Credit risk control method and device based on multi-source data fusion and reinforcement learning

Abstract

The invention discloses a credit risk control method and device based on multi-source data fusion and reinforcement learning, and relates to the field of financial science and technology or other related fields, and the method comprises the steps: responding to a loan application initiated by a borrower, obtaining a credit data set from a plurality of data sources, carrying out the preprocessing of the credit data set, and obtaining a credit data set; determining a weight vector of each piece of credit data in the credit data set by adopting an analytic hierarchy process strategy, fusing all the credit data based on the weight vector to obtain a target feature vector, inputting the target feature vector into a pre-constructed risk assessment model to obtain a model output result, the model output result comprising a risk assessment score, and when the capital return rate of the target financial capital applied by the borrower is greater than a preset return rate threshold value, the loan application of the borrower is approved, and the loan amount, the interest rate and the deadline are determined based on the risk assessment score. According to the invention, the technical problem that credit risk assessment cannot comprehensively and accurately reflect the credit status of the borrower in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the field of financial technology or other related fields, and in particular, to a credit risk control method and device based on multi-source data fusion and enhanced learning. Background Art

[0002] With the rapid development of Internet finance, the demand for credit is growing, and there are higher requirements for the convenience and immediacy of credit services. Financial institutions have moved the credit approval process online, pursuing a more efficient and faster approval mechanism. However, although online and Internet-based processes have improved efficiency, they have also brought new challenges, especially the prevention and control of credit risks. In the massive online credit applications, accurately identifying and assessing risks is an important issue for financial institutions.

[0003] In related technologies, the credit risk assessment strategies of financial institutions usually rely on a single data source and a fixed assessment model. A single data source cannot fully capture the credit status of borrowers, and fixed assessment models lack flexibility and self-updating capabilities. In the context of the rapid digitalization of financial services and the increasingly complex market environment, this limitation is more prominent and cannot fully and accurately reflect the credit status of borrowers.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiment of the present invention provides a credit risk control method and device based on multi-source data fusion and reinforcement learning, so as to at least solve the technical problem in the related art that credit risk assessment cannot fully and accurately reflect the credit status of the borrower.

[0006] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a credit risk control method based on multi-source data fusion and reinforcement learning is provided, the method comprising: responding to a loan application initiated by a borrower, obtaining a credit data set of the borrower from multiple data sources, and preprocessing the credit data set, wherein the credit data set at least includes: basic customer information, financial statements, credit records and transaction flows; using a hierarchical analysis strategy to determine the weight vector of each credit data in the preprocessed credit data set, and based on the weight vector, fusing all credit data to obtain a target feature vector; inputting the target feature vector into a pre-constructed risk assessment model to obtain a model output result, wherein the model output result at least includes: a risk assessment score; in the case where the capital return rate of the target financial capital applied for by the borrower is greater than a preset return rate threshold, approving the borrower's loan application, and determining the loan amount, interest rate and term based on the risk assessment score.

[0007] Optionally, the step of obtaining a credit data set of a borrower from multiple data sources includes: obtaining basic customer information from an internal information system of a financial institution or a third-party agency, wherein the basic customer information includes: the borrower's age, occupation and income; obtaining financial statements from a data file provided by the borrower or a third-party agency, wherein the financial statements include: the borrower's assets and liabilities; obtaining credit records from an internal information system of a financial institution or a third-party agency, wherein the credit records include: historical lending information, number of overdue payments, amount of debt owed and violation records; obtaining transaction flows from an internal information system of a financial institution, wherein the transaction flows include: the borrower's historical transaction business, transaction frequency, transaction amount and capital flow information.

[0008] Optionally, the data types of various credit data in the credit data set include: numerical data and categorical data, and the step of preprocessing the credit data set includes: removing duplicate data from the various credit data in the credit data set, and the step of removing duplicate data includes: calculating the similarity value between any two credit data in the credit data, and removing any one of the two data when the similarity value is greater than a preset similarity threshold; filling missing values ​​in the credit data with the mean, median or mode; when the credit data is categorical data, encoding the categorical data and converting the categorical data into numerical data; when the credit data is numerical data, performing data standardization on the numerical data and mapping the numerical data to a target interval; and normalizing the various credit data in the credit data set.

[0009] Optionally, the steps of using a hierarchical analysis strategy to determine the weight vectors of various credit data in a credit data set include: obtaining the importance ratios between the credit data, constructing a judgment matrix based on the importance ratios, and normalizing the judgment matrix; calculating the eigenvectors of the judgment matrix, normalizing the eigenvectors, and obtaining the weight vectors of the credit data, wherein the weight vectors corresponding to each credit data are different; calculating the consistency ratio of the weight vectors, and when the consistency ratio is greater than a preset index threshold, adjusting the judgment matrix and recalculating the weight vectors, wherein the consistency ratio is used to measure the logical consistency of the judgment matrix.

[0010] Optionally, the risk assessment model is trained in the following manner: the risk assessment model is initialized using a reinforcement learning algorithm, and a state space, an action space, a reward function, and a Q value are defined, wherein the state space includes: a target feature vector, the action space includes: loan approval action, loan rejection action, and loan amount adjustment action, the reward function includes: expected return and expected risk, and the Q value is the expected return of selecting the target action under the current state; the risk assessment model is trained based on a historical target feature vector, and the training steps include: inputting the historical target feature vector into the risk assessment model, and updating the risk assessment model based on the output Q value; when the Q value reaches the convergence condition, the training of the risk assessment model is terminated.

[0011] Optionally, the step of inputting the target feature vector into the risk assessment model and calculating the risk assessment score includes: inputting the target feature vector into the risk assessment model to obtain the default risk probability and the default risk loss; obtaining the borrower's historical credit score; and calculating the risk assessment score based on the default risk probability, the default risk loss and the historical credit score.

[0012] Optionally, the method also includes: obtaining the target financial capital applied for by the borrower, and extracting the capital loan interest rate and capital loan term from the model output results, and calculating the expected return based on the target financial capital, the capital loan interest rate and the capital loan term; obtaining the loan amount cost based on the target financial capital, obtaining the loan base cost from the internal information system of the financial institution, and calculating the operating cost based on the loan amount cost and the loan base cost; calculating the risk exposure value based on the target financial capital and the probability of default risk, wherein the risk exposure value is used to measure the risk of lending by the financial institution; obtaining the risk-free interest rate value from the internal information system of the financial institution, wherein the risk-free interest rate value is used to measure the return obtained by the financial institution under risk-free conditions; calculating the capital return rate of the target financial capital applied for by the borrower based on the expected return, operating cost, risk exposure value and risk-free interest rate value.

[0013] Optionally, the step of determining the loan amount, interest rate and term based on the risk assessment score includes: obtaining a benchmark loan amount in the model output results; obtaining a benchmark interest rate and a benchmark term from the internal information system of the financial institution; and determining the loan amount, interest rate and term based on the benchmark loan amount, benchmark interest rate and benchmark term according to the risk assessment score, wherein the loan amount and term are directly proportional to the risk assessment score, and the interest rate is inversely proportional to the risk assessment score.

[0014] According to another aspect of an embodiment of the present invention, a credit risk control device based on multi-source data fusion and reinforcement learning is also provided, and the device includes: an acquisition unit, which is used to respond to a loan application initiated by a borrower, obtain the borrower's credit data set from multiple data sources, and pre-process the credit data set, wherein the credit data set at least includes: basic customer information, financial statements, credit records and transaction flows; a fusion unit, which is used to determine the weight vector of each credit data in the pre-processed credit data set using a hierarchical analysis strategy, and fuse all credit data based on the weight vector to obtain a target feature vector; an input unit, which is used to input the target feature vector into a pre-built risk assessment model to obtain a model output result, wherein the model output result at least includes: a risk assessment score; a decision unit, which is used to approve the borrower's loan application when the capital return rate of the target financial capital applied by the borrower is greater than a preset return rate threshold, and determine the loan amount, interest rate and term based on the risk assessment score.

