Performance risk early warning method and system based on user data

Abstract

The invention discloses a user data-based performance risk early warning method and system, and the method comprises the steps: collecting the credit behavior change data of a to-be-detected user in a specific time period, mixing the credit behavior change data with a risk credit behavior change data set containing the information of a known default user, forming a mixed data set, processing the mixed data set through a clustering analysis technology, and carrying out the early warning of the performance risk. Obtaining risk classification data, screening out potential default users according to the risk classification data, inputting historical credit data of the potential default users into a performance risk value prediction algorithm, calculating a performance risk value, when the performance risk value exceeds a preset threshold value, further checking whether the collateral price change data meets a set condition, and if not, judging whether the collateral price change data meets the set condition. And if yes, determining that the user is a high-risk user. According to the invention, the credit behavior of the user and the collateral price change data are integrated, and the clustering analysis and risk prediction algorithm is used, so that the performance risk is efficiently and accurately evaluated, and the risk management capability of a financial institution is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of contract performance risk warning, and in particular to a contract performance risk warning method and system based on user data. Background Art

[0002] In the existing performance risk assessment field, financial institutions and other credit service providers are facing increasing data processing challenges. Traditional performance risk assessment methods usually require a separate assessment of each user's credit history and behavior patterns, which is not only computationally intensive but also inefficient. With the development of Internet finance, user behavior data is growing exponentially. How to quickly and accurately identify potential defaulting users from massive data has become an urgent problem to be solved in the industry.

[0003] Current methods often rely on manual review or simple statistical analysis, which may lead to subjective judgment bias in the evaluation process and make it difficult to reflect the user's latest credit status in real time. Summary of the invention

[0004] In order to solve at least one of the technical problems mentioned above, the present invention provides a method and system for early warning of contract performance risks based on user data.

[0005] In a first aspect, the present invention provides a contract performance risk early warning method based on user data, the method comprising:

[0006] Obtaining credit behavior change data and collateral price change data for different users to be detected within a preset time period, wherein the credit behavior change data includes change information on borrowing frequency, change information on the number of lending institutions, and change information on repayment methods;

[0007] Mixing a plurality of credit behavior change data with a preset risk credit behavior change data set to obtain a mixed data set, wherein the risk credit behavior change data set includes credit behavior change data of a plurality of known defaulting users within a preset time period;

[0008] Perform clustering on the mixed data set to obtain risk classification data, and screen out potential defaulting users based on the risk classification data;

[0009] Obtain the historical credit data of potential defaulting users, input the historical credit data and credit behavior change data into the preset performance risk value prediction algorithm, and output the performance risk value;

[0010] If the performance risk value is greater than the preset risk threshold, it is determined whether the collateral price change data of the potential defaulting user meets the preset conditions. If not, the performance risk of the potential defaulting user is confirmed to be high risk.

[0011] Preferably, the performance risk value prediction algorithm is expressed as:

[0012] V k =V history +V0,

[0013]

[0014] V0=ln(1+e F+L+M ),

[0015] Where V k represents the performance risk value of user k, V history represents the performance impact item caused by historical credit data, V0 represents the performance impact item caused by credit behavior change data, M k represents the total number of historical credit data of user k, t i Indicates the time distance between the i-th historical credit data and the current time, represents the weight of the i-th historical credit data, A i is the loan amount of the i-th historical credit data, ΔA i is the difference between the loan amount of the i-th historical credit data and the current loan amount, x i represents the overdue status of the i-th historical credit data, and x i =0 means not overdue, x i =1 means overdue, h i represents the number of days overdue for the i-th historical credit data, H i It represents the total number of days of the loan term of the i-th historical credit data. F, L and M are the normalized change information of borrowing frequency, change information of the number of lending institutions and change information of repayment method, respectively. Among them, M=0 represents automatic repayment and M=1 represents manual repayment.

[0016] Preferably, clustering the mixed data set to obtain risk classification data, and screening out potential defaulting users based on the risk classification data, includes:

[0017] Determine the number of classification clusters based on the number of risk types in the risk credit data set;

[0018] Based on the number of classification clusters, the K-means algorithm is used to cluster the mixed data set to obtain risk classification data;

[0019] According to the distribution of the credit data of the user to be detected in the risk classification data, it is determined whether the user to be detected is a potential default user.

