Post-loan risk warning method, device, electronic device and computer-readable medium

Abstract

The present invention discloses a post-loan risk warning method, device, electronic device and computer-readable medium, and relates to the field of big data intelligent analysis technology. A specific implementation of the method includes: performing sample derivation on each sample loan based on the warning time point to obtain a derived sample loan; marking the label of each derived sample loan; wherein the label includes a positive sample and a negative sample; training a model according to each derived sample loan and its corresponding label, thereby obtaining a post-loan risk warning model; and calculating the probability of the target loan becoming a bad debt loan at the future warning time point through the post-loan risk warning model. This implementation can solve the technical problem of inaccurate post-loan risk warning results.

Description

Technical Field

[0001] The present invention relates to the technical field of big data intelligent analysis, and in particular to a post-loan risk early warning method, device, electronic equipment and computer-readable medium. Background Art

[0002] Post-loan risk warning has always been one of the important tools for banks to monitor and manage post-loan. Different from the traditional post-loan management model, banks are now gradually increasing the application of big data analysis technology in post-loan risk warning. By analyzing customers' credit status, changes in business conditions, and behavioral characteristics from multiple dimensions, they can find the key factors that affect customers' post-loan risks, thereby establishing post-loan risk warning rules or models.

[0003] The logic of sample screening and label definition that meets business objectives is the basis for modeling the post-loan risk warning model. At present, the common practice is to refer to the label definition of the access model and use the final quality of the customer as the label for analysis and modeling. This approach is obviously not in line with the needs of the post-loan risk scenario. In fact, the performance of post-loan risk and pre-loan risk is quite different.

[0004] In the process of implementing the present invention, the inventors found that there are at least the following problems in the prior art:

[0005] Currently, traditional financial institutions have two main methods for post-loan risk warning models. One method is to establish post-loan risk warning rules or post-loan risk warning models based on expert experience. The business interpretability of this model is strong, but the effect of the warning is difficult to grasp, and it is easy to have problems such as too high a proportion of warned users or unclear ratios of good and bad users between warned and non-warned users. The other method is to use machine learning algorithms to establish a post-loan risk warning model. Before modeling, it is necessary to define the positive and negative labels of users and select modeling samples.

[0006] However, from the current post-loan risk warning model, there are unreasonable aspects in the definition of positive and negative labels and sample screening. The definition of positive and negative labels is generally consistent with the positive and negative labels of the pre-loan access model. The user number is used as the sample primary key, and whether the user has overdue payments during the entire repayment period is used as a sign to distinguish between positive and negative samples, thereby screening samples based on the user dimension. This will lead to inaccurate post-loan risk warning results, which in turn leads to greater pressure on post-loan risk warning processing. Summary of the invention

[0007] In view of this, embodiments of the present invention provide a post-loan risk warning method, device, electronic device and computer-readable medium to solve the technical problem of inaccurate post-loan risk warning results.

[0008] To achieve the above object, according to one aspect of an embodiment of the present invention, a post-loan risk early warning method is provided, comprising:

[0009] Perform sample derivation on each sample loan based on the early warning time point to obtain derived sample loans;

[0010] Labeling each of the derivative sample loans; wherein the labels include positive samples and negative samples;

[0011] A model is trained based on each of the derivative sample loans and their corresponding labels, thereby obtaining a post-loan risk warning model;

[0012] The post-loan risk early warning model is used to calculate the probability that the target loan will become a bad debt loan at a future early warning time point.

[0013] Optionally, sample derivation is performed on each sample loan based on the early warning time point to obtain a derived sample loan, including:

[0014] Filter out loans with repayment periods that reach the performance period;

[0015] Calculate the bad debt date of each loan, and select sample loans according to the bad debt date of each loan;

[0016] For each sample loan, a derivative sample loan is generated according to the early warning time point of the sample loan.

[0017] Optionally, calculating the bad debt date of each loan, and selecting sample loans according to the bad debt date of each loan, includes:

[0018] For each loan, the starting date after N days of overdue is taken as the bad debt date of the loan;

[0019] Eliminate loans with bad debt dates within M months after loan disbursement, thereby selecting sample loans;

[0020] Among them, N is greater than or equal to 25 and less than or equal to 45, and M is greater than or equal to 2 and less than or equal to 6.