[0015] Optionally, the acquisition unit includes: a first acquisition module, used to acquire basic customer information from the internal information system of a financial institution or a third-party institution, wherein the basic customer information includes: the borrower's age, occupation and income; a second acquisition module, used to acquire financial statements from a data file provided by the borrower or a third-party institution, wherein the financial statements include: the borrower's assets and liabilities; a third acquisition module, used to acquire credit records from the internal information system of a financial institution or a third-party institution, wherein the credit records include: historical lending information, number of overdue payments, amount of debt and violation records; a fourth acquisition module, used to acquire transaction flows from the internal information system of a financial institution, wherein the transaction flows include: the borrower's historical transaction business, transaction frequency, transaction amount and capital flow information.

[0016] Optionally, the acquisition unit also includes: a first preprocessing module, which is used for the data types of various credit data in the credit data set, including numerical data and categorical data, to remove duplicate data from the various credit data in the credit data set, and the step of removing duplicate data includes: calculating the similarity value between any two credit data in the credit data, and removing any one of the two data when the similarity value is greater than a preset similarity threshold; using the mean, median or mode to fill the missing values ​​in the credit data; a second preprocessing module, which is used for encoding the categorical data and converting the categorical data into numerical data when the credit data is categorical data; a third preprocessing module, which is used for data standardization of the numerical data and mapping the numerical data to a target interval when the credit data is numerical data; a fourth preprocessing module, which is used to normalize the various credit data in the credit data set.

[0017] Optionally, the fusion unit includes: a construction module, used to obtain the importance ratio between credit data, construct a judgment matrix based on the importance ratio, and normalize the judgment matrix; a first calculation module, used to calculate the eigenvector of the judgment matrix, normalize the eigenvector, and obtain the weight vector of the credit data, wherein the weight vector corresponding to each credit data is different; a second calculation module, used to calculate the consistency ratio of the weight vector, when the consistency ratio is greater than a preset index threshold, adjust the judgment matrix and recalculate the weight vector, wherein the consistency ratio is used to measure the logical consistency of the judgment matrix.

[0018] Optionally, the input unit includes: a definition module, which is used to initialize the risk assessment model using a reinforcement learning algorithm, define a state space, an action space, a reward function and a Q value, wherein the state space includes: a target feature vector, the action space includes: loan approval action, loan rejection action, loan amount adjustment action, the reward function includes: expected return and expected risk, and the Q value is the expected return of selecting the target action under the current state; a training module, which trains the risk assessment model based on historical target feature vectors, and the training steps include: inputting the historical target feature vectors into the risk assessment model, and updating the risk assessment model based on the output Q value; when the Q value reaches the convergence condition, the training of the risk assessment model is ended.

[0019] Optionally, the input unit also includes: a fifth acquisition module, used to input the target feature vector into the risk assessment model to obtain the default risk probability and the default risk loss; a sixth acquisition module, used to obtain the borrower's historical credit score; and a third calculation module, which calculates the risk assessment score based on the default risk probability, the default risk loss and the historical credit score.

[0020] Optionally, the decision unit includes: a fourth calculation module, which is used to obtain the target financial capital applied for by the borrower, and extract the capital loan interest rate and capital loan term in the model output results, and calculate the expected return based on the target financial capital, capital loan interest rate and capital loan term; a fifth calculation module, which obtains the loan amount cost based on the target financial capital, obtains the loan base cost from the internal information system of the financial institution, and calculates the operating cost based on the loan amount cost and the loan base cost; a sixth calculation module, which calculates the risk exposure value based on the target financial capital and the default risk probability, wherein the risk exposure value is used to measure the risk of lending by the financial institution; a seventh acquisition module, which is used to obtain the risk-free interest rate value from the internal information system of the financial institution, wherein the risk-free interest rate value is used to measure the return obtained by the financial institution under risk-free conditions; the seventh calculation module, which calculates the capital return rate of the target financial capital applied for by the borrower based on the expected return, operating cost, risk exposure value and risk-free interest rate value.

[0021] Optionally, the decision unit also includes: an eighth acquisition module, used to obtain the benchmark loan amount in the model output result; a ninth acquisition module, used to obtain the benchmark interest rate and benchmark term from the internal information system of the financial institution; a determination module, based on the benchmark loan amount, benchmark interest rate and benchmark term, determines the loan amount, interest rate and term according to the risk assessment score, wherein the loan amount and term are directly proportional to the risk assessment score, and the interest rate is inversely proportional to the risk assessment score.

[0022] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned credit risk control methods based on multi-source data fusion and reinforcement learning.

[0023] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any one of the above-mentioned credit risk control methods based on multi-source data fusion and reinforcement learning.

[0024] According to another aspect of an embodiment of the present invention, a computer program product is provided, including a computer program, which, when executed by a processor, implements the steps of any one of the above-mentioned credit risk control methods based on multi-source data fusion and reinforcement learning.

[0025] In the present disclosure, by responding to the loan application initiated by the borrower, the credit data set of the borrower is obtained from multiple data sources, and the credit data set is preprocessed, and then the hierarchical analysis strategy is used to determine the weight vector of each credit data in the preprocessed credit data set, and all credit data are fused based on the weight vector to obtain the target feature vector, and then the target feature vector is input into a pre-built risk assessment model to obtain the model output result, wherein the model output result at least includes: risk assessment score, and finally, if the capital return rate of the target financial capital applied by the borrower is greater than the preset return rate threshold, the borrower's loan application is approved, and the loan amount, interest rate and term are determined based on the risk assessment score. The present disclosure can form a comprehensive and multi-dimensional credit portrait of the borrower by fusing multi-source credit data, and use a risk assessment model to predict the loan risk, and can accurately assess the credit status of the borrower, thereby solving the technical problem that the credit risk assessment in the related technology cannot comprehensively and accurately reflect the credit status of the borrower. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0027] Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing a credit risk control method based on multi-source data fusion and enhanced learning is shown;

[0028] Figure 2 is a flowchart of an optional credit risk control method based on multi-source data fusion and reinforcement learning according to an embodiment of the present invention;

[0029] Figure 3 is a schematic diagram of another optional credit risk control method based on multi-source data fusion and reinforcement learning according to an embodiment of the present invention;

[0030] Figure 4 is a schematic diagram of an optional credit risk control device based on multi-source data fusion and reinforcement learning according to an embodiment of the present invention;

[0031] Figure 5 It is a structural block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0033] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0034] It should be noted that the credit risk control method based on multi-source data fusion and reinforcement learning and the device thereof in the present disclosure can be used in the field of financial technology. When credit risk control is performed based on multi-source data fusion and reinforcement learning, it can also be used in any field other than the field of financial technology. When credit risk control is performed based on multi-source data fusion and reinforcement learning, the present disclosure does not limit the application field of the credit risk control method based on multi-source data fusion and reinforcement learning and the device thereof.

[0035] It should be noted that the information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) collected by this disclosure are information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data are in compliance with the relevant laws, regulations and standards of the relevant regions, necessary confidentiality measures are taken, and public order and good customs are not violated, and corresponding operation portals are provided for users to choose to authorize or refuse. For example, an interface is set up between this system and relevant users or organizations. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or organization through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or organization.

[0036] It should be noted that in the present disclosure, when collecting and analyzing customer information, corresponding operation entrances are provided for users to choose to agree or reject the automated decision results; if the user chooses to reject, the expert decision-making process is entered.