[0020] In a second aspect, the present invention further provides a contract performance risk early warning system based on user data, the system comprising:

[0021] A data acquisition module is used to acquire credit behavior change data and collateral price change data of different users to be detected within a preset time period, wherein the credit behavior change data includes change information of borrowing frequency, change information of the number of lending institutions and change information of repayment method;

[0022] A risk data mixing module, used to mix a plurality of credit behavior change data with a preset risk credit behavior change data set to obtain a mixed data set, wherein the risk credit behavior change data set includes credit behavior change data of a plurality of known defaulting users within a preset time period;

[0023] A potential default user screening module is used to perform clustering processing on the mixed data set to obtain risk classification data, and screen out potential default users based on the risk classification data;

[0024] The risk value calculation module for potential default users is used to obtain the historical credit data of potential default users, input the historical credit data and credit behavior change data into the preset performance risk value prediction algorithm, and output the performance risk value;

[0025] The risk prediction module for potential defaulting users is used to determine whether the collateral price change data of the potential defaulting user meets the preset conditions if the performance risk value is greater than the preset risk threshold. If not, the performance risk of the potential defaulting user is confirmed to be high risk.

[0026] Preferably, the performance risk value prediction algorithm is expressed as:

[0027] V k =V history +V0,

[0028]

[0029] V0=ln(1+e F+L+M ),

[0030] Where V k represents the performance risk value of user k, V ntstory represents the performance impact item caused by historical credit data, V0 represents the performance impact item caused by credit behavior change data, M k represents the total number of historical credit data of user k, t i Indicates the time distance between the i-th historical credit data and the current time, represents the weight of the i-th historical credit data, A i is the loan amount of the i-th historical credit data, ΔA i is the difference between the loan amount of the i-th historical credit data and the current loan amount, x i represents the overdue status of the i-th historical credit data, and x i=0 means not overdue, x i =1 means overdue, h i represents the number of days overdue for the i-th historical credit data, H i It represents the total number of days of the loan term of the i-th historical credit data. F, L and M are the normalized change information of borrowing frequency, change information of the number of lending institutions and change information of repayment method, respectively. Among them, M=0 represents automatic repayment and M=1 represents manual repayment.

[0031] Preferably, clustering the mixed data set to obtain risk classification data, and screening out potential defaulting users based on the risk classification data, includes:

[0032] Determine the number of classification clusters based on the number of risk types in the risk credit data set;

[0033] Based on the number of classification clusters, the K-means algorithm is used to cluster the mixed data set to obtain risk classification data;

[0034] According to the distribution of the credit data of the user to be detected in the risk classification data, it is determined whether the user to be detected is a potential default user.

[0035] In a third aspect, the present invention further provides an electronic device comprising a processor and a memory, wherein the memory is used to store computer program code, and the computer program code comprises computer instructions. When the processor executes the computer instructions, the electronic device executes the method as described in the first aspect above and any possible implementation thereof.

[0036] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, wherein the computer program includes program instructions, and when the program instructions are executed by a processor of an electronic device, the processor executes the method as described in the first aspect above and any possible implementation thereof.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1) The present invention forms a mixed data set by mixing the credit behavior change data of the user to be detected with the data of the known defaulting users, and adopts a clustering algorithm for risk classification, so as to efficiently screen out potential defaulting users, and then combines historical credit data with the performance risk prediction algorithm to accurately evaluate the performance risk value. When the risk value exceeds the preset threshold and the price change of the collateral fails to provide sufficient protection, the high-risk users can be identified in time, thereby significantly improving the efficiency and accuracy of the performance risk warning, and effectively reducing the credit loss risk faced by financial institutions.

[0039] 2) The performance risk value prediction algorithm in the present invention can accurately quantify the user's performance risk level by integrating historical credit data and recent credit behavior change information. It uses weighted average and exponential decay to emphasize the importance of recent credit events, and normalizes the credit behavior change information to ensure that information of different dimensions is compared on the same scale. Furthermore, a logarithmic function is used to increase the sensitivity to small changes in credit behavior change data, thereby achieving accurate prediction of performance risk, taking into account both historical performance and the latest credit behavior dynamics.

