Intelligent risk identification method based on large model

Through the intelligent risk identification method of big model, users' risk levels are calculated and loan applications are segmented, which solves the problem that banks find it difficult to quantify user loan risks, realizes segmentation assessment of user risks and adjusts risk control strategies, and reduces bank bad debt risks .

CN119090628AInactive Publication Date: 2024-12-06JIANGSU RED NET TECH CO LTD

Patent Information

Application Number
CN202411357827.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-27
Publication Date
2024-12-06
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

When banks receive loan applications, since the user's loan risk involves many aspects, it is difficult for tellers to quantify the user's loan risk, especially when there is a lack of multiple or continuous overdue records in the user's credit report, it is difficult to determine whether the user has loan risk. This may affect the operation of the bank.

Method used

Using intelligent risk identification method based on big models, the user's risk level is calculated by obtaining the user's identity proof, income proof and liability situation, including the first risk level, the second risk level and the third risk level, and the user is analyzed based on these levels. loan application, subdivide users' risk levels, and adjust risk control strategies through overdue inference index.

Benefits of technology

It has realized a segmented risk assessment of user loan applications, helping banks understand users' risk levels, avoid one-size-fits-all rejection of loan applications, timely adjust risk control strategies, and reduce bank bad debt risks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119090628A_ABST
    Figure CN119090628A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of risk identification, and particularly relates to an intelligent risk identification method based on a large model, and the method comprises the following steps: 1, obtaining user information which comprises an identity certificate, an income certificate, a debt condition and a credit report; by setting a first risk level, a second risk level and a third risk level, dividing users submitting loan applications into three intervals, determining whether the users are in high-risk levels according to portraits of the users, and analyzing the risk levels of the users based on user information relative to the loan applications of the users through subjective auditing. The risk levels of the users can be subdivided to distinguish different users, and when the credit reports of the users do not have overdue records or have few overdue records, the users can be subdivided through the risk identification model, so that the risk levels of the users can be obtained, a bank can be assisted to know the users, and the user experience is improved. And determining whether the loan application of the user is passed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of risk identification, and specifically is an intelligent risk identification method based on a large model. Background Art

[0002] For example, when a bank receives a user's loan application, it needs to obtain user information in advance in order to avoid the user's overdue repayments, which would lead to an increase in the bank's bad debts.

[0003] Analyze whether the user has loan risks based on the user's identity proof, income proof, debt situation and credit report. Through identity proof, the identity of the lender is verified first to prevent others from using information to make loans. Through income proof, it can be measured whether the user has the ability to repay. Through debt situation and credit report, it can assist banks in understanding the user's historical loan information and credit.

[0004] However, in actual application, the above review of user loan applications is mainly verified by tellers, which has the following problems: when banks receive user loan applications, since the user's loan risk involves multiple aspects, it is difficult for tellers to quantify the user's loan risk. It is foreseeable that when a user's credit report has multiple overdue records or continuous overdue records, it can be easily determined that the user has loan risks, but when the user's credit report does not have multiple overdue records or continuous overdue records, it is difficult to determine whether the user has loan risks. If the loan application of users with loan overdue records is rejected in a one-size-fits-all manner, it may affect the operation of the bank;

[0005] To this end, the present invention provides an intelligent risk identification method based on a large model. Summary of the invention

[0006] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0007] The technical solution adopted by the present invention to solve the technical problem is: an intelligent risk identification method based on a large model described in the present invention comprises the following steps:

[0008] Step 1: Obtain user information, including identity proof, income proof, debt status and credit report;

[0009] Step 2: Verify user information and analyze the user's risk level based on the user information;

[0010] Step 3: Obtain the user's risk assessment results based on the risk level, including:

[0011] A first assessment result, wherein the first assessment result satisfies a risk level of a first risk level;

[0012] A second assessment result, wherein the second assessment result satisfies the risk level as a second risk level;

[0013] A third assessment result, wherein the third assessment result satisfies the risk level as the third risk level;

[0014] Step 4: Obtain the risk assessment results of the user, approve the loan applications of users who meet the first and second assessment results, and reject the loan applications of users who meet the third assessment results.

