Loan quality assessment method, device, electronic equipment, medium and product

Abstract

The present application provides a loan quality assessment method, device, electronic device, medium and product, which relate to the field of artificial intelligence. The method includes: obtaining time series data and non-time series data of the loan enterprise; wherein, the time series data includes the operating data of the enterprise and the industry data of the industry in which the enterprise is located, and the non-time series data includes the current static data of the enterprise; inputting the time series data into a pre-trained first assessment model to obtain a first score output by the first assessment model; inputting the non-time series data into a pre-trained second assessment model to obtain a second score output by the second assessment model; wherein, different enterprises have different corresponding first assessment models; and evaluating the loan quality of the enterprise based on the first score and the second score. The method of the present application flexibly and accurately evaluates the quality of enterprise loans.

Description

Technical Field

[0001] The present application relates to artificial intelligence technology, and in particular to a loan quality assessment method, device, electronic device, medium and product. Background Art

[0002] The quality of corporate loans is a crucial component of financial security. It not only impacts the safety of banks' assets but also the stability of financial markets and the healthy development of the economy. Corporate loan quality assessment is a crucial measure for banks to ensure loan quality and asset security. Banks need to comprehensively assess the quality and risk level of corporate loans to formulate sound credit policies and management measures to ensure stable bank operations and healthy economic development.

[0003] In existing technologies, loan quality assessments are mostly based on financial theories and use fixed calculation methods to calculate corporate loan-related data, resulting in limited accuracy of the assessment results. Summary of the Invention

[0004] The present application provides a loan quality assessment method, device, electronic equipment, medium and product for flexibly and accurately assessing the quality of corporate loans.

[0005] In one aspect, the present application provides a loan quality assessment method, comprising:

[0006] Obtaining time series data and non-time series data of the loan enterprise; wherein the time series data includes the business data of the enterprise and the industry data of the industry in which the enterprise is located, and the non-time series data includes the current static data of the enterprise;

[0007] Input the time series data into a pre-trained first evaluation model to obtain a first score output by the first evaluation model; input the non-time series data into a pre-trained second evaluation model to obtain a second score output by the second evaluation model; wherein different companies correspond to different first evaluation models;

[0008] The loan quality of the enterprise is evaluated based on the first score and the second score.

[0009] In another aspect, the present application provides a loan quality assessment device, comprising:

[0010] An acquisition module, configured to acquire time series data and non-time series data of a loan enterprise; wherein the time series data includes the business data of the enterprise and the industry data of the industry in which the enterprise is located, and the non-time series data includes the current static data of the enterprise;

[0011] A prediction module is configured to input the time series data into a pre-trained first evaluation model to obtain a first score output by the first evaluation model; and input the non-time series data into a pre-trained second evaluation model to obtain a second score output by the second evaluation model; wherein different first evaluation models correspond to different enterprises;

[0012] An evaluation module is used to evaluate the loan quality of the enterprise based on the first score and the second score.

[0013] On the other hand, the present application provides an electronic device, comprising: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method as described above.

[0014] On the other hand, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to implement the method as described above when executed by a processor.

[0015] On the other hand, the present application provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0016] In the loan quality assessment method, device, electronic device, medium and product provided in this application, the enterprise's time series data and non-time series data are integrated to assess the enterprise's loan quality, which can more comprehensively and accurately assess the enterprise's loan quality; using a first assessment model, a first score is obtained based on the enterprise's time series data prediction; using a second assessment model, a second score is obtained based on the enterprise's non-time series data prediction, which can accurately extract the characteristics of time series data and non-time series data, deeply grasp the correlation between enterprise loan-related data, and fully and reliably assess the enterprise's loan quality; different first assessment models can be used for different enterprises to achieve personalized model training and evaluation, effectively improving the accuracy and flexibility of loan quality assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0018] Figure 1 hereinafter is a flow chart showing a method for evaluating loan quality according to an embodiment of the present invention;

[0019] Figure 2 exemplarily shows a structural diagram of the first evaluation model provided in an embodiment of the present application;

[0020] Figure 3exemplarily shows a structural diagram of the convolution module 1 provided in an embodiment of the present application;

[0021] Figure 4 exemplarily shows a structural diagram of the second evaluation model provided in an embodiment of the present application;

[0022] Figure 5 Schematic diagram of a scenario of loan quality assessment provided by an embodiment of the present application is exemplarily shown in FIG.