[0040] 3) The present invention constructs a mixed data set and combines it with cluster analysis technology, using the K-means algorithm to efficiently and accurately identify user groups with similar risk characteristics, thereby achieving rapid screening of potential defaulting users and greatly improving the efficiency and accuracy of performance risk assessment. At the same time, it dynamically adapts to changes in new risk characteristics, continuously optimizes the risk assessment model, and effectively enhances the risk management capabilities and market stability of financial institutions.

[0041] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the background technology, the drawings required for use in the embodiments of the present invention or the background technology will be described below.

[0043] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and are used to illustrate the technical solutions of the present disclosure together with the specification.

[0044] Figure 1 A schematic diagram of a flow chart of a contract performance risk early warning method based on user data provided by an embodiment of the present invention;

[0045] Figure 2 A schematic diagram of the structure of a contract performance risk warning system based on user data provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0046] 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 are within the scope of protection of the present invention.

[0047] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0048] Traditional performance risk assessment methods usually require a separate assessment of each user's credit history and behavior patterns, which has the problems of huge computational complexity and low efficiency.

[0049] See also Figure 1 , Figure 1 The following is a flow chart of a method for early warning of contract performance risk based on user data provided by an embodiment of the present invention. Figure 1 As shown, the method includes:

[0050] S100, obtaining credit behavior change data and collateral price change data of different users to be detected within a preset time period, wherein the credit behavior change data includes change information of borrowing frequency, change information of the number of lending institutions and change information of repayment method;

[0051] In this embodiment, the credit behavior change data of different users to be detected within a preset time period and the price change data of the collateral they hold are collected from multiple channels. The "credit behavior change data" specifically refers to the change in the number of borrowing times of the user during this period (i.e., the frequency of borrowing information), the number of service institutions that provide loans to the user (the number of lending institutions), and whether the repayment method selected by the user has changed (repayment method change information), while the "collateral price change data" refers to the market value fluctuation of the items or assets provided by the user as loan collateral. The core of this step is the comprehensiveness and timeliness of the data, ensuring that the collected data can accurately reflect the credit status of the user and the value status of its collateral. The data acquisition at this stage is not limited to data in digital form, but also includes any information that can reflect the changes in the user's credit status, such as the number of loan applications, the diversity of lending platforms, whether the repayment method has changed from equal principal to equal principal and interest, automatic repayment or manual repayment, etc. At the same time, the market value fluctuation of the collateral will be tracked and recorded to fully grasp the changing trend of the user's performance ability. It should be noted that collateral refers to the assets provided by users as collateral during the lending process, such as real estate, vehicles, etc., and the price change data reflects the changes in the value of these assets over a certain period of time.

[0052] S200, mixing a plurality of credit behavior change data with a preset risk credit behavior change data set to obtain a mixed data set, wherein the risk credit behavior change data set includes credit behavior change data of a plurality of known defaulting users within a preset time period;

[0053] The multiple credit behavior change data collected above are mixed with the preset risk credit behavior change data set to construct a mixed data set containing a large number of credit behavior change data of known default users within a preset time period. The "risk credit behavior change data set" here refers to a collection of credit behavior data of users who have defaulted that has been accumulated over a long period of time and marked by an expert team. It is used to train the model to identify default patterns. The purpose is to enhance the learning ability of the model by introducing a large number of known cases, and to find the similarities between current user data and historical default user data by comparison, thereby identifying potential default risks.

[0054] S300, clustering the mixed data set to obtain risk classification data, and screening out potential defaulting users based on the risk classification data;

[0055] The constructed mixed data set is clustered, and the intrinsic connection between the data is found through the clustering algorithm in the data mining technology. The data with similar characteristics are classified into one category, and finally the risk classification data is obtained. Then, based on the risk classification data, potential default users who show characteristics similar to those of known default users are screened out. This embodiment uses the unsupervised learning method in machine learning to perform preliminary screening of a large amount of data and discover potential default risks through natural grouping between data.

[0056] S400, obtaining historical credit data of potential defaulting users, inputting historical credit data and credit behavior change data into a preset performance risk value prediction algorithm, and outputting a performance risk value;

[0057] For the potential defaulting users screened out, their historical credit data is further collected, including but not limited to previous repayment records, credit score change trends, etc., and the historical credit data and credit behavior change data are input into the preset performance risk value prediction algorithm to calculate a specific quantitative performance risk value. The performance risk value prediction algorithm uses machine learning algorithms including but not limited to random forests, support vector machines, etc., to estimate the performance risk value of each potential defaulting user through learning from historical data, thereby providing a quantitative default risk assessment result.