[0015] Preferably, the risk level in step 2 includes a first risk level, a second risk level and a third risk level, and the method for determining the risk level is:

[0016] Calculates the user's first debt-to-income ratio , and statistics of historical overdue information of users; the historical overdue information includes the number of overdue times and continuous overdue records;

[0017] If the user has a continuous overdue record, the user is judged to be at the third risk level;

[0018] If the user's overdue number of , the user is judged to be at the third risk level;

[0019] If the user's overdue number of ,and , the user is judged to be at the third risk level;

[0020] If the user's overdue number of ,and , the user is judged to be at the second risk level;

[0021] If the user's overdue number of ,and , the user is judged to be at the second risk level;

[0022] If the user's overdue number of ,and , the user is judged to be at the first risk level;

[0023] Among them, the first debt-to-income ratio Calculated according to the formula:

[0024] in, is the user's first month debt amount, is the user's first month income, is the first threshold, is the second threshold, and .

[0025] Preferably, the method further comprises step 5: calculating the debt-to-income ratio according to the first debt-to-income ratio. Calculate the user's maximum new loan share .

[0026] Preferably, the maximum new loan share in step 5 is The method to obtain is:

[0027] Obtain user debt status, income proof, and credit report;

[0028] According to the formula:

[0029]

[0030] in, is the first debt-to-income ratio, is the first threshold;

[0031] Calculate the user's maximum new loan share .

[0032] Preferably, the method further comprises step six:

[0033] Data collection, obtaining users’ historical repayment records;

[0034] Data preprocessing: cleaning and preprocessing the data to remove missing values ​​and outliers;

[0035] Extract features and obtain the number of overdue payments in the user's historical repayment records ;

[0036] Based on the number of overdue payments Calculate the user's overdue speculation index .

[0037] Preferably, the user's overdue speculation index in step 6 The method to obtain is:

[0038] Get the user's overdue speculation index base , overdue growth coefficient and number of overdue ;

[0039] According to the formula:

[0040]

[0041] in, is the number of loan periods, is the base period, is the correction factor, represents the overdue risk growth coefficient, is the basic risk factor, is the growth factor;

[0042] Calculate the user's overdue speculation index .

[0043] Preferably, the overdue speculation index base Q is set according to the risk level of the user, and includes a first index base and a second index base.

[0044] The second index base number corresponds to users whose risk level is the second risk level, and the first index base number corresponds to users whose risk level is the first risk level.

[0045] Preferably, the method further includes step 7: updating the user's overdue speculation index base according to the income certificate submitted by the user periodically. , the method is:

[0046] The user submits income proof information on a regular basis, and the income proof information includes the second month's income amount;

[0047] Calculate the user's second debt-to-income ratio based on the second month's income ;

[0048] According to the second debt-to-income ratio , re-determine the user's risk level;

[0049] Update the user's overdue speculation index base based on the re-determined risk level .

[0050] Preferably, the method further comprises step eight:

[0051] Based on the user's overdue index , obtain the overdue risk determination result;

[0052] The first determination result satisfies the user's overdue speculation index. Lower than the first preset estimation index;

[0053] The second determination result satisfies the user's overdue speculation index. Not less than the first preset estimation index and less than the second preset estimation index;

[0054] The third determination result satisfies the user's overdue speculation index Not less than the second preset estimation index;

[0055] The second preset inference index is greater than the first preset inference index.

[0056] Preferably, when the overdue risk determination result is the third determination result, an early warning instruction is output to issue an early warning.

[0057] The beneficial effects of the present invention are as follows:

[0058] 1. The present invention discloses an intelligent risk identification method based on a large model, which divides users who submit loan applications into three intervals by setting a first risk level, a second risk level and a third risk level. It can determine whether the user is at a high risk level based on the user's portrait. Compared with the subjective review of the user's loan application, the user's risk level is analyzed based on user information, and then the user's loan application is reviewed based on the risk level. The user's risk level can be subdivided to distinguish different users. When there is no overdue record or a small number of overdue records in the user's credit report, the user can be subdivided through the above-mentioned risk identification model, and then the user's risk level can be known, which assists the bank in understanding the user and determining whether to approve the user's loan application.

