 4 This is a schematic diagram of the structure of an electronic device 410 that implements the training method for the business processing evaluation model of the embodiments of this application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or claimed herein.
[0060] like Figure 4 As shown, the electronic device 410 includes at least one processor 411 and a memory, such as a read-only memory 412 or a random access memory 413, communicatively connected to the at least one processor 411. The memory stores computer programs executable by the at least one processor. The processor 411 can perform various appropriate actions and processes based on the computer program stored in the read-only memory 412 or loaded from storage unit 418 into the random access memory 413. The random access memory 413 can also store various programs and data required for the operation of the electronic device 410. The processor 411, read-only memory 412, and random access memory 413 are interconnected via a bus 414. An input / output interface 415 is also connected to the bus 414.
[0061] Multiple components in electronic device 410 are connected to input / output interface 415, including: input unit 416, such as keyboard, mouse, etc.; output unit 417, such as various types of monitors, speakers, etc.; storage unit 418, such as disk, optical disk, etc.; and communication unit 419, such as network card, modem, wireless transceiver, etc. Communication unit 419 allows electronic device 410 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0062] Processor 411 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 411 include, but are not limited to, central processing units, graphics processing units, various special-purpose artificial intelligence computing chips, various processors running machine learning model algorithms, digital signal processors, and any suitable processor, controller, microcontroller, etc. Processor 411 performs the various methods and processes described above, such as the training methods for business processing evaluation models.
[0063] In some embodiments, the training method for the business processing evaluation model can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 418. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 410 via read-only memory 412 and / or communication unit 419. When the computer program is loaded into random access memory 413 and executed by processor 411, one or more steps of the training method for the business processing evaluation model described above can be performed. Alternatively, in other embodiments, processor 411 can be configured as the training method for the business processing evaluation model by any other suitable means (e.g., by means of firmware).
[0064] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays, application-specific integrated circuits (ASICs), application-specific standard products (ASICs), systems-on-a-chip (SoCs), payload programmable logic devices, computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.
[0065] Computer programs used to implement the methods of this application may be written in any combination of one or more programming languages. These computer programs may be provided to the processor of a general-purpose computer, special-purpose computer, or other training device for a programmable business processing evaluation model, such that when executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.
[0066] In the context of this application, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fibers, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0067] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a cathode ray tube or liquid crystal display monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).
[0068] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.
[0069] A computing system can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product within the cloud computing service system to address the shortcomings of traditional physical hosts and virtual private servers, such as high management difficulty and weak business scalability.
[0070] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this application can be achieved, and this is not limited herein.
[0071] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.
Claims
1. A training method for a business processing evaluation model, characterized in that, include: Obtain historical transaction data for the target banking business, and construct business sample data based on the historical transaction data; wherein, the business sample data includes real transaction tags; Based on the business sample data, the pre-built machine learning model is trained with the first model until the training end condition of the first model is met, and a candidate business processing evaluation model is obtained. Using the candidate business processing evaluation model, business processing prediction is performed on different combinations of business dimension features in the historical business processing data to obtain candidate predicted business processing labels, and the feature importance of different business dimension features in the historical business processing data is determined based on the predicted business processing labels. Based on the feature importance and the historical business processing data, the candidate business processing evaluation model is trained as a second model until the training end condition of the second model is met, thus obtaining the target business processing evaluation model.
2. The method according to claim 1, characterized in that, The step of using the candidate business processing evaluation model to predict business processing for different combinations of business dimension features in the historical business processing data, obtaining candidate predicted business processing labels, and determining the feature importance of different business dimension features in the historical business processing data based on the predicted business processing labels includes: The historical business processing data is input into the candidate business processing evaluation model for feature extraction, generating a business dimension feature set and at least one business dimension feature subset; wherein, different business dimension feature subsets contain different combinations of business dimension features; the business dimension feature set refers to a feature combination that includes all business dimension features; Business processing predictions are performed on the business dimension feature set and the business dimension feature subset respectively to obtain the corresponding candidate predicted business processing labels. The feature importance of each business dimension feature is determined based on the business dimension feature set, the business dimension feature subset, and the candidate predicted business processing labels corresponding to the business dimension feature set and the business dimension feature subset.
3. The method according to claim 1, characterized in that, The business dimension feature subset is a subset of the business dimension feature set, and different business dimension feature subsets have the same number of feature types; the difference in the number of feature types between the business dimension feature subset and the business dimension feature set is 1.
4. The method according to claim 1, characterized in that, The step of training a second model for the candidate business processing evaluation model based on the feature importance and the historical business processing data includes: The historical business processing data is input into the candidate business processing evaluation model, and features are extracted from the historical business processing data to generate a set of business dimension features. Each business dimension feature in the business dimension feature set is fused with its corresponding feature importance to generate a target fused feature set. The target fusion feature set is used to predict business processing, and the target predicted business processing label is obtained. Based on the target predicted business processing label and the real business processing label, the candidate business processing evaluation model is trained as a second model.
5. The method according to claim 1, characterized in that, Before training the first model on the pre-built machine learning model based on the aforementioned business sample data, the process also includes: A preset model parameter optimization algorithm is used to optimize the metadata model structure parameters in the machine learning model.
6. The method according to claim 1, characterized in that, The target loss function in the training of the first model and the second model is determined based on the deviation between the real business processing label and the predicted business processing label, as well as the regularization parameter that characterizes the model structure of the machine learning model.
7. A training device for a business processing evaluation model, characterized in that, include: The sample generation module is used to acquire historical business processing data of the target banking business and construct business sample data based on the historical business processing data; wherein, the business sample data includes real business processing tags; The first training module is used to train a pre-built machine learning model based on the business sample data until the first model training termination condition is met, thereby obtaining a candidate business processing evaluation model. The feature importance module is used to use the candidate business processing evaluation model to predict the business processing of different business dimension feature combinations in the historical business processing data, obtain candidate predicted business processing labels, and determine the feature importance of different business dimension features in the historical business processing data based on the predicted business processing labels. The second training module is used to train the candidate business processing evaluation model based on the feature importance and the historical business processing data, until the training end condition of the second model is met, and the target business processing evaluation model is obtained.
8. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the training method for the business processing evaluation model as described in any one of claims 1-6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the training method of the business processing evaluation model as described in any one of claims 1-6.
10. A computer program product comprising a computer program that, when executed by a processor, implements a training method for a business processing evaluation model according to any one of claims 1-6.