A Loan Evaluation Method and System Based on Multi-Objective Optimization

Abstract

The present invention relates to the technical field of data processing, and specifically to a loan evaluation method and system based on multi-objective optimization. The method includes obtaining the loan type of a customer and the initial customer information; when the initial customer information contains abnormal information, adjusting the initial customer information according to a preset abnormal analysis model to obtain the adjusted customer information; selecting a corresponding quota evaluation model according to the loan type selected by the customer to analyze and process the customer information, that is, determining the qualification rating, risk rating and quota rating of the customer according to the customer information; and on the basis of the determined customer qualification rating, risk rating and quota rating, determining the credit quota of the customer through a preset risk quota balancing rule. By the above method, the present invention can make the loan evaluation quota of the customer reasonable, and make the risk tolerance of the enterprise relatively balanced with the loan quota of the customer, so as to enable the long-term and sustainable development of the loan business of the enterprise.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a loan evaluation method and system based on multi-objective optimization. Background Art

[0002] In the risk control scenario of credit loans, differentiated credit limit management has become the main technical means for financial institutions to reduce loan risks while complying with regulatory requirements. Scientific and rational customer credit limit management is crucial to improving customer satisfaction and reducing loan risks.

[0003] However, there are two existing loan quota assessment strategies. The first strategy is to provide customers with excessively high loan quotas, which will increase the risk that the company cannot recover the loan. The second strategy is to reduce the risk of the company recovering the loan by giving customers a smaller loan quota, which will reduce the customer's trust in the company and lead to customer loss. In the long run, these two unreasonable loan quota assessment strategies will cause losses to the company and are not conducive to the long-term development of the company. In addition, companies often use a one-size-fits-all approach to assess customers' qualifications, credit, etc. based on their abnormal loan information, without analyzing the customer's actual situation, resulting in the customer's qualifications and credit being lower than the actual situation, which in turn reduces customer satisfaction.

[0004] Therefore, the present invention provides a loan evaluation method and system based on multi-objective optimization to solve the above problems. Summary of the invention

[0005] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a loan evaluation method and system based on multi-objective optimization to solve the problem that the above-mentioned traditional loan amount evaluation method evaluates the user's loan amount unreasonably, which is not conducive to the long-term development of the enterprise.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] In a first aspect, the present invention provides a loan evaluation method based on multi-objective optimization, comprising:

[0008] Obtain the customer's loan type and initial customer information, and store the loan type and initial customer information in the qualification database, loan behavior database, risk feature database, and supplementary database according to the preset classification standards;

[0009] When the initial customer information contains abnormal information, the initial customer information is adjusted according to the preset abnormal analysis model to determine whether to remove the abnormal information to obtain the adjusted customer information;

[0010] Select the corresponding quota evaluation model according to the loan type selected by the customer to analyze and process the customer information to determine the credit quota of the customer; specifically including: determining the qualification rating, risk rating and quota rating of the customer according to the customer information; determining multiple evaluation parameters in the preset risk quota equilibrium rule according to the qualification rating, risk rating and quota rating of the customer; determining the credit quota of the customer according to the preset risk quota equilibrium rule.

[0011] Preferably, when the initial customer information contains abnormal information, adjust the initial customer information according to the preset abnormal analysis model, and judge whether to eliminate the abnormal information to obtain the adjusted customer information; determining the qualification rating, risk rating and quota rating of the customer according to the customer information includes:

[0012] Obtain the abnormal information in the customer information and the supplementary information related to the abnormal information, analyze the abnormal information and supplementary information of the customer in chronological order to determine the cause of the abnormality; judge the abnormal type of the abnormal information according to the cause of the abnormality; when the abnormal type belongs to the preset abnormal type, clear the abnormal information.

[0013] Preferably, the analyzing the abnormal information and supplementary information of the customer in chronological order to determine the cause of the abnormality includes: extracting the characteristics of the abnormal information and supplementary information and sorting them in chronological order to obtain the abnormal sequence characteristics; using the association analysis method to determine the association relationship between the characteristics in the abnormal sequence characteristics; forming the cause of the abnormality according to the abnormal information, supplementary information and association relationship.