[0059] 2. The present invention describes an intelligent risk identification method based on a large model. By setting an overdue speculation index base according to the risk level of the user when the loan application is approved, the corresponding overdue risk growth coefficient will change according to the number of overdue payments of the user during the repayment period, causing the user's overdue speculation index to increase. As the number of overdue payments of the user increases during the repayment period, the overdue speculation index will gradually increase. When the overdue speculation index corresponding to the user gradually increases, the bank will pay more attention to the user's economic changes and adjust the risk control strategy in a timely manner, including collection and application for freezing of user property and other operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The present invention will be further described below in conjunction with the accompanying drawings.

[0061] Figure 1 is a flow chart of the present invention; DETAILED DESCRIPTION

[0062] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0063] like Figure 1 As shown, an intelligent risk identification method based on a large model according to an embodiment of the present invention comprises the following steps:

[0064] Step 1: Obtain user information, including identity proof, income proof, debt status and credit report;

[0065] Step 2: Verify user information and analyze the user's risk level based on the user information;

[0066] In one embodiment, the risk level in step 2 includes a first risk level, a second risk level, and a third risk level, and the method for determining the risk level is:

[0067] Calculates the user's first debt-to-income ratio , and statistics of historical overdue information of users; the historical overdue information includes the number of overdue times and continuous overdue records;

[0068] If the user has a continuous overdue record, the user is judged to be at the third risk level;

[0069] If the user's overdue number of , the user is judged to be at the third risk level;

[0070] If the user's overdue number of ,and , the user is judged to be at the third risk level;

[0071] If the user's overdue number of ,and , the user is judged to be at the second risk level;

[0072] If the user's overdue number of ,and , the user is judged to be at the second risk level;

[0073] If the user's overdue number of ,and , the user is judged to be at the first risk level;

[0074] Among them, the first debt-to-income ratio Calculated according to the formula:

[0075] in, is the user's first month debt amount, is the user's first month income, is the first threshold, is the second threshold, and .

[0076] Theoretically, the monthly income and monthly debt of a user affect whether the user's new loan can be repaid on time or whether the user has the ability to repay the new loan. Assuming that the ratio of the user's monthly income to the monthly debt is close to 50%, it is difficult for the user to repay the new loan, that is, the user is a medium-risk or high-risk loan object. In addition, in addition to the ratio of the monthly income to the monthly debt, it is also necessary to consider whether the user has overdue payments. Assuming that there is an overdue record in the user's credit report, it means that the user is at great risk and may be overdue again, that is, a medium-risk or high-risk loan object; Specifically, in one embodiment of the present invention, the user's risk level can be analyzed according to the user's monthly debt, monthly income and whether there is an overdue situation, so as to achieve the user's profile determination. In one embodiment, when the user has a continuous overdue record, the user is directly determined to be at the third risk level. Correspondingly, if the user's overdue number exceeds 3 times, it means that the user's credit is poor and may continue to be overdue, so the user is also determined to be at the third risk level. When the user is continuously overdue or the number of overdue times exceeds 3 times, there is no need to consider the user's first debt-to-income ratio , and when the user has no continuous overdue record or the number of overdue times is less than 3 times, it is necessary to combine the user's first debt-to-income ratio Analyze the user's risk level, as described above, , Debt-to-income thresholds set for banks, specifically, , ;

[0077] Assume there are users A, B, and C;

[0078] User A's monthly income is 10,000, monthly debt is 1,500, and the number of overdue payments is 2;

[0079] User B's monthly income is 8,000, monthly debt is 2,500, and the number of overdue payments is 1;

[0080] User C's monthly income is 15,000, monthly debt is 3,000, and the number of overdue payments is 0;

[0081] User D's monthly income is 12,000, monthly debt is 3,000, and the number of overdue payments is 0;

[0082] According to the calculation, user A's first debt-to-income ratio ;

[0083] User B's first debt-to-income ratio ;

[0084] User C's first debt-to-income ratio ;

[0085] User D's First Debt-to-Income Ratio ;

[0086] Based on the above, user A is at the second risk level, user B is at the third risk level, user C is at the first risk level, and user D is at the second risk level.

[0087] Step 3: Obtain the user's risk assessment results based on the risk level, including:

[0088] A first assessment result, wherein the first assessment result satisfies a risk level of a first risk level;

[0089] A second assessment result, wherein the second assessment result satisfies the risk level as a second risk level;

[0090] A third assessment result, wherein the third assessment result satisfies the risk level as the third risk level;

[0091] Step 4: Obtain the risk assessment results of the user, approve the loan applications of users who meet the first and second assessment results, and reject the loan applications of users who meet the third assessment results.