[0037] The following embodiments of the present invention can be applied to various systems / applications / devices for credit risk control based on multi-source data fusion and reinforcement learning. The present invention obtains a credit data set of a borrower from multiple data sources, pre-processes the credit data set, and then uses a hierarchical analysis strategy to determine the weight vector of each credit data in the pre-processed credit data set, and fuses all credit data based on the weight vector to obtain a target feature vector, thereby forming a comprehensive and multi-dimensional credit profile of the borrower, and then inputs the target feature vector into a pre-built risk assessment model, which can accurately assess the credit status of the borrower, approve the borrower's loan application when the capital return rate of the target financial capital applied by the borrower is greater than the preset return rate threshold, and determine the loan amount, interest rate and term based on the risk assessment score, completing a comprehensive and accurate loan process.

[0038] The present invention is described in detail below in conjunction with various embodiments.

[0039] Embodiment 1

[0040] According to an embodiment of the present invention, an embodiment of a credit risk control method based on multi-source data fusion and reinforcement learning is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0041] The credit risk control method embodiment based on multi-source data fusion and reinforcement learning provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 The hardware structure block diagram of a computer terminal (or mobile device) for implementing a credit risk control method based on multi-source data fusion and reinforcement learning is shown. Figure 1 As shown, the computer terminal 10 (or mobile device) may include one or more ( Figure 1 The computer system 102 includes a processor 102 (the processor 102 may include but is not limited to a microprocessor MCU (Microcontroller Unit) or a programmable logic device FPGA (Field Programmable Gate Array)), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a Universal Serial Bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. A person skilled in the art can understand that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0042] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10 (or mobile device). As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0043] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the credit risk control method based on multi-source data fusion and enhanced learning in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, the credit risk control method based on multi-source data fusion and enhanced learning is realized. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0044] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0045] The display may be, for example, a touch screen liquid crystal display (LCD), which enables a user to interact with a user interface of the computer terminal 10 (or mobile device).

[0046] Under the above operating environment, this application provides Figure 2 The credit risk control method based on multi-source data fusion and reinforcement learning is shown. Figure 2 is a flowchart of an optional credit risk control method based on multi-source data fusion and reinforcement learning according to an embodiment of the present invention, such as Figure 2 As shown, the method comprises the following steps:

[0047] Step S201, in response to a loan application initiated by a borrower, obtain a credit data set of the borrower from multiple data sources, and pre-process the credit data set, wherein the credit data set includes at least: basic customer information, financial statements, credit records and transaction flow.

[0048] In an embodiment of the present invention, after a borrower initiates a loan application, the embodiment of the present invention responds to the application and obtains a credit data set related to the borrower from multiple data sources, aiming to build a comprehensive and detailed credit profile to support more accurate risk assessment. After the credit data set is collected, it is pre-processed to ensure the quality and availability of the data.

[0049] Multi-source data fusion refers to the collection and processing of data from different channels, different formats, and different qualities to form a comprehensive, multi-dimensional and accurate data set. In credit risk control, multi-source data includes internal data of financial institutions, credit data, social network data, e-commerce platform transaction data, etc. These data can provide information about the borrower's financial status, credit history, social behavior, consumption habits, etc., and thus more comprehensively reflect the borrower's credit status. The strategy of multi-source data fusion can overcome the information limitations brought by a single data source and improve the accuracy and comprehensiveness of credit risk assessment. At the same time, pre-processing the collected multi-source data can effectively remove noise and redundant information in the data, providing high-quality data support for subsequent risk assessment models.

[0050] In an embodiment of the present invention, the multi-source data obtained from multiple data sources include: basic customer information, financial statements, credit records and transaction flows, which constitute a credit data set, providing a solid data foundation for subsequent risk assessment and risk control. It should be noted that the multi-source data of the embodiment of the present invention is not limited to basic customer information, financial statements, credit records and transaction flows, but may also include any other data that helps assess the borrower's lending risk. The comprehensive use of such multi-source data can not only improve the comprehensiveness and accuracy of credit risk assessment, but also help discover potential credit risk points, thereby helping financial institutions to generate more reasonable credit policies, effectively control credit risks, and improve the quality of credit business.

[0051] Step S202, using a hierarchical analysis strategy to determine the weight vector of each credit data in the preprocessed credit data set, and based on the weight vector, all credit data are merged to obtain a target feature vector.

[0052] Among the multi-source data collected, each type of data (including basic customer information, financial statements, credit records and transaction flows) comes from an independent data source. In order to effectively integrate these credit data from different channels and with different properties, the embodiment of the present invention adopts hierarchical analysis to determine the weight vectors of various credit data in the credit data set, and then gives each type of credit data an appropriate degree of importance during the fusion process, ensuring that the final target feature vector is both comprehensive and accurate.

[0053] After determining the weight vector of each credit data in the credit data set, all credit data are fused using a weighted average strategy to obtain the target feature vector. The fusion formula is:

[0054]

[0055] Among them, F is the target feature vector, n is the number of credit data, D i is the characteristic vector of the i-th credit data, w i is the weight vector of the i-th credit data. It should be noted that the feature vector of the credit data is obtained after feature extraction of the credit data. The feature extraction strategy depends on the type of credit data and can be implemented by using deep learning models or mathematical transformation strategies.

[0056] Step S203: input the target feature vector into a pre-built risk assessment model to obtain a model output result, wherein the model output result at least includes: a risk assessment score.

[0057] After collecting, preprocessing and integrating multi-source credit data to form a comprehensive and accurate target feature vector, the target feature vector is input into the pre-built risk assessment model to generate a quantitative evaluation of the borrower's credit risk. When the target feature vector is input into the model, the risk assessment model starts running. Combining previous learning experience and the input target feature vector, the risk assessment model will calculate the risk assessment score for each loan application. The risk assessment score is obtained by analyzing the comprehensive characteristics of the borrower. The risk assessment model predicts the borrower's default risk probability and default risk loss based on multi-source data such as the borrower's personal information, financial status, credit record, and transaction flow, and calculates the risk assessment score based on these predicted data. The higher the risk assessment score, the lower the risk of the borrower, and the more suitable for the loan. The lower the risk assessment score, the higher the risk of the borrower, and the more careful the loan strategy needs to be considered.

[0058] Step S204: if the capital return rate of the target financial capital applied by the borrower is greater than the preset return rate threshold, the borrower's loan application is approved, and the loan amount, interest rate and term are determined based on the risk assessment score.

[0059] In an embodiment of the present invention, the indicator for determining whether to approve a loan for a borrower is the size of the capital return rate. The capital return rate is an indicator for measuring credit risk and return. The embodiment of the present invention obtains the capital return rate of the target financial capital applied for by the borrower, and compares it with a preset return rate threshold. If the capital return rate is greater than the preset return rate threshold, the borrower's loan application is approved; if the capital return rate is not greater than the preset return rate threshold, the borrower's loan application is rejected.

[0060] After approving a loan application, the embodiment of the present invention dynamically adjusts key conditions such as loan amount, interest rate and term based on the risk assessment score, i.e., the quantitative assessment result of the borrower's credit risk. This dynamic adjustment mechanism of loan conditions based on risk assessment scores enables the embodiment of the present invention to effectively control credit risks and achieve optimal allocation of credit resources while ensuring the profit target. Through this dynamic adjustment mechanism, the embodiment of the present invention can provide personalized and differentiated loan services based on the specific credit status of each borrower, which not only meets the needs of borrowers, but also ensures the safe and efficient operation of the credit business of financial institutions.