[0014] Preferably, determining the qualification rating, risk rating and quota rating of the customer according to the customer information includes: automatically discretizing the customer information according to the risk gradient and ROA gradient to obtain feature vectors, and fusing the feature vectors of the same type to obtain the average feature vector of the same type of data;

[0015] Using the graph neural network algorithm to conduct association analysis on the types of each average vector to respectively determine the association relationship between the types of data in the qualification rating, risk rating and quota rating, and form three mapping relationship graphs; determining the weights of the types of data in the graph according to the mapping relationship graph; determining the qualification rating, risk rating and quota rating according to the weights of the types of data and each average feature vector respectively.

[0016] Preferably, determining the weights of the types of data according to the mapping relationship graph includes: using the number of lines connected to the nodes in the mapping relationship graph as the initial weight of the type corresponding to the node; normalizing each initial weight to obtain the corresponding weights of the types of data.

[0017] Preferably, determining the qualification rating, risk rating, and quota rating according to the weights of various types of data and each average feature vector respectively includes: processing each average feature vector and the weights of various types of data to construct a weight set and a feature vector matrix; using the fuzzy analysis method to perform fuzzy processing on the weight set and the feature vector matrix to obtain the qualification rating, risk rating, and quota rating of the customer.

[0018] Preferably, determining the credit limit of the customer according to the preset risk limit balance rule includes:

[0019] Obtaining the corresponding average feature vector in the preset loss function and filling it; calculating the credit limit of the customer according to the preset loss function, and the formula of the preset loss function is:

[0020] ,

[0021] where, is the optimization function, is the return function, is the quota scale function, is the quota function, is the i-th average feature vector in the total average feature vector, is the weight coefficient, is the risk function.

[0022] Preferably, a loan evaluation method based on multi-objective optimization further includes: regularly obtaining the change information of the customer to update the corresponding data in each database, and dynamically adjusting the corresponding average feature vector and the mapping relationship graph according to the updated data; when the change information belongs to the preset change condition, determining the secondary credit limit of the customer according to the updated average feature vector and the mapping relationship graph; and taking the secondary credit limit as the current credit limit of the user.

[0023] Preferably, a loan evaluation method based on multi-objective optimization further includes: encrypting the preset risk limit balance rule by using the key encryption method; decrypting with the corresponding key when accessing the preset risk limit balance rule; and randomly generating a new key after the key is used.

[0024] In a second aspect, the present invention provides a loan evaluation system based on multi-objective optimization. The loan evaluation system is used to execute the loan evaluation method described in any one of the above technical solutions, and includes:

[0025] A data acquisition module, which acquires the loan type and initial customer information of the customer, and stores the loan type and initial customer information into the qualification database, loan behavior database, risk characteristic database, and supplementary database according to the preset classification standard;

[0026] Anomaly analysis module: When the initial customer information contains anomaly information, it adjusts the initial customer information according to a preset anomaly analysis model, and determines whether to eliminate the anomaly information to obtain the adjusted customer information.

[0027] Quota evaluation module: According to the loan type selected by the customer, it selects the corresponding quota evaluation model to analyze and process the customer information to determine the customer's credit quota. Specifically, it includes: determining the customer's qualification rating, risk rating, and quota rating according to the customer information; determining multiple evaluation parameters in the preset risk quota balance rule according to the customer's qualification rating, risk rating, and quota rating; and determining the customer's credit quota according to the preset risk quota balance rule.

[0028] The beneficial effects of the present invention are as follows:

[0029] 1. The present invention determines the customer's qualification rating, risk rating, and quota rating according to the customer information; and on the basis of the determined customer qualification rating, risk rating, and quota rating, determines the customer's credit quota through a preset risk quota balance rule. In this way, the present invention can make the loan evaluation quota of the customer reasonable, make the risk tolerance of the enterprise and the loan quota of the customer relatively balanced, so as to enable the long-term and sustainable development of the enterprise's loan business.