[0023] Figure 6 Schematic diagram of the structure of the loan quality assessment device provided in an embodiment of the present application is shown in FIG.

[0024] Figure 7 Schematic diagram of the structure of the electronic device for loan quality assessment provided in an embodiment of the present application is shown in FIG.

[0025] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0026] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0027] A module in this application refers to a functional module or a logical module. It can be in software form, where a processor executes program code to implement its functionality, or it can be in hardware form. "And / or" describes the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. The character " / " generally indicates that the associated objects are in an "or" relationship.

[0028] The quality of corporate loans is a crucial component of financial security. It not only impacts the safety of banks' assets but also the stability of financial markets and the healthy development of the economy. Corporate loan quality assessment is a crucial measure for banks to ensure loan quality and asset security. Banks need to comprehensively assess the quality and risk level of corporate loans to formulate sound credit policies and management measures to ensure stable bank operations and healthy economic development.

[0029] In existing technologies, loan quality assessments are mostly based on financial theories and use fixed calculation methods to calculate corporate loan-related data, resulting in limited accuracy of the assessment results.

[0030] The following specific embodiments are used to illustrate the technical solution of the present application. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0031] Figure 1 This is a flow chart of the loan quality assessment method provided in the embodiment of this application. Figure 1 As shown, the loan quality assessment method provided in this embodiment may include:

[0032] S101. Obtain time series data and non-time series data of the loan enterprise; wherein the time series data includes the enterprise's operating data and industry data of the enterprise's industry, and the non-time series data includes the enterprise's current static data;

[0033] S102. Input the time series data into a pre-trained first evaluation model to obtain a first score output by the first evaluation model; input the non-time series data into a pre-trained second evaluation model to obtain a second score output by the second evaluation model; wherein different enterprises correspond to different first evaluation models; wherein different enterprises correspond to different first evaluation models;

[0034] S103. Evaluate the loan quality of the enterprise based on the first score and the second score.

[0035] In practical applications, the executing entity of this embodiment can be a loan quality assessment device, which can be implemented through a computer program, such as application software, etc.; or, it can also be implemented as a medium storing relevant computer programs, such as a USB flash drive, a cloud disk, etc.; or, it can also be implemented through a physical device integrated or installed with relevant computer programs, such as a chip, a server, etc.

[0036] In a specific implementation, the time series data and non-time series data of the loan enterprise can be obtained; among them, the time series data includes the enterprise's operating data and the industry data of the industry in which the enterprise is located, and the non-time series data includes the enterprise's current static data.

[0037] Illustratively, operating data may include at least one of stock price data, cash flow data, and loan data. For example, stock price data may include, but is not limited to, a company's market capitalization, volume ratio, commission ratio, and total listing time at different times. Cash flow data may include, but is not limited to, the company's transaction amount, transaction type (transfer, income, purchase, etc.), balance, and time interval since the last transaction at different times. Loan data may include current loan data and total loan data. Current loan data may include, but is not limited to, the remaining repayment amount of the current loan, the change amount of the current loan, the time of the change, the reason for the change (repayment, new loan, etc.), and the time interval since the last change. Total loan data may include, but is not limited to, the total amount of all loans, the change amount of all loans, the time of all loan changes, the reason for all loan changes, and the time interval since the last change of all loans. These data are easily accessible, can effectively reflect the company's operating conditions, and are closely related to the loan, thus serving as a basis for judging loan quality.

[0038] For example, industry data includes industry loan data. For example, industry loan data may include, but is not limited to, loan interest rates, time periods, differences from previous periods, and time since previous periods for the industry at different times. This data is easily accessible and can reflect the market trends of the industry, thus serving as a basis for judging loan quality.

[0039] For example, static data may include at least one of basic enterprise information, credit data, and risk data. For example, basic enterprise information may include, but is not limited to, total market capitalization, years of establishment, and current financial statements; risk data may include, but is not limited to, misappropriation counts, serious violations, legal proceedings, and abnormal operating events; credit data may include, but is not limited to, current credit rating, legal entity credit rating, and collateral value; static data may also include loan application information, such as loan purpose and collateral value. Based on this static data, an accurate enterprise profile can be formed, thereby assessing loan quality.