[0061] Optionally, the step of obtaining a credit data set of a borrower from multiple data sources includes: obtaining basic customer information from an internal information system of a financial institution or a third-party agency, wherein the basic customer information includes: the borrower's age, occupation and income; obtaining financial statements from a data file provided by the borrower or a third-party agency, wherein the financial statements include: the borrower's assets and liabilities; obtaining credit records from an internal information system of a financial institution or a third-party agency, wherein the credit records include: historical lending information, number of overdue payments, amount of debt owed and violation records; obtaining transaction flows from an internal information system of a financial institution, wherein the transaction flows include: the borrower's historical transaction business, transaction frequency, transaction amount and capital flow information.

[0062] In step S201, in order to comprehensively and accurately assess the risk of the borrower, the embodiment of the present invention obtains the credit data set of the borrower from multiple data sources. First, the basic information of the customer is obtained from the internal information system of the financial institution or a third-party institution, and the basic information of the customer includes the age, occupation and income of the borrower; second, the financial statement is obtained from the data file provided by the borrower or a third-party institution, and the financial statement includes the assets and liabilities of the borrower; third, the credit record is obtained from the internal information system of the financial institution or a third-party institution, and the credit record includes the historical loan information, number of overdue payments, amount of debt and violation records of the borrower; finally, the transaction flow is obtained from the internal information system of the financial institution, and the transaction flow includes the historical transaction business, transaction frequency, transaction amount and capital flow information of the borrower.

[0063] The embodiment of the present invention constructs a multi-dimensional credit data set by comprehensively and systematically collecting and organizing the borrower's basic customer information, financial statement data, credit records and transaction flow information, providing a solid data foundation for subsequent credit risk assessment and control. This process not only involves in-depth mining of the internal system data of financial institutions, but also covers extensive cooperation with external institutions (third-party institutions), ensuring the richness and accuracy of the credit data set, thereby providing more scientific and refined risk assessment services for the credit business of financial institutions.

[0064] It should be noted that the embodiments of the present invention can also set the frequency and scope of data collection according to the characteristics of different data sources. For example, basic customer information can be updated regularly at long intervals, and transaction flows and credit records need to be updated in real time or at a high frequency to ensure the timeliness, completeness and accuracy of the data.

[0065] Optionally, the data types of various credit data in the credit data set include: numerical data and categorical data, and the step of preprocessing the credit data set includes: removing duplicate data from the various credit data in the credit data set, and the step of removing duplicate data includes: calculating the similarity value between any two credit data in the credit data, and removing any one of the two data when the similarity value is greater than a preset similarity threshold; filling missing values ​​in the credit data with the mean, median or mode; when the credit data is categorical data, encoding the categorical data and converting the categorical data into numerical data; when the credit data is numerical data, performing data standardization on the numerical data and mapping the numerical data to a target interval; and normalizing the various credit data in the credit data set.

[0066] In step S201, the data types of the collected credit data include numerical data and categorical data. For example, the income and assets of the borrower belong to numerical data, and the occupation and gender of the borrower belong to categorical data. When preprocessing the credit data, the credit data is first processed to remove duplicate data. The embodiment of the present invention determines whether there is duplicate data by calculating the similarity value of two data, and the similarity calculation is performed for numerical data and categorical data respectively. For numerical data, Euclidean distance or cosine similarity can be used to measure the proximity between the two data; for categorical data, Hamming distance or Jaccard distance can be used to measure the proximity between the two data. When judging whether there is duplicate data, a preset similarity threshold is usually set, such as 0.85. If the calculated similarity value is greater than the preset similarity threshold, they are considered to be duplicate data, and any one of the two data needs to be removed, and only one of the data is retained to maintain the purity of the credit data.

[0067] After deduplication of credit data, if there are missing values, they need to be filled to ensure the integrity of the credit data. For numerical data, the mean or median can be used to fill missing values; for categorical data, the mode can be used to fill missing values.

[0068] After filling the credit data, the categorical data also needs to be encoded. Hot encoding can be used to convert the categorical data into numerical data. For numerical data, it needs to be standardized and mapped within the target range. The calculation formula for standardization is:

[0069]

[0070] Among them, x′ is the standardized credit data, x is the credit data, and x min is the minimum value in each credit data set, x max is the maximum value in each credit data set. This calculation formula can map numerical data to the interval [0,1] to facilitate the subsequent processing of credit data of different magnitudes.

[0071] Finally, the credit data is normalized. The calculation formula for normalization in the embodiment of the present invention is:

[0072]

[0073] Among them, x″ is the normalized credit data, x′ is the credit data after deduplication, padding, and standardization (or encoding), μ is the mean of each credit data set, and σ is the standard deviation of each credit data set. Normalization can ensure that the credit data are on the same scale, which is convenient for the subsequent risk assessment model processing.

[0074] Through a series of preprocessing operations, the accuracy and consistency of credit data are ensured, providing a solid data foundation for subsequent risk assessment.

[0075] It should be noted that the credit data set includes not only numerical data and categorical data, but also other types of data, such as time series data or text data. After preprocessing the time series data or text data, these data can be merged with the numerical data and categorical data.

[0076] Optionally, the steps of using a hierarchical analysis strategy to determine the weight vectors of various credit data in a credit data set include: obtaining the importance ratios between the credit data, constructing a judgment matrix based on the importance ratios, and normalizing the judgment matrix; calculating the eigenvectors of the judgment matrix, normalizing the eigenvectors, and obtaining the weight vectors of the credit data, wherein the weight vectors corresponding to each credit data are different; calculating the consistency ratio of the weight vectors, and when the consistency ratio is greater than a preset index threshold, adjusting the judgment matrix and recalculating the weight vectors, wherein the consistency ratio is used to measure the logical consistency of the judgment matrix.

[0077] In step S202, when using the hierarchical analysis strategy to determine the weight vectors of various credit data in the credit data set, it is first necessary to obtain the importance ratio between the credit data. The importance ratio is usually set based on experience, and the importance ratio of one credit data relative to another credit data is determined by comparing the importance between the two credit data. When determining the importance ratio, a scaling strategy is usually adopted, where "1" represents that the two credit data are equally important, "9" represents that one credit data is much more important than the other credit data, and the values ​​between "0 to 9" reflect the difference in importance to different degrees.

[0078] Based on the determined importance ratio, a judgment matrix A is constructed, which is:

[0079]

[0080] Where n is the number of types of credit data, A ij is the importance ratio of the i-th credit data to the j-th credit data, for example, “A ij =3" means that the importance of the i-th credit data is three times that of the j-th credit data, "A ij =1 / 3" means that the importance of the i-th type of credit data is 1 / 3 of the j-th type of credit data.

[0081] After constructing the judgment matrix, the judgment matrix is ​​normalized, and then the eigenvector of the judgment matrix, that is, the relative weight of various credit data, is calculated. The calculation formula is:

[0082]

[0083] Among them, W i is the relative weight of the i-th credit data, B ij is the j-th element in the i-th row of the judgment matrix, and n is the number of elements in the i-th row (i.e., the number of types of credit data).

[0084] Then, based on the relative weights of all credit data, the relative weights of various credit data are normalized to obtain the weight vectors of various credit data. At the same time, since the judgment matrix is ​​set based on experience, in order to ensure the logical consistency of the judgment matrix, the weight vector needs to be checked for consistency. In the process of consistency checking, the maximum eigenvalue of the judgment matrix is ​​first calculated, and then the consistency index is calculated. The calculation formula of the consistency index is:

[0085]

[0086] Among them, CI is the consistency index, λ maxis the maximum eigenvalue of the judgment matrix, and n is the number of types of credit data. After calculating the consistency index, the consistency ratio is calculated. The calculation formula of the consistency ratio is:

[0087]

[0088] Among them, CR is the consistency ratio, CI is the consistency index, and RI is the random consistency index, where the random consistency index is obtained from the hierarchical analysis strategy standard table according to the number of categories n. After the consistency ratio is calculated, it is compared with the preset index threshold. The preset index threshold can be set to 0.1. When the consistency ratio is less than or equal to the preset index threshold, it is considered that the consistency check has passed. When the consistency ratio is greater than the preset index threshold, it is necessary to adjust the judgment matrix and recalculate the weight vector of each credit data.