[0058] S500, if the performance risk value is greater than the preset risk threshold, determine whether the collateral price change data of the potential defaulting user meets the preset conditions. If not, confirm that the performance risk of the potential defaulting user is high risk.

[0059] When the performance risk value exceeds a pre-set risk threshold, the price change data of the collateral of the potential defaulting user is further checked to see if it meets the preset conditions. If the value of these collaterals is not enough to cover the possible default losses, the performance risk of the user will be finally determined to be high. This determination process not only improves the accuracy of risk assessment, but also provides timely early warning signals for financial institutions, enabling them to take appropriate measures to reduce losses before risks occur.

[0060] In this embodiment, a mixed data set is formed by mixing the credit behavior change data of the user to be detected with the data of known defaulting users, and a clustering algorithm is used for risk classification to efficiently screen out potential defaulting users. Then, the historical credit data and the performance risk prediction algorithm are combined to accurately assess the performance risk value. When the risk value exceeds the preset threshold and the price change of the collateral fails to provide sufficient protection, high-risk users can be identified in a timely manner, thereby significantly improving the efficiency and accuracy of performance risk warning and effectively reducing the credit loss risk faced by financial institutions.

[0061] Preferably, the performance risk value prediction algorithm is expressed as:

[0062] V k =V history +V0,

[0063]

[0064] V0=ln(1+e F+L+M ),

[0065] Where V k represents the performance risk value of user k, V history represents the performance impact item caused by historical credit data, V0 represents the performance impact item caused by credit behavior change data, M k represents the total number of historical credit data of user k, t i Indicates the time distance between the i-th historical credit data and the current time, represents the weight of the i-th historical credit data, A i is the loan amount of the i-th historical credit data, ΔA i is the difference between the loan amount of the i-th historical credit data and the current loan amount, x i represents the overdue status of the i-th historical credit data, and x i =0 means not overdue, x i =1 means overdue, h i represents the number of days overdue for the i-th historical credit data, H iIt represents the total number of days of the loan term of the i-th historical credit data. F, L and M are the normalized change information of borrowing frequency, change information of the number of lending institutions and change information of repayment method, respectively. Among them, M=0 represents automatic repayment and M=1 represents manual repayment.

[0066] Performance impact item V caused by historical credit data history The first term of the middle molecule It is the weighted sum of overdue situations, taking into account the overdue situations and overdue days of all historical credit data, where: The term is used to adjust the importance of historical data. As time goes by, the weight gradually decreases, which means that recent behavior has a higher weight than long-term behavior. If the user has overdue behavior in the historical record, the value will increase, and the more overdue days, the greater the value. This term emphasizes the user's tendency to default. It is a weighted sum of changes in loan amounts, taking into account changes in user debt. If the user's recent loan amount has increased, then this value will increase, indicating that the user's financial pressure has increased and the possibility of default has also increased accordingly. The denominator is M k , is the total number of historical credit data of the user, which plays a normalization role.

[0067] In the performance impact item V0 caused by the credit behavior change data, F represents the normalized change information of the borrowing frequency, which reflects the frequency of the user's borrowing activities over a period of time. The larger the F value, the higher the borrowing frequency; L represents the normalized change information of the number of lending institutions, which reflects the user's borrowing behavior between different financial institutions. The larger the value, the more different the user borrows from; M represents the normalized change information of the repayment method. If the user changes from automatic repayment to manual repayment, or changes the repayment method frequently, it reflects the uncertainty or complexity of the user in repayment. The larger the value, the more changes in the repayment method; F, L and M range between [0,1]. The use of logarithmic function enhances the sensitivity to small changes in credit behavior change data, so that the algorithm can capture and reflect changes in user credit status in a timely manner.

[0068] In this embodiment, the performance risk value prediction algorithm can accurately quantify the user's performance risk level by integrating historical credit data and recent credit behavior change information. It uses weighted average and exponential decay to emphasize the importance of recent credit events, and normalizes the credit behavior change information to ensure that information of different dimensions is compared on the same scale. Furthermore, a logarithmic function is used to increase the sensitivity to small changes in credit behavior change data, thereby achieving accurate prediction of performance risk, taking into account both historical performance and the latest credit behavior dynamics.