[0092] When receiving a user's loan application, in order to avoid the user's overdue repayment, which may lead to an increase in the bank's bad debts, the bank needs to obtain user information in advance and analyze whether the user has loan risks based on the user's identity proof, income proof, debt situation and credit report. Through identity proof, the identity of the lender is verified first to prevent others from using information to apply for loans. Through income proof, it can be measured whether the user has the ability to repay. Through debt situation and credit report, it can assist the bank to understand the user's historical loan information and credit; However, in actual application, the above review of user loan applications is mainly verified by tellers, and there are the following problems: When a bank receives a user's loan application, since the user's loan risk involves multiple aspects, it is difficult for the teller to quantify the user's loan risk. It can be foreseen that when the user's credit report has multiple overdue records or continuous overdue records, it can be easily determined that the user has loan risks, but when the user's credit report does not have multiple overdue records or continuous overdue records, it is difficult to determine whether the user has loan risks. If the loan application of users with loan overdue records is rejected in a one-size-fits-all manner, it may affect the operation of the bank;

[0093] In one embodiment of the present invention, a risk identification model based on user income, debt and credit is set. The risk identification model is based on the user's first debt-to-income ratio. , historical overdue information is used to analyze the risk level of users, that is, to construct profiles of different users, the profiles specifically include the first risk level, the second risk level and the third risk level. In one embodiment, the first risk level, the second risk level and the third risk level are set, and users who submit loan applications are divided into three intervals. According to the user's profile, it can be determined whether the user is at a high risk level. Compared with the subjective review of the user's loan application, the user's risk level is analyzed based on user information, and then the user's loan application is reviewed based on the risk level. The user's risk level can be subdivided to distinguish different users. When there is no overdue record or a small number of overdue records in the user's credit report, the user can be subdivided through the above risk identification model, and then the user's risk level can be known, which assists the bank in understanding the user and determining whether to approve the user's loan application;

[0094] It is worth noting that, in one embodiment, the first risk level is low risk, the second risk level is medium risk, and the third risk level is high risk.

[0095] In one embodiment, the step further includes: Calculate the user's maximum new loan share .

[0096] The ratio of the user's monthly debt to monthly income is the user's debt-to-income ratio. By calculating the user's debt-to-income ratio, the user's repayment ability can be reflected, and the user's maximum new loan share can be calculated based on the debt-to-income ratio. It is worth noting that in theory, banks will set a maximum debt-to-income ratio for loan users. For example, a bank sets a maximum debt-to-income ratio of 36%. Assuming that user A has a monthly income of 10,000 yuan and an existing debt that requires a monthly repayment of 1,500 yuan, his current debt-to-income ratio is 15%. This means that 15% of user A's monthly income is used to pay off existing debts. According to the first threshold set by the bank, , then the user A can theoretically only afford a maximum of RMB 2,100 in new monthly loan payments, and this is based on the user's stable monthly income.

[0097] In one embodiment, the maximum new loan share in step 5 is The method to obtain is:

[0098] Obtain user debt status, income proof, and credit report;

[0099] According to the formula:

[0100]

[0101] in, is the first debt-to-income ratio, is the first threshold;

[0102] Calculate the user's maximum new loan share .

[0103] According to the above, in one embodiment of the present invention, by obtaining the highest debt-to-income ratio set by the bank, that is, the first threshold , and then calculate the difference between the user's current debt-to-income ratio and multiply it by the user's monthly income to get the user's new monthly loan share, that is, the user's maximum new loan share. It is worth noting that each bank has a different threshold for the first threshold. The settings can be different;

[0104] As mentioned above, user A's monthly debt is 1500, and the maximum new loan share is obtained through calculation. Therefore, when the loan application submitted by user A is approved, based on its highest new loan share, if its total loan share is high, it will need to extend the loan period; if its total loan share is low, its loan period will also be relatively low;

[0105] For example, if the total loan share of user A is 120,000, then according to his maximum new loan share, user A may need to repay the loan in 60 installments; and when the total loan share of user A is 70,000, according to the maximum new loan share, user A may need to repay the loan in 36 installments.