[0030] 2. When there is anomaly information in the initial customer information, the present invention uses a preset anomaly analysis model to judge the anomaly information to determine whether to eliminate the anomaly information; after eliminating the anomaly information, it uses the remaining initial customer information to analyze the customer's qualification rating, risk rating, and quota rating, and determine the customer's credit quota. In this way, the present invention can solve the problem that the enterprise evaluates the customer's qualification, credit, etc. based on the customer's abnormal loan information without analyzing in combination with the actual situation of the customer, resulting in the customer's qualification, credit, etc. being lower than the actual situation, which helps to improve the customer's credit quota and thus increase the customer's satisfaction.

[0031] 3. The present invention uses the graph neural network algorithm to perform correlation analysis on the types of each average vector to determine the correlation relationship between various types of data in the qualification rating, the correlation relationship between various types of data in the risk rating, and the correlation relationship between various types of data in the quota rating; and then processes the correlation relationship and each average feature vector to determine the customer's qualification rating, risk rating, and quota rating. In this way, the present invention can objectively evaluate the customer's qualification rating, risk rating, and quota rating based on the correlation relationship between various types of data of the customer, making the evaluation result more in line with the actual situation of the customer. Description of the Drawings

[0032] Figure 1Schematic flowchart of a loan assessment method based on multi-objective optimization according to the present invention;

[0033] Figure 2 Schematic flowchart of determining the credit limit of a customer according to the present invention;

[0034] Figure 3 Module diagram of a loan assessment system based on multi-objective optimization according to the present invention. Detailed implementation manners

[0035] Next, reference will be made to Figure 1 to Figure 3 to explain each embodiment of the present invention in detail. Those skilled in the art should understand that these implementation manners are only used to explain the technical principle of the present invention and are not intended to limit the protection scope of the present invention.

[0036] A loan assessment method based on multi-objective optimization, as shown in Figure 1 and Figure 2 shown, includes the following steps:

[0037] Step S11: Obtain the loan type and initial customer information of the customer, and store the loan type and initial customer information into the qualification database, loan behavior database, risk characteristic database, and supplementary database according to a preset classification standard;

[0038] Among them, the initial customer information includes customer personal information, customer transaction information, historical loan information, work information, business information, and credit information; the data in the loan behavior database is used to evaluate the quota rating of the customer to limit the loan quota range of the customer;

[0039] Step S12: When the initial customer information contains abnormal information, adjust the initial customer information according to a preset abnormal analysis model, and determine whether to eliminate the abnormal information to obtain the adjusted customer information;

[0040] After executing step S12, that is, when the abnormal information is eliminated, the remaining initial customer information is used to analyze the customer's qualification rating, risk rating, and quota rating, and to determine the customer's credit limit. Through the above method, the present invention can solve the problem that enterprises usually use a one-size-fits-all method to evaluate the qualifications, credit, etc. of customers for abnormal loan information of customers, without analyzing in combination with the actual situation of the customers, resulting in the customer's qualifications, credit, etc. being lower than the actual situation, which helps to improve the customer's credit limit and thus increase the customer's satisfaction;

[0041] Step S13: Select the corresponding quota evaluation model according to the loan type selected by the customer to analyze and process the customer information to determine the customer's credit limit.

[0042] Preferably, the quota evaluation model includes an individual quota evaluation model, a sole proprietor quota evaluation model, and an enterprise quota evaluation model. For example, when targeting office workers, factors such as the stability of the customer's job, salary range, and the prospects of the industry in which they work are considered to evaluate the customer's qualification rating, risk rating, credit limit, etc.

[0043] For sole proprietors, factors such as the industry in which the customer is self-employed, the business conditions of the same or similar practitioners in the location, etc. are considered to evaluate the customer's qualification rating, risk rating, credit limit, etc. For enterprises, factors such as the industry of the enterprise, the enterprise's transaction records, the social insurance situation of the staff, the tax payment situation, the beneficiary, etc. are considered to objectively evaluate the enterprise's qualification rating, risk rating, credit limit, etc. In short, for different user groups, different evaluation strategies are adopted to evaluate the customer's credit limit.