[0069] Preferably, clustering the mixed data set to obtain risk classification data, and screening out potential defaulting users based on the risk classification data, includes:

[0070] Determine the number of classification clusters based on the number of risk types in the risk credit data set;

[0071] Based on the number of classification clusters, the K-means algorithm is used to cluster the mixed data set to obtain risk classification data;

[0072] According to the distribution of the credit data of the user to be detected in the risk classification data, it is determined whether the user to be detected is a potential default user.

[0073] In this embodiment, the number of classification clusters in the clustering process is determined based on the number of different types of risk users in the preset risk credit behavior change data set. Usually, the number of classification clusters is equal to the number of risk types plus one (plus the normal performance type). The number of risk types is the number of categories summarized based on the behavior patterns of known default users. Risk types include but are not limited to frequent borrowing, multiple overdue payments, and multi-institutional lending. Each type of risk type corresponds to a possible default behavior pattern. Determining the number of classification clusters is a basic step in cluster analysis, which directly affects the effect and accuracy of clustering. According to the determined number of classification clusters, the K-means algorithm is used to cluster the mixed data set. It should be noted that K-means is a commonly used unsupervised learning algorithm that divides the data set into K clusters through an iterative process, so that the data points in the cluster are as similar as possible to each other, and the data points between clusters are as different as possible. The K-means algorithm assigns the data points in the mixed data set to each classification cluster to generate risk classification data. The risk classification data can reflect the similarity relationship between data points and reveal potential risk user groups. Furthermore, based on the distribution of the credit data of the users to be detected in the risk classification data, it is possible to further determine which users are potential default users. Specifically, if a credit data point of a user to be detected is located in a classification cluster composed of known default users, then the user is considered to have a higher risk of default and is thus marked as a potential default user.

[0074] In this embodiment, a mixed data set is constructed and combined with cluster analysis technology, and the K-means algorithm is used to efficiently and accurately identify user groups with similar risk characteristics, thereby achieving rapid screening of potential default users, greatly improving the efficiency and accuracy of performance risk assessment, and dynamically adapting to changes in new risk characteristics, continuously optimizing risk assessment models, and effectively enhancing the risk management capabilities and market stability of financial institutions.

[0075] In summary, the method provided in this embodiment can at least achieve the following effects:

[0076] 1) The present invention forms a mixed data set by mixing the credit behavior change data of the user to be detected with the data of the known defaulting users, and adopts a clustering algorithm for risk classification, so as to efficiently screen out potential defaulting users, and then combines historical credit data with the performance risk prediction algorithm to accurately evaluate the performance risk value. When the risk value exceeds the preset threshold and the price change of the collateral fails to provide sufficient protection, the high-risk users can be identified in time, thereby significantly improving the efficiency and accuracy of the performance risk warning, and effectively reducing the credit loss risk faced by financial institutions.

[0077] 2) The performance risk value prediction algorithm in the present invention can accurately quantify the user's performance risk level by integrating historical credit data and recent credit behavior change information. It uses weighted average and exponential decay to emphasize the importance of recent credit events, and normalizes the credit behavior change information to ensure that information of different dimensions is compared on the same scale. Furthermore, a logarithmic function is used to increase the sensitivity to small changes in credit behavior change data, thereby achieving accurate prediction of performance risk, taking into account both historical performance and the latest credit behavior dynamics.

[0078] 3) The present invention constructs a mixed data set and combines it with cluster analysis technology, using the K-means algorithm to efficiently and accurately identify user groups with similar risk characteristics, thereby achieving rapid screening of potential defaulting users and greatly improving the efficiency and accuracy of performance risk assessment. At the same time, it dynamically adapts to changes in new risk characteristics, continuously optimizes the risk assessment model, and effectively enhances the risk management capabilities and market stability of financial institutions.