[0106] In one embodiment, step six is ​​also included:

[0107] Data collection, obtaining users’ historical repayment records;

[0108] Data preprocessing: cleaning and preprocessing the data to remove missing values ​​and outliers;

[0109] Extract features and obtain the number of overdue payments in the user's historical repayment records ;

[0110] Based on the number of overdue payments Calculate the user's overdue speculation index .

[0111] In order to avoid bad debts, it is also necessary to predict the overdue risk for all users who are currently repaying their loans, that is, to calculate the user's overdue speculation index based on the user's repayment record during the repayment process, specifically the number of overdue payments. It is worth noting that the user's overdue speculation index calculated here is used to measure whether the user's repayment in the current period will be overdue. By calculating the overdue speculation index, banks can understand the repayment enthusiasm of all users. For users with a higher overdue speculation index, banks can pay more attention to the user's economic changes and adjust risk control strategies in a timely manner, including collection and application to freeze user property.

[0112] In one embodiment, the overdue speculation index It is calculated based on the overdue inferred index base Q, and the overdue inferred index base Q is set according to the risk level of the user, including a first index base and a second index base.

[0113] The second index base number corresponds to users whose risk level is the second risk level, and the first index base number corresponds to users whose risk level is the first risk level.

[0114] Corresponding to the above, based on the construction of user portraits, the overdue speculation index bases of users who meet the first risk level, the second risk level and the third risk level are all different. In one embodiment of the present invention, different overdue speculation index bases are set for different users according to the user's risk level distribution, specifically including a first index base and a second index base, which correspond to users of the first risk level and the second risk level respectively. It is worth noting that in one embodiment, the second index base can be set to 60, and the first index base can be set to 30.

[0115] For example, if user A has a total loan amount of 120,000 and needs to repay the loan in 60 installments, then the overdue estimated index is calculated based on the above. .

[0116] In one embodiment, the user's overdue estimated index in step 6 is The method to obtain is:

[0117] Get the user's overdue speculation index base , overdue growth coefficient and number of overdue ;

[0118] According to the formula:

[0119]

[0120] in, is the number of loan periods, is the base period, is the correction factor, represents the overdue risk growth coefficient, is the basic risk factor, is the growth factor;

[0121] Calculate the user's overdue speculation index .

[0122] Based on the above, according to the overdue speculation index base, overdue growth coefficient and the number of overdue times set by the bank for the user, the user's overdue speculation index can be obtained in real time. In one embodiment of the present invention, the overdue speculation index base is set by the bank and is set according to the risk level of the user when the loan application is passed. It is worth noting that the overdue speculation index base of each user is different, while the basic risk coefficient, growth factor, etc. are all set by the bank and are the same for each user. According to the number of overdue times of the user during the repayment period, the corresponding overdue risk growth coefficient will change, causing the user's overdue speculation index to increase, and as the number of overdue times of the user during the repayment period increases, the overdue speculation index will gradually increase. When the overdue speculation index corresponding to the user gradually increases, the bank will pay more attention to the user's economic changes and adjust the risk control strategy in a timely manner, including collection and application for freezing of user property and other operations. Among them, the basic risk coefficient Set to 0.1, indicating the basic risk coefficient when there is no overdue, growth factor Set to 1.5, which means that for each additional overdue payment, the risk factor increases by 50%, and the basic risk factor , Growth Factor All are set by banks; n=12;

[0123] And according to the above, since the user has the highest new loan share, when the total loan share proposed by the user is high, it may lead to more loan periods. As the loan period increases, the user's monthly income fluctuates, so there is a greater risk of overdue payment. Based on the above, on the basis of the user's overdue speculation index base, consider the ratio of the user's total loan period to the basic period, and calculate the overdue speculation index base corresponding to users with different loan periods. When two users are at the same risk level, for example, the above-mentioned user A and user D are both at the second risk level, but the total loan period of user A is different from the total loan period of user D. At this time, the overdue speculation index bases of user A and user D are distinguished by the total loan period, so that the difference in overdue risk between users is magnified, thereby facilitating banks to manage and monitor users' repayments.

[0124] In one embodiment, the step 7 is also included: updating the user's overdue speculation index base according to the income certificate submitted by the user periodically , the method is:

[0125] The user submits income proof information on a regular basis, and the income proof information includes the second month's income amount;

[0126] Calculate the user's second debt-to-income ratio based on the second month's income ;

[0127] According to the second debt-to-income ratio , re-determine the user's risk level;

[0128] Update the user's overdue speculation index base based on the re-determined risk level .