[0044] Among them, the specific analysis process of step S13 includes the following steps:

[0045] Step S21: Determine the customer's qualification rating, risk rating, and quota rating based on the customer information;

[0046] Step S22: Determine multiple evaluation parameters in the preset risk quota equilibrium rule based on the customer's qualification rating, risk rating, and quota rating;

[0047] Among them, the determined evaluation parameters are the corresponding parameters in the return function, quota scale function, quota function, and risk function;

[0048] Step S23: Determine the customer's credit limit according to the preset risk quota equilibrium rule.

[0049] In steps S21 to S23, the present invention determines the customer's qualification rating, risk rating, and quota rating based on the customer information; and on the basis of the determined customer qualification rating, risk rating, and quota rating, determines the customer's credit limit through the preset risk quota equilibrium rule. By the above method, the present invention can make the loan evaluation quota of the customer reasonable, make the risk tolerance of the enterprise and the loan quota of the customer relatively balanced, so as to enable the long-term and sustainable development of the enterprise's loan business.

[0050] In an embodiment of the present invention, when the initial customer information contains abnormal information, the initial customer information is adjusted according to the preset abnormal analysis model, and it is judged whether to eliminate the abnormal information to obtain the adjusted customer information; determining the customer's qualification rating, risk rating, and quota rating based on the customer information includes: obtaining the abnormal information in the customer information and the supplementary information related to the abnormal information, analyzing the customer's abnormal information and supplementary information in chronological order to determine the cause of the abnormality; judging the abnormal type of the abnormal information according to the cause of the abnormality; when the abnormal type belongs to the preset abnormal type, the abnormal information is cleared.

[0051] Preferably, the abnormality type includes industry abnormality, turnover abnormality, operation abnormality, etc. The preset abnormality type includes turnover abnormality.

[0052] In this embodiment, the abnormal information and supplementary information related to the abnormal information in the customer information are obtained by extracting data of abnormal customer behavior from the initial customer information, such as overdue records, and then filtering out relevant supplementary information from the supplementary information based on the abnormal information. The supplementary information can be abnormal turnover transfer records, proof contracts for abnormal turnover items, explanatory text, etc.

[0053] Then, the abnormal information and supplementary information of the customer are analyzed in chronological order to determine the cause of the abnormality; according to the cause of the abnormality, whether the abnormal type of the abnormal information belongs to the preset abnormal type is determined; when the abnormal type belongs to the preset abnormal type, the abnormal information is cleared. When the abnormal type does not belong to the preset abnormal type, the abnormal information is not eliminated, and the customer's credit limit is evaluated by executing the subsequent loan evaluation method.

[0054] Through the above implementation mode, after eliminating the relevant abnormal information, the present invention uses the remaining initial customer information to analyze the customer's qualification rating, risk rating and credit rating, and determines the customer's credit limit. This can solve the problem that enterprises evaluate customers' qualifications, credit, etc. based on their abnormal loan information without analyzing the customers' actual conditions, resulting in the customer's qualifications, credit, etc. being lower than the actual conditions. This helps to improve the customer's credit limit, thereby increasing customer satisfaction.

[0055] In one embodiment of the present invention, the analysis of the abnormal information and supplementary information of the customer in chronological order to determine the abnormal cause includes: extracting features from the abnormal information and supplementary information and sorting them in chronological order to obtain abnormal sequence features; using an association analysis method to determine the association relationship between each feature in the abnormal sequence features; and forming the abnormal cause based on the abnormal information, supplementary information and the association relationship. The association relationship includes a causal relationship, a progressive relationship, etc.

[0056] Preferably, the feature extraction method may be to extract features based on a keyword library, or to use AIGC to process the abnormal information and supplementary information to automatically generate the abnormal cause.

[0057] Through the above-mentioned implementation mode, the present invention can determine the abnormal cause of the abnormal information based on the abnormal information and supplementary information, which is helpful to subsequently determine the abnormal type based on the abnormal cause, so as to further judge whether to eliminate the abnormal information, increase the customer's qualifications, credit, etc., and then help increase the customer's credit limit.