[0089] Different credit data have different impacts on credit risk, and the weight vectors of different credit data are not the same. Different weight vectors are calculated through hierarchical analysis strategy. In this process, the impact of multi-source data on the risk assessment model is comprehensively considered, making credit risk assessment more reasonable.

[0090] Optionally, the risk assessment model is trained in the following manner: the risk assessment model is initialized using a reinforcement learning algorithm, and a state space, an action space, a reward function, and a Q value are defined, wherein the state space includes: a target feature vector, the action space includes: loan approval action, loan rejection action, and loan amount adjustment action, the reward function includes: expected return and expected risk, and the Q value is the expected return of selecting the target action under the current state; the risk assessment model is trained based on a historical target feature vector, and the training steps include: inputting the historical target feature vector into the risk assessment model, and updating the risk assessment model based on the output Q value; when the Q value reaches the convergence condition, the training of the risk assessment model is terminated.

[0091] In step S203, the risk assessment model is pre-built and trained. First, the risk assessment model is initialized using a reinforcement learning algorithm to define the state space, action space, reward function, and Q value, where the state space includes the target feature vector, the action space includes loan actions such as loan approval action, loan rejection action, and loan amount adjustment action, the Q value is the expected return of selecting the target action in the current state, and the reward function includes expected return and expected risk, which are both output results of the risk assessment model.

[0092] When training the risk assessment model, the historical target feature vector is input into the risk assessment model, and the risk assessment model is updated based on the output Q value. This update process is continuously iterated, and the risk assessment model gradually maximizes the lending yield. Finally, when the Q value reaches the convergence condition, the training of the risk assessment model is terminated, and the trained risk assessment model is used in actual scenarios.

[0093] Optionally, the step of inputting the target feature vector into the risk assessment model and calculating the risk assessment score includes: inputting the target feature vector into the risk assessment model to obtain the default risk probability and the default risk loss; obtaining the borrower's historical credit score; and calculating the risk assessment score based on the default risk probability, the default risk loss and the historical credit score.

[0094] In step S203, the risk assessment score is calculated based on the output of the risk assessment model. After inputting the target feature vector into the risk assessment model, the embodiment of the present invention predicts the borrower's default risk probability and default risk loss. At the same time, the embodiment of the present invention obtains the borrower's historical credit score, and then calculates the risk assessment score based on the default risk probability, default risk loss and historical credit score. The calculation formula is:

[0095]

[0096] Among them, Score is the risk assessment score, P is the predicted default risk probability, L is the predicted default risk loss, R is the borrower's historical credit score, α, β, and γ are the weight coefficients of the default risk probability, default risk loss, and historical credit score, respectively, where the weight coefficient is set by yourself. It should be noted that R in this calculation formula may not be the historical credit score, and other credit risk factors that can measure the borrower's historical credit may be used. Through this calculation formula, the range of risk assessment scores is controlled within 0 to 1, which is used to measure the risk level of the borrower.

[0097] Optionally, the method also includes: obtaining the target financial capital applied for by the borrower, and extracting the capital loan interest rate and capital loan term from the model output results, and calculating the expected return based on the target financial capital, the capital loan interest rate and the capital loan term; obtaining the loan amount cost based on the target financial capital, obtaining the loan base cost from the internal information system of the financial institution, and calculating the operating cost based on the loan amount cost and the loan base cost; calculating the risk exposure value based on the target financial capital and the probability of default risk, wherein the risk exposure value is used to measure the risk of lending by the financial institution; obtaining the risk-free interest rate value from the internal information system of the financial institution, wherein the risk-free interest rate value is used to measure the return obtained by the financial institution under risk-free conditions; calculating the capital return rate of the target financial capital applied for by the borrower based on the expected return, operating cost, risk exposure value and risk-free interest rate value.

[0098] In step S204, the capital return rate is calculated based on the expected return, operating cost, risk exposure value and risk-free interest rate value, and the calculation formula is:

[0099]

[0100] In the above formula, RAROC is the return on capital, R is the expected return, C is the operating cost, E is the risk exposure value, and r is the risk-free interest rate value.

[0101] In the calculation parameters of the return on capital, the expected return is calculated based on the target financial capital, capital loan interest rate and capital loan term. First, obtain the target financial capital applied by the borrower, and then extract the capital loan interest rate and capital loan term from the output results of the risk assessment model. The calculation formula for the expected return is:

[0102] R=H×r loan ×T

[0103] In the above formula, H is the target financial capital, r loan is the capital loan interest rate, and T is the capital loan term.

[0104] The operating cost is calculated based on the loan amount cost and the loan base cost. First, the loan amount cost is obtained based on the target financial capital, and then the loan base cost is obtained from the internal information system of the financial institution. Finally, the operating cost is calculated based on the loan amount cost and the loan base cost. The calculation formula is:

[0105] C=C fixed +C var ×H

[0106] In the above formula, C is the operating cost, H is the target financial capital, and C fixed is the loan base cost, C var is the variable cost of the loan. The variable cost of the loan is also obtained from the internal information system of the financial institution.

[0107] The risk exposure value is calculated based on the target financial capital and the default risk probability output by the risk assessment model. The calculation formula is:

[0108] E=H×P

[0109] In the above formula, E is the risk exposure value, H is the target financial capital, and P is the probability of default risk. Risk exposure value is one of the indicators used to measure the lending risk of financial institutions.

[0110] The risk-free interest rate is an indicator used to measure the returns that financial institutions can obtain under risk-free conditions, and is usually obtained from the internal information system of the financial institution.

[0111] Optionally, the step of determining the loan amount, interest rate and term based on the risk assessment score includes: obtaining a benchmark loan amount in the model output results; obtaining a benchmark interest rate and a benchmark term from the internal information system of the financial institution; and determining the loan amount, interest rate and term based on the benchmark loan amount, benchmark interest rate and benchmark term according to the risk assessment score, wherein the loan amount and term are directly proportional to the risk assessment score, and the interest rate is inversely proportional to the risk assessment score.

[0112] In step S204, when the capital return rate is greater than the preset return rate threshold, the borrower's loan application is approved, and then the loan amount, interest rate and term are determined based on the risk assessment score. Specifically, the benchmark loan amount is first obtained from the output result of the risk assessment model, and then the benchmark interest rate and benchmark term are obtained from the internal information system of the financial institution. When determining the borrower's loan amount, based on the benchmark loan amount, the higher the risk assessment score, the larger the approved loan amount, and the lower the risk assessment score, the smaller the approved loan amount; when determining the borrower's loan term, based on the benchmark term, the higher the risk assessment score, the longer the approved loan term, and the lower the risk assessment score, the shorter the approved loan term; when determining the borrower's loan interest rate, based on the benchmark interest rate, the higher the risk assessment score, the smaller the approved loan interest rate, and the lower the risk assessment score, the larger the approved loan interest rate.

[0113] Figure 3 is a schematic diagram of another optional credit risk control method based on multi-source data fusion and reinforcement learning according to an embodiment of the present invention, such as Figure 3 As shown, the method first obtains multi-source data from the internal system of a financial institution or a third-party institution, and then pre-processes the obtained multi-source data including removing duplicate data, filling missing values, standardization and normalization, and then uses a hierarchical analysis strategy to determine the weight vector of the multi-source data after pre-processing, and then fuses the multi-source data based on the weight vector, inputs the fused target feature vector into a pre-built risk assessment model, obtains the model output results including the risk assessment score, and finally obtains the capital return rate related to the borrower, determines whether to approve the loan for the borrower based on the capital return rate, and then determines the loan amount, interest rate and term based on the risk assessment score. Finally, the embodiment of the present invention stores the evaluation results using a distributed system and a database, and displays the evaluation results in the form of charts.