[0129] During the repayment period, if the user has too many repayment periods, such as 24 repayment periods, 60 repayment periods, or even 300 repayment periods, the corresponding risks are not the same. In theory, as the number of repayment periods increases, the user's income fluctuates, which may cause the user's debt-to-income ratio to rise, and then exceed the maximum debt-to-income ratio set by the bank, which is the first threshold. In one embodiment of the present invention, during the repayment period, the user's second debt-to-income ratio can be obtained based on the income proof information submitted by the user periodically. , based on the user's second debt-to-income ratio , re-determine the user's risk level, and then adjust the user's overdue speculation index base ;

[0130] For example: The overdue estimated index base when user A submits a loan application , assuming that user A's income decreases, resulting in user A's second debt-to-income ratio Exceeding the first threshold , then user A changes from the second risk level to the third risk level, and the third index base corresponding to the third risk level is set to 90, then the overdue estimated index base of user A is .

[0131] In one embodiment, step eight is also included:

[0132] Based on the user's overdue index , obtain the overdue risk determination result;

[0133] The first determination result satisfies the user's overdue speculation index. Lower than the first preset estimation index;

[0134] The second determination result satisfies the user's overdue speculation index. Not less than the first preset estimation index and less than the second preset estimation index;

[0135] The third determination result satisfies the user's overdue speculation index Not less than the second preset estimation index;

[0136] wherein the second preset inference index is greater than the first preset inference index;

[0137] When the overdue risk determination result is the third determination result, an early warning instruction is output to issue an early warning.

[0138] Based on the above, when the user's overdue speculation index If it is higher than the second preset inference index, an early warning instruction needs to be output to remind the bank to pay attention to the user's repayment situation and economic changes. If necessary, an application can be made to freeze the user's property to avoid bad debts for the bank.

[0139] After the user's loan application is approved, the bank needs to pay attention to the user's repayment situation and record whether the user has any overdue payments. Based on the above, the calculated overdue speculation index of the user can assist the bank in analyzing whether the user can repay on time within the repayment period. In one embodiment of the present invention, when the calculated overdue speculation index of the user is not lower than the first preset speculation index, it indicates that the user has a history of overdue payments and is likely to be overdue at the next repayment period. Therefore, the bank needs to pay more attention to the user's economic changes and adjust the risk control strategy in time, including collection and application for freezing of user property. On the contrary, when the user's overdue speculation index is lower than the second preset speculation index, it is not necessary to pay more attention to the user's economic changes.

[0140] The first preset inference index is 60, and the second preset inference index is 90. According to the above, if the income of user A does not change significantly during the repayment period, the overdue inference index of user A is , is less than the second preset inference index and greater than the first preset inference index, then the overdue risk determination result of user A is the second determination result; assuming that user A's income decreases significantly during the repayment period, then the overdue inference index of user A is , is greater than the second preset prediction index, then the overdue risk determination result of user A is the third determination result.

[0141] Working principle: In one embodiment of the present invention, a risk identification model based on user income, debt and credit is set. The risk identification model is based on the user's first debt-to-income ratio. , analyze the risk level of users based on historical overdue information, that is, construct profiles of different users, and the profiles specifically include a first risk level, a second risk level, and a third risk level. In one embodiment, the first risk level, the second risk level, and the third risk level are set, and users who submit loan applications are divided into three intervals. According to the user's profile, it can be determined whether the user is at a high risk level. Compared with subjective review of the user's loan application, the user's risk level is analyzed based on user information, and then the user's loan application is reviewed based on the risk level. The user's risk level can be subdivided to distinguish different users. When there is no overdue record or a small number of overdue records in the user's credit report, the user can be subdivided through the above-mentioned risk identification model, and then the user's risk level can be known, which assists the bank in understanding the user and determining whether to approve the user's loan application.

[0142] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An intelligent risk identification method based on a large model, characterized in that: The following steps are involved: Step 1: Obtain user information, including identity proof, income proof, debt status and credit report; Step 2: Verify user information and analyze the user's risk level based on the user information; Step 3: Obtain the user's risk assessment results based on the risk level, including: A first assessment result, wherein the first assessment result satisfies a risk level of a first risk level; A second assessment result, wherein the second assessment result satisfies the risk level as a second risk level; A third assessment result, wherein the third assessment result satisfies the risk level as the third risk level; Step 4: Obtain the risk assessment results of the user, approve the loan applications of users who meet the first and second assessment results, and reject the loan applications of users who meet the third assessment results.