[0058] In one embodiment of the present invention, determining the qualification rating, risk rating, and quota rating of a customer based on customer information includes: automatically discretizing the customer information according to the risk gradient and ROA gradient to obtain feature vectors, and fusing the feature vectors of the same type to obtain the average feature vector of the same type of data; using a graph neural network algorithm to perform correlation analysis on the types of each average vector to respectively determine the correlation relationships between the types of data for the qualification rating, risk rating, and quota rating, and forming three mapping relationship graphs; determining the weights of the types of data in the mapping relationship graphs according to the mapping relationship graphs; and respectively determining the qualification rating, risk rating, and quota rating according to the weights of the types of data and each average feature vector.

[0059] When using the graph neural network algorithm to perform correlation analysis on the types of each average vector, a tree structure-based algorithm or an icon analysis method is combined to display the correlation relationships between the types of data.

[0060] In this embodiment, in the databases corresponding to the qualification rating, risk rating, and quota rating, the same data can be stored in different databases to facilitate data acquisition, quickly perform analysis and processing, and determine the relevant data of the evaluation indicators; the evaluation indicators include qualification, risk, and quota. In the score setting, corresponding scores are set for different types of data. In the liability item, negative scores are set for different liability amounts; in the item of repaying on time or repaying in advance, positive scores are set, and when the information is updated, the scores will change accordingly.

[0061] Through the setting method of this embodiment, the present invention can determine the qualification rating, risk rating, and quota rating of a customer based on customer information, and then facilitate determining multiple evaluation parameters in the preset risk quota balance rule according to the qualification rating, risk rating, and quota rating of the customer; and determining the credit quota of the customer according to the preset risk quota balance rule, making the loan evaluation quota of the customer reasonable, and further making the risk tolerance of the enterprise and the loan quota of the customer relatively balanced, so as to enable the long-term and sustainable development of the loan business of the enterprise.

[0062] In one embodiment of the present invention, determining the weights of the types of data according to the mapping relationship graph includes: using the number of lines connected to the nodes in the mapping relationship graph as the initial weight of the type corresponding to the node; and performing normalization processing on each initial weight to obtain the corresponding weights of the types of data.

[0063] Furthermore, the present invention uses the number of relationship lines between the nodes of the graph neural mapping relationship map as the initial weight. For different types of data, a non-linear normalization method, a linear normalization method, or a standard deviation normalization method is used to normalize each initial value to obtain the corresponding weights for each type of data. Through the setting method of this embodiment, the present invention can objectively determine the weights of each type of data, so that the calculated qualification rating, risk rating, and quota rating are objective, avoiding the interference of subjective factors.

[0064] In an embodiment of the present invention, determining the qualification rating, risk rating, and quota rating according to the weights of each type of data and each average feature vector respectively includes: processing each average feature vector and the weights of each type of data to construct a weight set and a feature vector matrix; using the fuzzy analysis method to perform fuzzy processing on the weight set and the feature vector matrix to obtain the qualification rating, risk rating, and quota rating of the customer.

[0065] In this embodiment, when using the fuzzy analysis method to perform fuzzy processing on the feature vector matrix, a consistency test is performed on the feature vector matrix. When the test fails, the feature vector matrix is adjusted; when the test passes, the weight set and the feature vector matrix are processed to obtain the qualification rating, risk rating, and quota rating of the customer. And the qualification assessment result, risk assessment result, and quota assessment result of the customer can also be determined.

[0066] By using the fuzzy analysis method, the present invention can objectively process and analyze the weights and each average feature vector, determine the qualification rating, risk rating, and quota rating of the customer, avoid the interference of human subjective factors, make the determined results objective, and the fuzzy analysis method can handle the uncertainty between data and improve the accuracy of the assessment.

[0067] In an embodiment of the present invention, determining the credit limit of the customer according to the preset risk limit balance rule includes: obtaining the corresponding average feature vector in the preset loss function and filling it; calculating the credit limit of the customer according to the preset loss function, and the formula of the preset loss function is:

[0068] ,

[0069] where is the optimization function, is the return function, is the quota scale function, is the quota function, is the i-th average feature vector in the total average feature vector, is the weight coefficient, It is a risk function. The total number of feature vectors is n; the credit limit value corresponding to the maximum value of the preset loss function is the finally determined credit limit.