[0114] The embodiment of the present invention can form a comprehensive and multi-dimensional credit profile of borrowers by fusing multi-source data, and build a risk assessment model using a reinforcement learning algorithm, which can more accurately predict the borrower's default risk. Based on this strategy, the embodiment of the present invention can automatically screen high-quality borrowers, identify potential risky borrowers, and provide financial institutions with detailed assessment reports and recommendation strategies, reduce the degree of manual intervention, and improve the processing efficiency of credit business. The embodiment of the present invention can reduce the overall risk level of the credit business of financial institutions by discovering potential risky borrowers and taking corresponding risk control measures in a timely manner. At the same time, the embodiment of the present invention supports customized risk thresholds and control rules, which can be adjusted and optimized according to the business needs and risk preferences of different financial institutions.

[0115] The steps provided by the credit risk control method based on multi-source data fusion and reinforcement learning are as follows: by responding to the loan application initiated by the borrower, the credit data set of the borrower is obtained from multiple data sources, and the credit data set is preprocessed, and then the weight vector of each credit data in the preprocessed credit data set is determined by using a hierarchical analysis strategy, and all credit data are fused based on the weight vector to obtain the target feature vector, and then the target feature vector is input into a pre-constructed risk assessment model to obtain the model output result, wherein the model output result at least includes: risk assessment score, and finally, when the capital return rate of the target financial capital applied by the borrower is greater than the preset return rate threshold, the borrower's loan application is approved, and the loan amount, interest rate and term are determined based on the risk assessment score. The embodiment of the present invention can form a comprehensive and multi-dimensional credit portrait of the borrower by fusing multi-source credit data, and use a risk assessment model to predict the loan risk, and can accurately assess the credit status of the borrower, thereby solving the technical problem that the credit risk assessment in the related technology cannot comprehensively and accurately reflect the credit status of the borrower.

[0116] The following is a detailed description in conjunction with another embodiment.

[0117] Embodiment 2

[0118] A credit risk control device based on multi-source data fusion and reinforcement learning provided in this embodiment includes multiple implementation units, each of which corresponds to each implementation step in the above-mentioned embodiment 1.

[0119] Figure 4 is a schematic diagram of an optional credit risk control device based on multi-source data fusion and reinforcement learning according to an embodiment of the present invention, such as Figure 4 As shown, the credit risk control device based on multi-source data fusion and enhanced learning may include: an acquisition unit 41, a fusion unit 42, an input unit 43, and a decision unit 44.

[0120] Among them, the acquisition unit 41 is used to respond to the loan application initiated by the borrower, obtain the borrower's credit data set from multiple data sources, and pre-process the credit data set, wherein the credit data set at least includes: customer basic information, financial statements, credit records and transaction flows.

[0121] The fusion unit 42 is used to determine the weight vector of each credit data in the pre-processed credit data set by adopting a hierarchical analysis strategy, and fuse all the credit data based on the weight vector to obtain a target feature vector.

[0122] The input unit 43 is used to input the target feature vector into a pre-built risk assessment model to obtain a model output result, wherein the model output result at least includes: a risk assessment score.

[0123] The decision unit 44 is used to approve the borrower's loan application when the capital return rate of the target financial capital applied by the borrower is greater than a preset return rate threshold, and determine the loan amount, interest rate and term based on the risk assessment score.

[0124] The above-mentioned credit risk control device based on multi-source data fusion and reinforcement learning can respond to the loan application initiated by the borrower through the acquisition unit 41, obtain the credit data set of the borrower from multiple data sources, and pre-process the credit data set, and then use the hierarchical analysis strategy to determine the weight vector of each credit data in the pre-processed credit data set through the fusion unit 42, and fuse all credit data based on the weight vector to obtain the target feature vector, and then input the target feature vector into the pre-constructed risk assessment model through the input unit 43 to obtain the model output result, wherein the model output result at least includes: risk assessment score, and finally, through the decision unit 44, if the capital return rate of the target financial capital applied by the borrower is greater than the preset return rate threshold, approve the borrower's loan application, and determine the loan amount, interest rate and term based on the risk assessment score. The embodiment of the present invention can form a comprehensive and multi-dimensional credit portrait of the borrower by fusing multi-source credit data, use the risk assessment model to predict the loan risk, and accurately assess the credit status of the borrower, thereby solving the technical problem that the credit risk assessment in the related technology cannot comprehensively and accurately reflect the credit status of the borrower.

[0125] Optionally, the acquisition unit 41 includes: a first acquisition module, used to acquire basic customer information from the internal information system of a financial institution or a third-party institution, wherein the basic customer information includes: the borrower's age, occupation and income; a second acquisition module, used to acquire financial statements from a data file provided by the borrower or a third-party institution, wherein the financial statements include: the borrower's assets and liabilities; a third acquisition module, used to acquire credit records from the internal information system of a financial institution or a third-party institution, wherein the credit records include: historical lending information, number of overdue payments, amount of debt and violation records; a fourth acquisition module, used to acquire transaction flows from the internal information system of a financial institution, wherein the transaction flows include: the borrower's historical transaction business, transaction frequency, transaction amount and capital flow information.

[0126] Optionally, the acquisition unit 41 also includes: a first preprocessing module, which is used for the data types of various credit data in the credit data set, including numerical data and categorical data, to remove duplicate data from the various credit data in the credit data set, and the step of removing duplicate data includes: calculating the similarity value between any two credit data in the credit data, and removing any one of the two data when the similarity value is greater than a preset similarity threshold; using the mean, median or mode to fill the missing values ​​in the credit data; a second preprocessing module, which is used for encoding the categorical data and converting the categorical data into numerical data when the credit data is categorical data; a third preprocessing module, which is used for data standardization of the numerical data and mapping the numerical data to a target interval when the credit data is numerical data; a fourth preprocessing module, which is used to normalize the various credit data in the credit data set.

[0127] Optionally, the fusion unit 42 includes: a construction module, used to obtain the importance ratio between credit data, construct a judgment matrix based on the importance ratio, and normalize the judgment matrix; a first calculation module, used to calculate the eigenvector of the judgment matrix, normalize the eigenvector, and obtain the weight vector of the credit data, wherein the weight vector corresponding to each credit data is different; a second calculation module, used to calculate the consistency ratio of the weight vector, when the consistency ratio is greater than a preset index threshold, adjust the judgment matrix and recalculate the weight vector, wherein the consistency ratio is used to measure the logical consistency of the judgment matrix.

[0128] Optionally, the input unit 43 includes: a definition module, which is used to initialize the risk assessment model using a reinforcement learning algorithm, define a state space, an action space, a reward function and a Q value, wherein the state space includes: a target feature vector, the action space includes: loan approval action, loan rejection action, loan amount adjustment action, the reward function includes: expected return and expected risk, and the Q value is the expected return of selecting the target action under the current state; a training module, which trains the risk assessment model based on a historical target feature vector, and the training steps include: inputting the historical target feature vector into the risk assessment model, and updating the risk assessment model based on the output Q value; when the Q value reaches the convergence condition, the training of the risk assessment model is ended.

[0129] Optionally, the input unit 43 also includes: a fifth acquisition module, used to input the target feature vector into the risk assessment model to obtain the default risk probability and the default risk loss; a sixth acquisition module, used to obtain the borrower's historical credit score; and a third calculation module, which calculates the risk assessment score based on the default risk probability, the default risk loss and the historical credit score.