[0079] See also Figure 2 In one embodiment, a contract performance risk early warning system based on user data is also provided, the system comprising:

[0080] The data acquisition module 100 is used to acquire the credit behavior change data and collateral price change data of different users to be detected within a preset time period, wherein the credit behavior change data includes the change information of borrowing frequency, the change information of the number of lending institutions and the change information of repayment method;

[0081] A risk data mixing module 200 is used to mix a plurality of credit behavior change data with a preset risk credit behavior change data set to obtain a mixed data set, wherein the risk credit behavior change data set includes credit behavior change data of a plurality of known defaulting users within a preset time period;

[0082] The potential default user screening module 300 is used to perform clustering processing on the mixed data set to obtain risk classification data, and screen out potential default users based on the risk classification data;

[0083] The potential default user risk value calculation module 400 is used to obtain the historical credit data of the potential default user, input the historical credit data and credit behavior change data into a preset performance risk value prediction algorithm, and output the performance risk value;

[0084] The potential default user risk prediction module 500 is used to determine whether the collateral price change data of the potential default user meets the preset conditions if the performance risk value is greater than the preset risk threshold. If not, the performance risk of the potential default user is confirmed to be high risk.

[0085] Preferably, the performance risk value prediction algorithm is expressed as:

[0086] V k =V history +V0,

[0087]

[0088] V0=ln(1+e F+L+M ),

[0089] Where V k represents the performance risk value of user k, V history represents the performance impact item caused by historical credit data, V0 represents the performance impact item caused by credit behavior change data, M k represents the total number of historical credit data of user k, t i Indicates the time distance between the i-th historical credit data and the current time, represents the weight of the i-th historical credit data, A i is the loan amount of the i-th historical credit data, ΔA i is the difference between the loan amount of the i-th historical credit data and the current loan amount, x i represents the overdue status of the i-th historical credit data, and x i =0 means not overdue, x i =1 means overdue, h i represents the number of days overdue for the i-th historical credit data, H i It represents the total number of days of the loan term of the i-th historical credit data. F, L and M are the normalized change information of borrowing frequency, change information of the number of lending institutions and change information of repayment method, respectively. Among them, M=0 represents automatic repayment and M=1 represents manual repayment.

[0090] Preferably, clustering the mixed data set to obtain risk classification data, and screening out potential defaulting users based on the risk classification data, includes:

[0091] Determine the number of classification clusters based on the number of risk types in the risk credit data set;

[0092] Based on the number of classification clusters, the K-means algorithm is used to cluster the mixed data set to obtain risk classification data;

[0093] According to the distribution of the credit data of the user to be detected in the risk classification data, it is determined whether the user to be detected is a potential default user.

[0094] It can be understood that the functions or modules included in the system provided in this embodiment can be used to execute the method described in the above method embodiment. Its specific implementation can refer to the description of the above method embodiment. For the sake of brevity, it will not be repeated here.

[0095] The present invention also provides an electronic device, including a processor and a memory, wherein the memory is used to store computer program code, and the computer program code includes computer instructions. When the processor executes the computer instructions, the electronic device executes a method as described in any possible implementation manner.

[0096] The present invention also provides a computer-readable storage medium, in which a computer program is stored. The computer program includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes a method as described in any possible implementation manner.

[0097] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0098] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. Those skilled in the art can also clearly understand that the descriptions of the various embodiments of the present invention have different focuses. For the convenience and brevity of description, the same or similar parts may not be repeated in different embodiments. Therefore, for parts not described or not described in detail in a certain embodiment, refer to the records of other embodiments.

Claims

1. A contract performance risk early warning method based on user data, characterized in that: The method comprises: Obtaining credit behavior change data and collateral price change data for different users to be detected within a preset time period, wherein the credit behavior change data includes change information on borrowing frequency, change information on the number of lending institutions, and change information on repayment methods; Mixing a plurality of credit behavior change data with a preset risk credit behavior change data set to obtain a mixed data set, wherein the risk credit behavior change data set includes credit behavior change data of a plurality of known defaulting users within a preset time period; Perform clustering on the mixed data set to obtain risk classification data, and screen out potential defaulting users based on the risk classification data; Obtain the historical credit data of potential defaulting users, input the historical credit data and credit behavior change data into the preset performance risk value prediction algorithm, and output the performance risk value; If the performance risk value is greater than the preset risk threshold, it is determined whether the collateral price change data of the potential defaulting user meets the preset conditions. If not, the performance risk of the potential defaulting user is confirmed to be high risk.