[0070] Through the calculation formula of the preset loss function in this embodiment, the present invention can reasonably use the loan evaluation amount of customers, so that the risk tolerance of the enterprise and the loan amount of customers can be relatively balanced, in order to enable the long-term and sustainable development of the enterprise's loan business.

[0071] In an embodiment of the present invention, a loan evaluation method based on multi-objective optimization further includes: regularly obtaining the change information of customers to update the corresponding data in each database, and dynamically adjusting the corresponding average feature vector and mapping relationship graph according to the updated data; when the change information belongs to the preset change condition, determining the secondary credit limit of the customer according to the updated average feature vector and mapping relationship graph; and using the secondary credit limit as the current credit limit of the user.

[0072] Preferably, the method for obtaining change information can be to obtain it from other platforms by using web crawler methods, information re-uploaded by customers, etc. Among them, the change information includes the real-time collection of relevant information such as the transaction information, loan information, work information, business information, and personal credit of customers, so as to evaluate the qualifications, creditworthiness, and loan repayment ability of customers in real time.

[0073] Preferably, the preset change conditions include the work status of customers, the financing situation of enterprises, business situations, etc., and there are specific conditions targeted at specific situations. For example, when the customer is a staff member and the company where they work changes, compare the salary of the current position in the company with the customer's original salary. When it exceeds the set ratio or drops by the set ratio, the change information belongs to the preset change condition.

[0074] Through the setting method of this embodiment, the present invention can regularly collect the latest information of customers, and then flexibly adjust the credit limit of customers according to the latest information, so as to ensure the reasonableness of the loan evaluation amount within the set period, make the risk tolerance of the enterprise and the loan amount of customers relatively balanced, in order to enable the long-term and sustainable development of the enterprise's loan business.

[0075] In an embodiment of the present invention, a loan evaluation method based on multi-objective optimization further includes: encrypting the preset risk limit balancing rule by using a key encryption method; decrypting it by using the corresponding key when accessing the preset risk limit balancing rule; and randomly generating a new password after the key is used. At the same time, encrypt the customer information by using the key encryption method.

[0076] In this embodiment, the present invention adopts a key encryption method to encrypt the preset risk quota balancing rule to ensure the security of customer information during transmission, processing, and retention, and at the same time ensure the security of the internal logic of the preset risk quota balancing rule.

[0077] The present invention also provides a loan evaluation system based on multi-objective optimization. As shown in the appendix Figure 3 The loan evaluation system is used to execute the loan evaluation method described in any one of the above embodiments, and includes a data acquisition module, an anomaly analysis module, and a quota evaluation module. The main contents of each module are as follows:

[0078] The data acquisition module acquires the loan type of the customer and the initial customer information, and stores the loan type and the initial customer information in the qualification database, the loan behavior database, the risk characteristic database, and the supplementary database according to the preset classification criteria;

[0079] The anomaly analysis module, when the initial customer information contains anomaly information, adjusts the initial customer information according to the preset anomaly analysis model, and judges whether to eliminate the anomaly information to obtain the adjusted customer information;

[0080] The quota evaluation module selects the corresponding quota evaluation model according to the loan type selected by the customer to analyze and process the customer information to determine the customer's credit quota; specifically includes: determining the customer's qualification rating, risk rating, and quota rating according to the customer information; determining multiple evaluation parameters in the preset risk quota balancing rule according to the customer's qualification rating, risk rating, and quota rating; determining the customer's credit quota according to the preset risk quota balancing rule.

[0081] Through the mutual cooperation between the modules of the present invention, the customer's qualification rating, risk rating, and quota rating can be determined according to the customer information; and on the basis of the determined customer qualification rating, risk rating, and quota rating, the customer's credit quota is determined through the preset risk quota balancing rule. In the above manner, the present invention can make the loan evaluation quota of the customer reasonable, make the risk tolerance of the enterprise and the loan quota of the customer relatively balanced, so as to enable the long-term and sustainable development of the enterprise's loan business.