[0130] Optionally, the decision unit 44 includes: a fourth calculation module, which is used to obtain the target financial capital applied for by the borrower, and extract the capital loan interest rate and capital loan term in the model output results, and calculate the expected return based on the target financial capital, the capital loan interest rate and the capital loan term; a fifth calculation module, which obtains the loan amount cost based on the target financial capital, obtains the loan base cost from the internal information system of the financial institution, and calculates the operating cost based on the loan amount cost and the loan base cost; a sixth calculation module, which calculates the risk exposure value based on the target financial capital and the default risk probability, wherein the risk exposure value is used to measure the risk of lending by the financial institution; a seventh acquisition module, which is used to obtain the risk-free interest rate value from the internal information system of the financial institution, wherein the risk-free interest rate value is used to measure the income obtained by the financial institution under risk-free conditions; the seventh calculation module, which calculates the capital return rate of the target financial capital applied for by the borrower based on the expected return, operating cost, risk exposure value and risk-free interest rate value.

[0131] Optionally, the decision unit 44 also includes: an eighth acquisition module, used to obtain the benchmark loan amount in the model output result; a ninth acquisition module, used to obtain the benchmark interest rate and benchmark term from the internal information system of the financial institution; a determination module, based on the benchmark loan amount, benchmark interest rate and benchmark term, determines the loan amount, interest rate and term according to the risk assessment score, wherein the loan amount and term are directly proportional to the risk assessment score, and the interest rate is inversely proportional to the risk assessment score.

[0132] The above-mentioned credit risk control device based on multi-source data fusion and reinforcement learning may also include a processor and a memory. The above-mentioned acquisition unit 41, fusion unit 42, input unit 43, decision unit 44, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize the corresponding functions.

[0133] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and credit risk control is performed based on multi-source data fusion and enhanced learning by adjusting kernel parameters.

[0134] The above-mentioned memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one storage chip.

[0135] Embodiment 3

[0136] An embodiment of the present application may provide an electronic device, Figure 5 is a structural block diagram of an electronic device according to an embodiment of the present application. Figure 5 As shown, the electronic device may include: one or more ( Figure 5 (only one is shown) processor 502, memory 504, storage controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0137] Among them, the memory can be used to store software programs and modules, such as program instructions / modules corresponding to the methods and devices in the embodiments of the present application, and the processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0138] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: respond to the loan application initiated by the borrower, obtain the borrower's credit data set from multiple data sources, and pre-process the credit data set, wherein the credit data set at least includes: basic customer information, financial statements, credit records and transaction flows; use a hierarchical analysis strategy to determine the weight vector of each credit data in the pre-processed credit data set, and fuse all credit data based on the weight vector to obtain a target feature vector; input the target feature vector into a pre-built risk assessment model to obtain a model output result, wherein the model output result at least includes: a risk assessment score; if the capital return rate of the target financial capital applied by the borrower is greater than a preset return rate threshold, approve the borrower's loan application, and determine the loan amount, interest rate and term based on the risk assessment score.

[0139] The processor can also call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain basic customer information from the internal information system of the financial institution or a third-party agency, wherein the basic customer information includes: the borrower's age, occupation and income; obtain financial statements from data files provided by the borrower or a third-party agency, wherein the financial statements include: the borrower's assets and liabilities; obtain credit records from the internal information system of the financial institution or a third-party agency, wherein the credit records include: historical loan information, number of overdue payments, amount of debt and violation records; obtain transaction flows from the internal information system of the financial institution, wherein the transaction flows include: the borrower's historical transaction business, transaction frequency, transaction amount and capital flow information.

[0140] The processor can also call the information and application programs stored in the memory through the transmission device to perform the following steps: the data types of various credit data in the credit data set include: numerical data and categorical data, and various credit data in the credit data set are processed to remove duplicate data. The step of removing duplicate data includes: calculating the similarity value between any two credit data in the credit data, and removing any one of the two data when the similarity value is greater than a preset similarity threshold; using the mean, median or mode to fill the missing values ​​in the credit data; when the credit data is categorical data, encoding the categorical data and converting the categorical data into numerical data; when the credit data is numerical data, data standardization is performed on the numerical data, and the numerical data is mapped to the target interval; various credit data in the credit data set are normalized.

[0141] The processor can also call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain the importance ratio between the credit data, construct a judgment matrix based on the importance ratio, and normalize the judgment matrix; calculate the eigenvector of the judgment matrix, normalize the eigenvector, and obtain the weight vector of the credit data, wherein the weight vector corresponding to each credit data is different; calculate the consistency ratio of the weight vector, and when the consistency ratio is greater than the preset index threshold, adjust the judgment matrix and recalculate the weight vector, wherein the consistency ratio is used to measure the logical consistency of the judgment matrix.

[0142] The processor can also call the information and application programs stored in the memory through the transmission device to perform the following steps: initialize the risk assessment model using a reinforcement learning algorithm, define the state space, action space, reward function and Q value, wherein the state space includes: target feature vector, the action space includes: loan approval action, loan rejection action, loan amount adjustment action, the reward function includes: expected return and expected risk, and the Q value is the expected return of selecting the target action in the current state; train the risk assessment model based on the historical target feature vector, the training step includes: inputting the historical target feature vector into the risk assessment model, and updating the risk assessment model based on the output Q value; when the Q value reaches the convergence condition, end the training of the risk assessment model.

[0143] The processor can also call the information and application programs stored in the memory through the transmission device to perform the following steps: input the target feature vector into the risk assessment model to obtain the default risk probability and default risk loss; obtain the borrower's historical credit score; calculate the risk assessment score based on the default risk probability, default risk loss and historical credit score.

[0144] The processor can also call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain the target financial capital applied for by the borrower, and extract the capital loan interest rate and capital loan term in the model output results, and calculate the expected return based on the target financial capital, capital loan interest rate and capital loan term; obtain the loan amount cost based on the target financial capital, obtain the loan base cost from the internal information system of the financial institution, and calculate the operating cost based on the loan amount cost and the loan base cost; calculate the risk exposure value based on the target financial capital and the default risk probability, wherein the risk exposure value is used to measure the risk of lending by the financial institution; obtain the risk-free interest rate value from the internal information system of the financial institution, wherein the risk-free interest rate value is used to measure the income obtained by the financial institution under risk-free conditions; calculate the capital return rate of the target financial capital applied for by the borrower based on the expected return, operating cost, risk exposure value and risk-free interest rate value.

[0145] The processor can also call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain the benchmark loan amount in the model output results; obtain the benchmark interest rate and benchmark term from the internal information system of the financial institution; based on the benchmark loan amount, benchmark interest rate and benchmark term, determine the loan amount, interest rate and term according to the risk assessment score, wherein the loan amount and term are directly proportional to the risk assessment score, and the interest rate is inversely proportional to the risk assessment score.

[0146] According to the embodiment of the present invention, a credit risk control method based on multi-source data fusion and reinforcement learning is provided. By responding to the loan application initiated by the borrower, the credit data set of the borrower is obtained from multiple data sources, and the credit data set is preprocessed. Then, the weight vector of each credit data in the preprocessed credit data set is determined by a hierarchical analysis strategy, and all credit data are fused based on the weight vector to obtain the target feature vector, and then the target feature vector is input into a pre-constructed risk assessment model to obtain the model output result, wherein the model output result at least includes: risk assessment score, and finally, when the capital return rate of the target financial capital applied by the borrower is greater than the preset return rate threshold, the borrower's loan application is approved, and the loan amount, interest rate and term are determined based on the risk assessment score. By fusing multi-source credit data, the embodiment of the present invention can form a comprehensive and multi-dimensional credit portrait of the borrower, use a risk assessment model to predict the loan risk, and accurately assess the credit status of the borrower, thereby solving the technical problem that the credit risk assessment in the related technology cannot comprehensively and accurately reflect the credit status of the borrower.