2. The contract performance risk early warning method based on user data according to claim 1 is characterized in that: The performance risk value prediction algorithm is expressed as: V k =V history +V0, V0=ln(1+e F+L+M ), Where V k represents the performance risk value of user k, V history represents the performance impact item caused by historical credit data, V0 represents the performance impact item caused by credit behavior change data, M k represents the total number of historical credit data of user k, t i Indicates the time distance between the i-th historical credit data and the current time, represents the weight of the i-th historical credit data, A i is the loan amount of the i-th historical credit data, ΔA i is the difference between the loan amount of the i-th historical credit data and the current loan amount, x i represents the overdue status of the i-th historical credit data, and x i =0 means not overdue, x i =1 means overdue, h i represents the number of days overdue for the i-th historical credit data, H i It represents the total number of days of the loan term of the i-th historical credit data. F, L and M are the normalized change information of borrowing frequency, change information of the number of lending institutions and change information of repayment method, respectively. Among them, M=0 represents automatic repayment and M=1 represents manual repayment.

3. The contract performance risk early warning method based on user data according to claim 1, characterized in that: The clustering process is performed on the mixed data set to obtain risk classification data, and potential defaulting users are screened out based on the risk classification data, including: Determine the number of classification clusters based on the number of risk types in the risk credit data set; Based on the number of classification clusters, the K-means algorithm is used to cluster the mixed data set to obtain risk classification data; According to the distribution of the credit data of the user to be detected in the risk classification data, it is determined whether the user to be detected is a potential default user.

4. A contract performance risk early warning system based on user data, characterized in that: The system comprises: A data acquisition module is used to acquire credit behavior change data and collateral price change data of different users to be detected within a preset time period, wherein the credit behavior change data includes change information of borrowing frequency, change information of the number of lending institutions and change information of repayment method; A risk data mixing module, used to mix a plurality of credit behavior change data with a preset risk credit behavior change data set to obtain a mixed data set, wherein the risk credit behavior change data set includes credit behavior change data of a plurality of known defaulting users within a preset time period; A potential default user screening module is used to perform clustering processing on the mixed data set to obtain risk classification data, and screen out potential default users based on the risk classification data; The risk value calculation module for potential default users is used to obtain the historical credit data of potential default users, input the historical credit data and credit behavior change data into the preset performance risk value prediction algorithm, and output the performance risk value; The risk prediction module for potential defaulting users is used to determine whether the collateral price change data of the potential defaulting user meets the preset conditions if the performance risk value is greater than the preset risk threshold. If not, the performance risk of the potential defaulting user is confirmed to be high risk.

5. The contract performance risk early warning system based on user data according to claim 4 is characterized in that: The performance risk value prediction algorithm is expressed as: V k =V history +V0, V0=ln(1+e F+L+M ), Where V k represents the performance risk value of user k, V history represents the performance impact item caused by historical credit data, V0 represents the performance impact item caused by credit behavior change data, M k represents the total number of historical credit data of user k, t i Indicates the time distance between the i-th historical credit data and the current time, represents the weight of the i-th historical credit data, A i is the loan amount of the i-th historical credit data, ΔA i is the difference between the loan amount of the i-th historical credit data and the current loan amount, x i represents the overdue status of the i-th historical credit data, and x i =0 means not overdue, x i =1 means overdue, h i represents the number of days overdue for the i-th historical credit data, H i It represents the total number of days of the loan term of the i-th historical credit data. F, L and M are the normalized change information of borrowing frequency, change information of the number of lending institutions and change information of repayment method, respectively. Among them, M=0 represents automatic repayment and M=1 represents manual repayment.

6. The contract performance risk early warning system based on user data according to claim 4 is characterized in that: The clustering process is performed on the mixed data set to obtain risk classification data, and potential defaulting users are screened out based on the risk classification data, including: Determine the number of classification clusters based on the number of risk types in the risk credit data set; Based on the number of classification clusters, the K-means algorithm is used to cluster the mixed data set to obtain risk classification data; According to the distribution of the credit data of the user to be detected in the risk classification data, it is determined whether the user to be detected is a potential default user.

7. An electronic device, characterized in that: include: A processor and a memory, wherein the memory is used to store computer program code, wherein the computer program code includes computer instructions, and when the processor executes the computer instructions, the electronic device executes the contract performance risk warning method based on user data as described in any one of claims 1 to 3.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes the user data-based performance risk warning method as described in any one of claims 1 to 3.

Images