[0021] Optionally, generating a derivative sample loan according to the early warning time point of the sample loan includes:

[0022] Randomly select the sample loans at the end of S months after the loan is disbursed as the warning time point; where S is greater than or equal to 2;

[0023] Derivative sample loans are generated according to various warning time points of the sample loans.

[0024] Optionally, generating a derivative sample loan according to each warning time point of the sample loan includes:

[0025] For each warning time point, generate a derivative sample loan corresponding to the warning time point;

[0026] For a sample loan of bad debt, if the warning time point of the sample loan is after the bad debt date of the sample loan, the derivative sample loan corresponding to the warning time point is screened out.

[0027] Optionally, the label for marking each of the derivative sample loans includes:

[0028] For a sample loan of bad debt, determine whether the time interval between the warning time point of the sample loan and the bad debt date of the sample loan is greater than T months; wherein T is greater than or equal to 2 and less than or equal to 6;

[0029] If yes, the derivative sample loan corresponding to the warning time point is marked as a positive sample;

[0030] If not, the derivative sample loans corresponding to the warning time point are marked as negative samples.

[0031] Optionally, a model is trained based on each of the derivative sample loans and their corresponding labels to obtain a post-loan risk warning model, including:

[0032] Randomly extracting derivative sample loans with positive labels and derivative sample loans with negative labels, so that the number ratio of the derivative sample loans with positive labels to the derivative sample loans with negative labels is a preset ratio;

[0033] The model is trained using the extracted derivative sample loans and their corresponding labels to obtain a post-loan risk warning model.

[0034] Optionally, the probability of the target loan becoming a bad debt loan at a future warning time point is calculated by the post-loan risk warning model, including:

[0035] The post-loan risk early warning model is used to calculate the probability that the target loan will become a bad debt loan at the end of each future month.

[0036] In addition, according to another aspect of the embodiment of the present invention, a post-loan risk early warning device is provided, comprising:

[0037] The derivative module is used to derive samples of each sample loan based on the early warning time point to obtain derived sample loans;

[0038] A labeling module, used to label each of the derivative sample loans; wherein the label includes positive samples and negative samples;

[0039] A training module, used for training a model based on each of the derivative sample loans and their corresponding labels, so as to obtain a post-loan risk warning model;

[0040] The early warning module is used to calculate the probability of the target loan becoming a bad debt loan at a future early warning point through the post-loan risk early warning model.

[0041] Optionally, the derivative module is further used for:

[0042] Filter out loans with repayment periods that reach the performance period;

[0043] Calculate the bad debt date of each loan, and select sample loans according to the bad debt date of each loan;

[0044] For each sample loan, a derivative sample loan is generated according to the early warning time point of the sample loan.

[0045] Optionally, the derivative module is further used for:

[0046] For each loan, the starting date after N days of overdue is taken as the bad debt date of the loan;

[0047] Eliminate loans with bad debt dates within M months after loan disbursement, thereby selecting sample loans;

[0048] Among them, N is greater than or equal to 25 and less than or equal to 45, and M is greater than or equal to 2 and less than or equal to 6.

[0049] Optionally, the derivative module is further used for:

[0050] Randomly select the sample loans at the end of S months after the loan is disbursed as the warning time point; where S is greater than or equal to 2;

[0051] Derivative sample loans are generated according to various warning time points of the sample loans.

[0052] Optionally, the derivative module is further used for:

[0053] For each warning time point, generate a derivative sample loan corresponding to the warning time point;

[0054] For a sample loan of bad debt, if the warning time point of the sample loan is after the bad debt date of the sample loan, the derivative sample loan corresponding to the warning time point is screened out.

[0055] Optionally, the marking module is further used for:

[0056] For a sample loan of bad debt, determine whether the time interval between the warning time point of the sample loan and the bad debt date of the sample loan is greater than T months; wherein T is greater than or equal to 2 and less than or equal to 6;

[0057] If yes, the derivative sample loan corresponding to the warning time point is marked as a positive sample;

[0058] If not, the derivative sample loans corresponding to the warning time point are marked as negative samples.

[0059] Optionally, the training module is further used to:

[0060] Randomly extracting derivative sample loans with positive labels and derivative sample loans with negative labels, so that the number ratio of the derivative sample loans with positive labels to the derivative sample loans with negative labels is a preset ratio;

[0061] The model is trained using the extracted derivative sample loans and their corresponding labels to obtain a post-loan risk warning model.

[0062] Optionally, the early warning module is further used for:

[0063] The post-loan risk early warning model is used to calculate the probability that the target loan will become a bad debt loan at the end of each future month.