[0040] Inputting time-series data into the first evaluation model effectively extracts its temporal characteristics and accurately predicts the first score. Inputting non-time-series data into the second evaluation model effectively predicts the second score based on the non-time-series data characteristics. Combining the first and second scores allows for an accurate and comprehensive assessment of corporate loan quality. Due to the large volume and complexity of time-series data, different first evaluation models can be used for different companies, enabling personalized assessments and accurate first scores, effectively improving the flexibility and reliability of the assessment. For non-time-series data, a universal second evaluation model can be used, effectively improving assessment efficiency and conserving resources.

[0041] In the embodiment of the present application, the enterprise's loan quality is evaluated by integrating the enterprise's time series data and non-time series data, which can more comprehensively and accurately evaluate the enterprise's loan quality; using a first evaluation model, a first score is obtained based on the enterprise's time series data prediction; using a second evaluation model, a second score is obtained based on the enterprise's non-time series data prediction, which can accurately extract the characteristics of time series data and non-time series data, deeply grasp the correlation between enterprise loan-related data, and fully and reliably evaluate the enterprise's loan quality; different first evaluation models can be used for different enterprises to achieve personalized model training and evaluation, effectively improving the accuracy and flexibility of loan quality evaluation.

[0042] In one possible implementation, the time series data includes multiple types;

[0043] Input the time series data into the pre-trained first evaluation model to obtain the first score output by the first evaluation model, including:

[0044] Through the first evaluation model, based on the attention mechanism, feature extraction is performed on each type of input time series data to obtain the time series features of each type of time series data;

[0045] According to the time series characteristics of each type of time series data, a first score is predicted.

[0046] In the specific implementation, since there is a certain degree of self-correlation within each type of time series data, the attention mechanism can be used to first extract the time series features of each type of time series data, and then the first score is predicted and output based on the time series features of each type of time series data, avoiding directly using global time series data for time series prediction while ignoring the field of view, which can effectively improve the accuracy and reliability of the prediction.

[0047] For example, the first evaluation model may include a convolutional module and a Transformer module. Transformer is a classic model for natural language processing (NLP) proposed by the Google team in 2017. It uses a self-attention mechanism and does not adopt the sequential structure of a recurrent neural network (RNN). This allows the model to be trained in parallel and has global information, but ignores the details of local features.

[0048] Figure 2 This is a schematic diagram of the structure of the first evaluation model provided in the embodiment of the present application. Figure 2As shown, taking the time series data including stock price data, flow data, current loan data, total loan data and industry loan data as an example, the first evaluation model includes multiple feature mapping modules, multiple convolution modules 1, multiple TCAN (Temporal Convolutional Attention-based The network consists of a temporal convolutional neural network (TCAN) module based on an attention mechanism, a convolution module 2, a Transformer predictor, and a fully connected layer. After inputting the time series data into the first evaluation model, multiple feature mapping modules convert each type of time series data into features. Multiple convolution modules 1 perform convolution processing on the features of each type of time series data. Multiple TCAN modules perform time series prediction convolution on the features of each type of time series data after convolution processing. The time series features of each type of time series data are extracted through the attention mechanism, and each type of sequence data is independently embedded into the features, which expands the local receptive field and effectively reflects the self-correlation within different types of time series data. This can avoid ignoring the local field of view due to excessive globalization when directly using the Transformer predictor for time series prediction. The features output by the TCAN module are input into the convolution module 2. Due to the large amount of time series data, the data features are reduced in dimension by the convolution module 2. The Transformer predictor models global information based on the self-attention mechanism, captures the relationship between cross-type data, and better performs feature processing. The fully connected layer integrates the features output by the Transformer predictor and finally outputs the first predicted score. Among them, the convolution module 1 may include multiple convolution units, each convolution unit includes two convolution (Conv) layers and a ReLU layer with a jump connection; the convolution module 2 may include multiple convolution layers and a concatenation (Concate) layer. Figure 3 This is a schematic diagram of the structure of the convolution module 1 provided in the embodiment of the present application. Figure 3 As shown, the convolution module 1 may include three convolution units. By combining the transformer and the convolutional neural network, local and global information are well combined, the relevance of the data can be more comprehensively obtained, and the first score can be accurately predicted based on the time series data.