2. The intelligent risk identification method based on a large model according to claim 1 is characterized by: The risk levels in step 2 include the first risk level, the second risk level and the third risk level. The method for determining the risk level is: Calculates the user's first debt-to-income ratio , and statistics of historical overdue information of users; the historical overdue information includes the number of overdue times and continuous overdue records; If the user has a continuous overdue record, the user is judged to be at the third risk level; If the user's overdue number of , the user is judged to be at the third risk level; If the user's overdue number of ,and , the user is judged to be at the third risk level; If the user's overdue number of ,and , the user is judged to be at the second risk level; If the user's overdue number of ,and , the user is judged to be at the second risk level; If the user's overdue number of ,and , the user is judged to be at the first risk level; Among them, the first debt-to-income ratio Calculated according to the formula: in, is the user's first month debt amount, is the user's first month income, is the first threshold, is the second threshold, and .

3. The intelligent risk identification method based on a large model according to claim 2 is characterized by: Also includes step 5: Based on the first debt-to-income ratio Calculate the user's maximum new loan share .

4. The intelligent risk identification method based on a large model according to claim 3 is characterized by: The maximum new loan share in step 5 The method to obtain is: Obtain user debt status, income proof, and credit report; According to the formula: in, is the first debt-to-income ratio, is the first threshold; Calculate the user's maximum new loan share .

5. The intelligent risk identification method based on a large model according to claim 1 is characterized in that: Also includes step six: Data collection, obtaining users’ historical repayment records; Data preprocessing: cleaning and preprocessing the data to remove missing values ​​and outliers; Extract features and obtain the number of overdue payments in the user's historical repayment records ; Based on the number of overdue payments Calculate the user's overdue speculation index .

6. The intelligent risk identification method based on a large model according to claim 5 is characterized by: The overdue estimated index of the user in step 6 The method to obtain is: Get the user's overdue speculation index base , overdue growth coefficient and number of overdue ; According to the formula: in, is the number of loan periods, is the base period, is the correction factor, represents the overdue risk growth coefficient, is the basic risk factor, is the growth factor; Calculate the user's overdue speculation index .

7. The intelligent risk identification method based on a large model according to claim 6 is characterized by: The overdue speculation index base Q is set according to the risk level of the user, including a first index base and a second index base; The second index base number corresponds to users whose risk level is the second risk level, and the first index base number corresponds to users whose risk level is the first risk level.

8. The intelligent risk identification method based on a large model according to claim 7 is characterized in that: Also includes step 7: Update the user's overdue speculation index base based on the income proof submitted by the user periodically , the method is: The user submits income proof information on a regular basis, and the income proof information includes the second month's income amount; Calculate the user's second debt-to-income ratio based on the second month's income ; According to the second debt-to-income ratio , re-determine the user's risk level; Update the user's overdue speculation index base based on the re-determined risk level .

9. The intelligent risk identification method based on a large model according to claim 8 is characterized by: Also includes step eight: Based on the user's overdue index , obtain the overdue risk determination result; The first determination result satisfies the user's overdue speculation index. Lower than the first preset estimation index; The second determination result satisfies the user's overdue speculation index. Not less than the first preset estimation index and less than the second preset estimation index; The third determination result satisfies the user's overdue speculation index Not less than the second preset estimation index; The second preset inference index is greater than the first preset inference index.

10. The method for intelligent risk identification based on a large model according to claim 9, characterized in that: When the overdue risk determination result is the third determination result, an early warning instruction is output to issue an early warning.

Citation Information

Patent Citations

  • Financial risk prediction method and device and electronic equipment

    CN111192131A

  • Risk assessment method, system and device and storage medium

    CN116452323A

  • Business information prediction method and device, storage medium and electronic device

    CN116664289A

  • Credit line determination method and device, electronic equipment and storage medium

    CN117252677A

  • Information service processing system for financial risk control

    CN117710081A

Cited By

  • Credit report AI intelligent analysis and visualization platform

    CN121660783A