[0064] According to another aspect of an embodiment of the present invention, there is further provided an electronic device, including:

[0065] one or more processors;

[0066] a storage device for storing one or more programs,

[0067] When the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any one of the above embodiments.

[0068] According to another aspect of an embodiment of the present invention, a computer-readable medium is provided, on which a computer program is stored. When the program is executed by a processor, the method described in any one of the above embodiments is implemented.

[0069] One embodiment of the above invention has the following advantages or beneficial effects: because the technical means of deriving samples of each sample loan based on the warning time point, obtaining derived sample loans, and marking the labels of each derived sample loan are adopted, the technical problem of inaccurate post-loan risk warning results in the prior art is overcome. The embodiment of the present invention fully considers the scene characteristics of post-loan risk warning, derives sample loans by warning time points, and screens out suitable samples and defines suitable labels, so that the post-loan risk warning model constructed by the derived samples can accurately calculate the probability of the loan becoming a bad debt at the future warning time point. Compared with the prior art that only predicts that the loan will eventually become a bad debt, the embodiment of the present invention can reduce the warning ratio each month, thereby reducing the pressure of post-loan risk warning processing.

[0070] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0072] Figure 1 is a schematic diagram of the main process of the post-loan risk early warning method according to an embodiment of the present invention;

[0073] Figure 2 It is a schematic diagram of the main process of a post-loan risk early warning method according to a reference embodiment of the present invention;

[0074] Figure 3 It is a schematic diagram of the main process of a post-loan risk early warning method according to another reference embodiment of the present invention;

[0075] Figure 4 It is a schematic diagram of the main process of a post-loan risk early warning method according to another reference embodiment of the present invention;

[0076] Figure 5 is a schematic diagram of the main modules of the post-loan risk early warning device according to an embodiment of the present invention;

[0077] Figure 6 is an exemplary system architecture diagram to which embodiments of the present invention may be applied;

[0078] Figure 7 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0079] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0080] From the current post-loan risk warning model, there are unreasonable aspects in the definition of positive and negative labels and sample screening. The definition of positive and negative labels is generally consistent with the positive and negative labels of the pre-loan access model, with the user number as the sample primary key, and whether the user has overdue payments during the entire repayment period as a sign to distinguish the positive and negative samples, thereby screening samples based on the user dimension. In fact, there are significant differences in the performance of post-loan risk and pre-loan risk, so the sample screening and label definition methods of the post-loan risk warning model are different from those of the access model. The user's post-loan risk has the characteristics of gradually appearing, and it will only appear as a bad debt after a certain point in time after the loan, and it will appear as a good debt before that point in time. Therefore, the sample should use the loan number + warning time point as the primary key for label definition and sample screening.

[0081] In order to solve the technical problems existing in the prior art, the embodiment of the present invention utilizes the characteristic that post-loan risk has the gradual emergence, and proposes a post-loan risk early warning method, which can accurately predict the probability of a loan becoming a bad debt at each point in time, thereby achieving the goal of big data post-loan early warning.

[0082] Figure 1 Schematic diagram of the main process of the post-loan risk warning method according to an embodiment of the present invention. Figure 1 As shown, the post-loan risk early warning method may include:

[0083] Step 101, performing sample derivation on each sample loan based on the early warning time point to obtain a derived sample loan.

[0084] In order to train a better post-loan risk warning model, it is necessary to first construct a suitable sample. The embodiment of the present invention derives the sample based on the existing sample to generate a derived sample loan.

[0085] Optionally, step 101 may include: screening out loans whose repayment periods have reached the performance period; calculating the bad debt date of each loan, and screening out sample loans according to the bad debt date of each loan; for each sample loan, generating a derivative sample loan according to the warning time point of the sample loan. In order to have both negative samples and positive samples in the sample, it is necessary to first screen out loans whose repayment periods have reached the performance period, then calculate the bad debt date of these loans, and screen out sample loans according to the bad debt date of the loans, and finally generate a derivative sample loan according to the warning time point of the sample loan, that is, each warning time point corresponds to a derivative sample loan.