[0049] Figure 4 This is a schematic diagram of the structure of the second evaluation model provided in the embodiment of the present application. Figure 4As shown, the second evaluation model includes a feature mapping module, a convolution module 1, a Transformer predictor, and a fully connected layer. Since the amount of non-time series data is small and has no time series features, after the non-time series data is input into the second evaluation model, the feature mapping module converts the global non-time series data mapping into features, the convolution module 1 performs convolution processing on the features of the non-time series data, and the Transformer predictor models the global information based on the self-attention mechanism; the fully connected layer integrates the features output by the Transformer predictor and finally outputs the predicted second score. Among them, the structure of the feature mapping module, convolution module 1, Transformer predictor and fully connected layer in the second evaluation model can be the same as the feature mapping module, convolution module 1, Transformer predictor and fully connected layer in the first evaluation model, and their specific parameters can be different.

[0050] In one possible implementation, the method further includes:

[0051] Obtain time series training data and non-time series training data, where the time series training data includes multiple sets of historical time series data and corresponding first scores, and the non-time series training data includes multiple sets of historical non-time series data and corresponding second scores;

[0052] Obtaining an initial first evaluation model and a second evaluation model;

[0053] The initial first evaluation model / second evaluation model is trained according to the time series training data / non-time series training data until the loss function converges, thereby obtaining a trained first evaluation model / second evaluation model.

[0054] In a specific implementation, both time-series training data and non-time-series training data can be obtained. The time-series training data includes multiple sets of historical time-series data and corresponding first scores, while the non-time-series training data includes multiple sets of historical non-time-series data and corresponding second scores. The first and second scores can be manually assigned or determined according to preset rules. For example, if the company's loan repayments are on time and without delay during a certain period, the first score is full. If there is an overdue payment, but the cash flow meets the current requirements, the first score is 80 points. If the cash flow is insufficient, but the balance meets the total deduction requirements, the first score is 40 points. Alternatively, a gradient deduction can be implemented during the overdue period, based on the cash flow percentage within the gradient. For example, if the cash flow meets half of the monthly requirements but the asset value meets the total requirements, the first score is 60 points. If there are signs of bad debt, such as no cash flow in a certain period and insolvency, the score can be adjusted, ranging from 0 (completely undeductible) to 40 (deductible). In actual applications, different scoring methods can be selected based on production needs and are not limited here.

[0055] After establishing the initial first evaluation model and the second evaluation model, the initial first evaluation model and the second evaluation model are trained respectively using time series training data and non-time series training data until the loss function converges. The training is then stopped, and the current first evaluation model / second evaluation model is used as the trained first evaluation model / second evaluation model.

[0056] For example, the loss function may include but is not limited to a mean square error (MSE) loss function, etc.

[0057] It should be noted that the first evaluation models corresponding to different enterprises are different. The corresponding first evaluation models can be trained separately using the time series training data of each enterprise, and personalized evaluation can be carried out for specific enterprises, effectively improving the accuracy of the first score obtained by using time series data prediction; the second evaluation model is universal, and uses unified standards to evaluate enterprise portraits to obtain the second score, which can effectively improve the evaluation efficiency.

[0058] In a possible implementation, the loan quality of the enterprise is evaluated based on the first score and the second score, including:

[0059] The loan quality of the enterprise is evaluated based on whether the first score and the second score are higher than the preset threshold, as well as the changing trend of the first score / second score at different times.

[0060] Figure 5 Schematic diagram of the scenario of loan quality assessment provided in the embodiment of this application. Figure 5 As shown, in a specific implementation, the first score output by the first evaluation model and the second score output by the second evaluation model can both be input into the business module. The business module can perform corresponding business processing according to preset rules. For example, assuming that the first score / second score is higher than the first threshold, the loan quality is judged to be high, and the enterprise can consider rapid approval and loan disbursement when applying for a new loan subsequently; assuming that the first score / second score is lower than the second threshold, the loan quality is judged to be low, and lending can be tightened, review can be strengthened, communication can be carried out to urge repayment, and the situation can be understood; if the first score / second score shows a continuous upward or downward trend for multiple times, further analysis can be carried out to evaluate the loan quality from multiple angles.