[0147] It can be understood by those skilled in the art that Figure 5 The structure shown is for illustration only, and the electronic device may also be a smart phone, a tablet computer, a PDA, a mobile Internet device (MID), a PAD, or other terminal devices. Figure 5 The structure of the electronic device is not limited. Figure 5 More or fewer components (such as network interfaces, display devices, etc.) shown in, or having Figure 5 Different configurations are shown.

[0148] A person skilled in the art can understand that all or part of the steps in the various credit risk control methods based on multi-source data fusion and reinforcement learning in the above-mentioned embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0149] Embodiment 4

[0150] The embodiment of the present application further provides a storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the credit risk control method based on multi-source data fusion and enhanced learning provided in the first embodiment.

[0151] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is further provided, the computer-readable storage medium including a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the credit risk control methods based on multi-source data fusion and reinforcement learning in the above-mentioned embodiment 1.

[0152] Optionally, in this embodiment, the above storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0153] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the credit risk control method based on multi-source data fusion and reinforcement learning described in each embodiment of the present application.

[0154] The present application also provides a computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the credit risk control method based on multi-source data fusion and reinforcement learning described in each embodiment of the present application are implemented.

[0155] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0156] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0157] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0158] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0159] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0160] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0161] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A credit risk control method based on multi-source data fusion and reinforcement learning, characterized in that: include: In response to a loan application initiated by a borrower, a credit data set of the borrower is obtained from multiple data sources, and the credit data set is preprocessed, wherein the credit data set includes at least: basic customer information, financial statements, credit records and transaction flow; A hierarchical analysis strategy is used to determine a weight vector of each credit data in the preprocessed credit data set, and all the credit data are fused based on the weight vector to obtain a target feature vector; Inputting the target feature vector into a pre-built risk assessment model to obtain a model output result, wherein the model output result at least includes: a risk assessment score; In the case that the capital return rate of the target financial capital applied for by the borrower is greater than a preset return rate threshold, the borrower's loan application is approved, and the loan amount, interest rate and term are determined based on the risk assessment score.

2. The credit risk control method according to claim 1, characterized in that: The step of obtaining the borrower's credit data set from multiple data sources includes: Obtaining the basic information of the customer from the internal information system of the financial institution or a third-party institution, wherein the basic information of the customer includes: the age, occupation and income of the borrower; Obtaining the financial statement from a data file provided by the borrower or a third-party institution, wherein the financial statement includes: the assets and liabilities of the borrower; Obtain the credit record from the internal information system of a financial institution or a third-party institution, wherein the credit record includes: historical loan information, number of overdue payments, amount of debt owed, and violation records; The transaction flow is obtained from the internal information system of a financial institution, wherein the transaction flow includes: the borrower's historical transaction business, transaction frequency, transaction amount and capital flow information.

3. The credit risk control method according to claim 1, characterized in that: The data types of various credit data in the credit data set include: numerical data and categorical data. The step of preprocessing the credit data set includes: Deduplication processing is performed on various credit data in the credit data set, and the step of deduplication processing includes: calculating the similarity value between any two credit data in the credit data, and removing any one of the two data when the similarity value is greater than a preset similarity threshold; filling the missing values ​​in the credit data with the mean, median or mode; In the case where the credit data is the classified data, encoding the classified data and converting the classified data into numerical data; In the case where the credit data is the numerical data, performing data standardization processing on the numerical data and mapping the numerical data into a target interval; The various credit data in the credit data set are normalized.

4. The credit risk control method according to claim 1, characterized in that: The step of using a hierarchical analysis strategy to determine the weight vectors of various credit data in the credit data set includes: Obtaining importance ratios between the credit data, constructing a judgment matrix based on the importance ratios, and normalizing the judgment matrix; Calculating the eigenvector of the judgment matrix, normalizing the eigenvector, and obtaining the weight vector of the credit data, wherein the weight vector corresponding to each credit data is different; The consistency ratio of the weight vector is calculated, and when the consistency ratio is greater than a preset index threshold, the judgment matrix is ​​adjusted and the weight vector is recalculated, wherein the consistency ratio is used to measure the logical consistency of the judgment matrix.

5. The credit risk control method according to claim 1, characterized in that: The risk assessment model is trained in the following way: The risk assessment model is initialized by using a reinforcement learning algorithm, and a state space, an action space, a reward function and a Q value are defined, wherein the state space includes: the target feature vector, the action space includes: loan approval action, loan rejection action, and loan amount adjustment action, the reward function includes: expected return and expected risk, and the Q value is the expected return of selecting the target action in the current state; The risk assessment model is trained based on the historical target feature vector, and the training steps include: inputting the historical target feature vector into the risk assessment model, updating the risk assessment model based on the output Q value; and ending the training of the risk assessment model when the Q value reaches a convergence condition.

6. The credit risk control method according to claim 1, characterized in that: The step of inputting the target feature vector into the risk assessment model and calculating the risk assessment score comprises: Inputting the target feature vector into the risk assessment model to obtain the default risk probability and default risk loss; Obtaining the historical credit score of the borrower; The risk assessment score is calculated based on the default risk probability, the default risk loss and the historical credit score.

7. The credit risk control method according to claim 1, characterized in that: Also includes: Obtaining the target financial capital applied for by the borrower, extracting the capital loan interest rate and the capital loan term from the output result of the model, and calculating the expected return based on the target financial capital, the capital loan interest rate and the capital loan term; Obtaining a loan amount cost based on the target financial capital, obtaining a loan base cost from an internal information system of a financial institution, and calculating an operating cost based on the loan amount cost and the loan base cost; Calculate the risk exposure value based on the target financial capital and the default risk probability, wherein the risk exposure value is used to measure the risk of lending by financial institutions; Obtaining a risk-free interest rate value from an internal information system of a financial institution, wherein the risk-free interest rate value is used to measure the income obtained by the financial institution under a risk-free situation; The capital return rate of the target financial capital applied for by the borrower is calculated based on the expected return, the operating cost, the risk exposure value and the risk-free interest rate value.

8. The credit risk control method according to claim 1, characterized in that: The step of determining the loan amount, interest rate and term based on the risk assessment score comprises: Obtaining the benchmark loan amount in the output result of the model; Obtain benchmark interest rates and benchmark terms from the internal information systems of financial institutions; Based on the benchmark loan amount, the benchmark interest rate and the benchmark term, the loan amount, interest rate and term are determined according to the risk assessment score, wherein the loan amount and term are directly proportional to the risk assessment score, and the interest rate is inversely proportional to the risk assessment score.

9. A credit risk control device based on multi-source data fusion and reinforcement learning, characterized in that: include: An acquisition unit, configured to respond to a loan application initiated by a borrower, acquire a credit data set of the borrower from multiple data sources, and pre-process the credit data set, wherein the credit data set includes at least: basic customer information, financial statements, credit records, and transaction flow; A fusion unit, used for determining a weight vector of each credit data in the preprocessed credit data set by adopting a hierarchical analysis strategy, and fusing all the credit data based on the weight vector to obtain a target feature vector; An input unit, used to input the target feature vector into a pre-built risk assessment model to obtain a model output result, wherein the model output result at least includes: a risk assessment score; The decision-making unit is used to approve the borrower's loan application when the capital return rate of the target financial capital applied for by the borrower is greater than a preset return rate threshold, and determine the loan amount, interest rate and term based on the risk assessment score.

10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the credit risk control method based on multi-source data fusion and reinforcement learning as described in any one of claims 1 to 7.

Images