[0086] In an embodiment of the present invention, for example, if the performance period is 6 months (whether each loan has gone bad can only be determined in the 6th month after the loan is disbursed), the loan disbursed 6 months ago is selected; for example, if the performance period is 12 months (whether each loan has gone bad can only be determined in the 12th month after the loan is disbursed), the loan disbursed 12 months ago is selected; if the performance period is 13 months (whether each loan has gone bad can only be determined in the 13th month after the loan is disbursed), the loan disbursed 13 months ago is selected; if the performance period is 18 months (whether each loan has gone bad can only be determined in the 18th month after the loan is disbursed), the loan disbursed 18 months ago is selected.

[0087] Before defining labels, first screen out suitable samples, exclude samples that obviously do not meet the warning requirements, retain samples that can show positive and negative results, and then derive samples based on the warning time point.

[0088] Optionally, the bad debt date of each loan is calculated, and sample loans are screened out according to the bad debt date of each loan, including: for each loan, taking the starting date after N days of overdue as the bad debt date of the loan; screening out loans with bad debt dates within M months after the loan is issued, thereby screening out sample loans; wherein N is greater than or equal to 25 and less than or equal to 45, and M is greater than or equal to 2 and less than or equal to 6. In an embodiment of the present invention, the date on which the loan becomes a bad debt (i.e., the bad debt date) is defined as the starting date after N days of overdue, indicating that the loan is considered a bad debt from this day onwards. If N=30, the starting date after 30 days of overdue is taken as the bad debt date of the loan; if N=35, the starting date after 35 days of overdue is taken as the bad debt date of the loan; and N=45, the starting date after 45 days of overdue is taken as the bad debt date of the loan. Since some loans may have the risk of fraud, it is necessary to screen out loans with a greater risk of fraud, and the retained loans are sample loans. For example, if a loan goes bad within 3 months after it is disbursed, then the possibility of fraud risk for this loan is high, and it is difficult for the post-loan risk warning model to predict this risk. For example, if the loan date of user A is June 9, 2021, and it starts to be overdue one month later until it is 30 days overdue, then this loan should be screened out when screening samples to avoid affecting the prediction effect of the model.

[0089] Optionally, generating a derivative sample loan based on the warning time point of the sample loan includes: randomly selecting the sample loan at the end of S months after the loan is disbursed as the warning time point; wherein S is greater than or equal to 2; generating a derivative sample loan based on each warning time point of the sample loan. Since the post-loan risk warning is triggered every month after the loan is disbursed, the end of N months after the loan can be randomly selected for each loan as the warning time point. If S is 3, the sample loan is randomly selected at the end of 3 months after the loan is disbursed as the warning time point. For example, the warning time point may be the end of 3 months, 4 months, and 8 months after the loan for loan A, the warning time point corresponding to loan B is the end of 4 months, 7 months, and 11 months after the loan, and the warning time point corresponding to loan C is the end of 5 months, 8 months, and 11 months after the loan. If S is 4, then the randomly selected sample loans will use the end of 4 months after the loan as the warning time point. For example, loan A may use the end of 3 months, 5 months, 8 months, and 13 months after the loan as the warning time point, the corresponding warning time point for loan B may be the end of 2 months, 6 months, 7 months, and 12 months after the loan, and the corresponding warning time point for loan C may be the end of 5 months, 9 months, 11 months, and 15 months after the loan.

[0090] Optionally, derivative sample loans are generated according to each warning time point of the sample loan, including: for each warning time point, generating a derivative sample loan corresponding to the warning time point; for bad debt sample loans, if the warning time point of the sample loan is after the bad debt date of the sample loan, then the derivative sample loan corresponding to the warning time point is screened out. For normal loans, since there are N month-end warning time points, one sample loan will correspond to N warning time points, forming N derivative sample loans. However, for bad debt loans, since no warning is required after the loan deteriorates, the warning time points after the deterioration date can be deleted. At this time, the sample is a sample with loan number + warning time point as the primary key. In other words, one loan will generate 1 to 3 derivative sample loans.

[0091] For example, Loan A is a normal loan with a loan disbursement date of June 9, 2021. If September 9, 2021, December 9, 2021, and March 9, 2022 are randomly selected as warning time points, Loan A will be directly expanded to 3 derivative sample loans; Loan B is a bad debt loan with a loan disbursement date of June 9, 2021 and a bad debt date of January 9, 2022, then the warning time point of March 9, 2022 is no longer needed, so the derivative sample loans corresponding to Loan B are 2.

[0092] Step 102, labeling each of the derivative sample loans; wherein the label includes positive samples and negative samples.