[0061] Figure 6 This is a schematic diagram of the structure of the loan quality assessment device provided in the embodiment of the present application. Figure 6 As shown, the loan quality assessment device 600 provided in this embodiment may include:

[0062] Acquisition module 61 is used to acquire time series data and non-time series data of the loan enterprise; wherein the time series data includes the business data of the enterprise and the industry data of the industry in which the enterprise is located, and the non-time series data includes the current static data of the enterprise;

[0063] Prediction module 62 is configured to input the time series data into a pre-trained first evaluation model to obtain a first score output by the first evaluation model; and input the non-time series data into a pre-trained second evaluation model to obtain a second score output by the second evaluation model; wherein different first evaluation models correspond to different enterprises;

[0064] The evaluation module 63 is configured to evaluate the loan quality of the enterprise based on the first score and the second score.

[0065] In one possible implementation, the time series data includes multiple types; the prediction module 62 is configured to:

[0066] By using the first evaluation model, extracting features of each type of input time series data based on the attention mechanism to obtain time series features of each type of time series data;

[0067] The first score is predicted based on the time series features of the time series data of each type.

[0068] In one possible implementation, the prediction module 62 is further configured to

[0069] Acquire time series training data and non-time series training data, wherein the time series training data includes multiple groups of historical time series data and corresponding first scores, and the non-time series training data includes multiple groups of historical non-time series data and corresponding second scores;

[0070] Obtaining an initial first evaluation model and a second evaluation model;

[0071] The initial first evaluation model / second evaluation model is trained according to the time series training data / non-time series training data until the loss function converges, thereby obtaining a trained first evaluation model / second evaluation model.

[0072] In a possible implementation, the evaluation module 63 is specifically configured to:

[0073] The loan quality of the enterprise is evaluated based on whether the first score and the second score are higher than a preset threshold and the changing trend of the first score / second score at different times.

[0074] In a possible implementation, the operating data includes at least one of stock price data, cash flow data, and loan data, and the industry data includes industry loan interest rate data.

[0075] In a possible implementation, the static data includes at least one of basic enterprise information, credit data, and risk data.

[0076] In actual applications, the loan quality assessment device can be implemented through a computer program, such as application software, etc.; or, it can be implemented as a medium storing relevant computer programs, such as a USB flash drive, a cloud disk, etc.; or, it can be implemented through a physical device integrated or installed with relevant computer programs, such as a chip, a server, etc.

[0077] It should be noted that the loan evaluation device is used to execute the loan evaluation method as described above, and its specific implementation method is as described above and will not be repeated here.

[0078] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application, such as Figure 7 As shown, the electronic device includes:

[0079] The electronic device includes a processor 291 and a memory 292. It may also include a communication interface 293 and a bus 294. The processor 291, memory 292, and communication interface 293 can communicate with each other via bus 294. Communication interface 293 can be used for information transmission. The processor 291 can invoke logic instructions in memory 292 to execute the methods of the above embodiments.

[0080] In addition, the logic instructions in the memory 292 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product.

[0081] Memory 292, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as program instructions / modules corresponding to the methods in the embodiments of the present application. Processor 291 executes the software programs, instructions, and modules stored in memory 292 to perform functional applications and data processing, thereby implementing the methods in the above-mentioned method embodiments.

[0082] Memory 292 may include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function; the data storage area may store data generated based on the use of the terminal device. Memory 292 may also include high-speed random access memory and non-volatile memory.

[0083] An embodiment of the present application provides a non-transitory computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are executed by a processor, they are used to implement the method described in the above embodiment.

[0084] An embodiment of the present application provides a computer program product, including a computer program. When the computer program is executed by a processor, the method provided in any of the above embodiments of the present application is implemented.

[0085] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all optional embodiments, and the actions and modules involved are not necessarily required by this application.

[0086] It should be further noted that, although the various steps in the flowchart are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps may be performed in other orders. Moreover, at least a portion of the steps in the flowchart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily performed at the same time, but may be performed at different times. The execution order of these sub-steps or stages is not necessarily to be performed in sequence, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0087] It should be understood that the above-described device embodiments are merely illustrative, and the device of the present application may also be implemented in other ways. For example, the division of units / modules in the above-described embodiments is merely a logical functional division, and actual implementations may employ other division methods. For example, multiple units, modules, or components may be combined or integrated into another system, or some features may be omitted or not implemented.