[0082] Moreover, when there is anomaly information in the initial customer information, the present invention adopts a method of judging the anomaly information through a preset anomaly analysis model to judge whether to eliminate the anomaly information; after eliminating the anomaly information, the remaining initial customer information is used to analyze the customer's qualification rating, risk rating, and quota rating, and determine the customer's credit quota. In the above manner, the present invention can solve the problem that the enterprise evaluates the customer's qualification, credit, etc. based on the customer's abnormal loan information without analyzing in combination with the actual situation of the customer, resulting in the customer's qualification, credit, etc. being lower than the actual situation, which helps to improve the customer's credit quota and thus increase the customer's satisfaction.

[0083] The various embodiments of the systems and techniques described above in this specification can be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems-on-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that receives data and instructions from, and transmits data and instructions to, a storage system, at least one input device, and at least one output device.

[0084] It should be noted that in the description of the present invention, the terms "first", "second", and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0085] The program code for implementing the methods of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the program codes are executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program codes can be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0086] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, 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 disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0087] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the 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 acoustic input, speech input, or tactile input).

[0088] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), and the Internet.

[0089] A computer system can include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. 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, a server of a distributed system, or a server incorporating a blockchain.

[0090] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A loan evaluation method based on multi-objective optimization, characterized in that: include: Obtain the customer's loan type and initial customer information, and store the loan type and initial customer information in the qualification database, loan behavior database, risk feature database, and supplementary database according to the preset classification standards; When the initial customer information contains abnormal information, the initial customer information is adjusted according to a preset abnormal analysis model to determine whether to remove the abnormal information to obtain adjusted customer information, including: extracting features from the abnormal information and supplementary information and sorting them in chronological order to obtain abnormal sequence features; using an association analysis method to determine the association relationship between the features in the abnormal sequence features; forming an abnormal cause based on the abnormal information, supplementary information and the association relationship; determining the abnormal type of the abnormal information based on the abnormal cause; and when the abnormal type belongs to a preset abnormal type, clearing the abnormal information; According to the loan type selected by the customer, the corresponding credit assessment model is selected to analyze and process the customer information to determine the customer's credit limit, including: automatically discretizing the customer information according to the risk gradient and ROA gradient to obtain the feature vector, fusing the feature vectors of the same type to obtain the average feature vector of the same type of data; using the graph neural network algorithm to perform association analysis on the types of each average vector to respectively determine the association relationship between the various types of data in the qualification rating, risk rating and credit rating, and form three mapping relationship maps; according to the number of lines connecting the nodes in the mapping relationship map, the initial weight of the corresponding type of the node is used; each initial weight is normalized to obtain the corresponding weight of each type of data; each average feature vector and the weight of each type of data are processed to construct a weight set and a feature vector matrix; the fuzzy analysis method is used to fuzzy the weight set and the feature vector matrix to obtain the customer's qualification rating, risk rating and credit rating; when the fuzzy analysis method is used to fuzzy the feature vector matrix, the consistency test is performed on the feature vector matrix, and when the test fails, the feature vector matrix is ​​adjusted; Determine multiple evaluation parameters in the preset risk quota balancing rules based on the customer's qualification rating, risk rating and quota rating; Obtain the corresponding average eigenvector in the preset loss function and fill it in; calculate the customer's credit limit based on the preset loss function, the preset loss function The calculation formula is: ， in, is the optimization function, is the reward function, is the quota size function, is the amount function, is the i-th average eigenvector in the total average eigenvector, is the weight coefficient, is the risk function.

2. The loan assessment method according to claim 1, characterized in that: Also includes: Regularly obtain customer change information to update the corresponding data in each database, and dynamically adjust the corresponding average feature vector and mapping relationship map according to the updated data; When the change information falls within the preset change conditions, the customer's secondary credit limit is determined based on the updated average feature vector and mapping relationship map; The secondary credit limit is used as the user's current credit limit.

3. The loan assessment method according to claim 1, characterized in that: Also includes: The preset risk quota balancing rules are encrypted using a key encryption method; When accessing the preset risk quota balancing rules, the corresponding key is used for decryption; And randomly generate new keys after the keys are used.

4. A loan evaluation system based on multi-objective optimization, wherein the loan evaluation system is used to execute the loan evaluation method according to any one of claims 1 to 3.

Images