[0088] In addition, unless otherwise specified, the functional units / modules in the various embodiments of the present application may be integrated into a single unit / module, each unit / module may exist physically separately, or two or more units / modules may be integrated together. The aforementioned integrated units / modules may be implemented in the form of hardware or software program modules.

[0089] If an integrated unit / module is implemented in hardware, the hardware may be digital circuits, analog circuits, etc. The physical implementation of the hardware structure includes, but is not limited to, transistors, memristors, etc. Unless otherwise specified, the processor may be any appropriate hardware processor, such as a CPU, GPU, FPGA, DSP, and ASIC. Unless otherwise specified, the storage unit may be any appropriate magnetic storage medium or magneto-optical storage medium, such as resistive random access memory (RRAM), dynamic random access memory (DRAM), static random access memory (SRAM), enhanced dynamic random access memory (EDRAM), high-bandwidth memory (HBM), hybrid memory cube (HMC), etc.

[0090] If the integrated unit / module is implemented in the form of a software program module and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, 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. The computer software product is stored in a memory and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, magnetic disk, or optical disk, etc., various media that can store program code.

[0091] In the above embodiments, the description of each embodiment has its own emphasis. For parts not described in detail in a particular embodiment, please refer to the relevant description of other embodiments. The technical features of the above embodiments can be combined in any way. To keep the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0092] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0093] It should be understood that the present application is not limited to the exact structure described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A loan quality assessment method, characterized in that: include: Obtaining time series data and non-time series data of the loan enterprise; wherein the time series data includes the business data of the enterprise and the industry data of the industry in which the enterprise is located, and the non-time series data includes the current static data of the enterprise; Input the time series data into a pre-trained first evaluation model to obtain a first score output by the first evaluation model; input the non-time series data into a pre-trained second evaluation model to obtain a second score output by the second evaluation model; wherein different companies correspond to different first evaluation models; evaluating the loan quality of the enterprise based on the first score and the second score; The time series data includes multiple types; inputting the time series data into a pre-trained first evaluation model to obtain a first score output by the first evaluation model includes: By using the first evaluation model, extracting features of each type of input time series data based on the attention mechanism to obtain time series features of each type of time series data; Predicting the first score according to the time series features of the time series data of each type; The first evaluation model includes a convolution module and a Transformer module.

2. The method according to claim 1, characterized in that The method further comprises: Acquire time series training data and non-time series training data, wherein the time series training data includes multiple groups of historical time series data and corresponding first scores, and the non-time series training data includes multiple groups of historical non-time series data and corresponding second scores; Obtaining an initial first evaluation model and a second evaluation model; The initial first evaluation model / second evaluation model is trained according to the time series training data / non-time series training data until the loss function converges, thereby obtaining a trained first evaluation model / second evaluation model.

3. The method according to claim 1, characterized in that The evaluating the loan quality of the enterprise according to the first score and the second score includes: The loan quality of the enterprise is evaluated based on whether the first score and the second score are higher than a preset threshold and the changing trend of the first score / second score at different times.

4. The method according to claim 1, wherein The operating data includes at least one of stock price data, cash flow data and loan data, and the industry data includes industry loan interest rate data.

5. The method according to any one of claims 1 to 4, characterized in that The static data includes at least one of basic enterprise information, credit data and risk data.

6. A loan quality assessment device, characterized in that: include: An acquisition module, configured to acquire time series data and non-time series data of a loan enterprise; wherein the time series data includes the business data of the enterprise and the industry data of the industry in which the enterprise is located, and the non-time series data includes the current static data of the enterprise; A prediction module is configured to input the time series data into a pre-trained first evaluation model to obtain a first score output by the first evaluation model; and input the non-time series data into a pre-trained second evaluation model to obtain a second score output by the second evaluation model; wherein different first evaluation models correspond to different enterprises; an evaluation module, configured to evaluate the loan quality of the enterprise based on the first score and the second score; The time series data includes multiple types; the prediction module is specifically used to: By using the first evaluation model, extracting features of each type of input time series data based on the attention mechanism to obtain time series features of each type of time series data; Predicting the first score according to the time series features of the time series data of each type; The first evaluation model includes a convolution module and a Transformer module.

7. An electronic device comprising: a processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, wherein the computer-executable instructions are used to implement the method according to any one of claims 1 to 5 when executed by a processor.

9. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images