[0093] The post-loan risk warning model is based on the multi-dimensional data performance of users after lending, and predicts the possibility of the loan becoming a bad debt at each warning time point (usually the end of each month). Based on this goal, it is necessary to define the positive and negative labels of the samples at each warning time point. Therefore, after generating the derivative sample loans, it is necessary to mark the labels of these derivative sample loans one by one, and the labels are positive samples or negative samples.

[0094] Optionally, step 102 may include: for a sample loan of a bad debt, determining whether the time interval between the warning time point of the sample loan and the bad debt date of the sample loan is greater than T months; wherein T is greater than or equal to 2 and less than or equal to 6; if so, marking the derivative sample loan corresponding to the warning time point as a positive sample; if not, marking the derivative sample loan corresponding to the warning time point as a negative sample. In an embodiment of the present invention, the sample loans are divided into good and bad debts. For a sample loan of a bad debt, determining whether the time interval between the warning time point of the sample loan and the bad debt date of the sample loan is greater than T months, if so, marking the derivative sample loan corresponding to the warning time point as a positive sample, and if not, marking the derivative sample loan corresponding to the warning time point as a negative sample. For example, if T is 3, if the time interval between the warning time point of the sample loan and the bad debt date of the sample loan is greater than 3 months, the label of the derivative sample loan corresponding to the warning time point is a positive sample, and only when the time interval between the warning time point and the bad debt date is less than or equal to 3 months, the derivative sample loan corresponding to the warning time point is marked as a negative sample.

[0095] For example, loan B is a bad debt loan, the loan date is June 9, 2021, and the bad debt date is January 9, 2022. Then the label of the derivative sample loan corresponding to the warning time point of September 9, 2021 is a positive sample, and the label of the derivative sample loan corresponding to the warning time point of December 9, 2021 is a negative sample.

[0096] Step 103: training a model based on the derivative sample loans and their corresponding labels, thereby obtaining a post-loan risk warning model.

[0097] The model uses the relevant information of each derivative sample loan (such as user credit, age, occupation, loan amount, etc.) as the input, and the label corresponding to each derivative sample loan as the output of the model for supervised training. The model learns iteratively and finally generates a post-loan risk model.

[0098] Optionally, step 103 may include: randomly extracting derivative sample loans with positive labels and derivative sample loans with negative labels, so that the ratio of the number of derivative sample loans with positive labels to the number of derivative sample loans with negative labels is a preset ratio; using the extracted derivative sample loans and their corresponding labels to train the model, thereby obtaining a post-loan risk warning model. For example, samples are randomly extracted from the derivative sample loans according to a certain positive-negative ratio (5:1 or 10:1) to model the post-loan risk warning model.

[0099] Step 104, calculating the probability of the target loan becoming a bad debt loan at a future warning point through the post-loan risk warning model.

[0100] The relevant information of the target loan (such as user credit, age, occupation, loan amount, etc.) is input into the post-loan risk warning model, and the post-loan risk warning model can output the probability of the loan becoming a bad debt loan at a future warning point.

[0101] Optionally, step 104 may include: calculating the probability of the target loan becoming a bad debt loan at the end of each future month by the post-loan risk warning model. In order to better monitor the risk of the loan, the post-loan risk warning model may also output the probability of the target loan becoming a bad debt loan at the end of each future month.

[0102] According to the various embodiments described above, it can be seen that the embodiment of the present invention solves the technical problem of inaccurate post-loan risk warning results in the prior art by deriving samples of each sample loan based on the warning time point to obtain derived sample loans and marking the labels of each derived sample loan. The embodiment of the present invention fully considers the scene characteristics of post-loan risk warning, derives sample loans by warning time point, and screens out suitable samples and defines suitable labels, so that the post-loan risk warning model constructed by the derived samples can accurately calculate the probability of the loan becoming a bad debt at the future warning time point. Compared with the prior art that only predicts that the loan will eventually become a bad debt, the embodiment of the present invention can reduce the warning ratio each month, thereby reducing the pressure of post-loan risk warning processing.

[0103] Figure 2 Schematic diagram of the main process of a post-loan risk early warning method according to a reference embodiment of the present invention. As another embodiment of the present invention, Figure 2 As shown, the post-loan risk early warning method may include:

[0104] Step 201, screen out loans whose repayment periods have reached the performance period.

[0105] Step 202: for each loan, the starting date after N days of overdue is used as the bad debt date of the loan; wherein N is greater than or equal to 25 and less than or equal to 45, and M is greater than or equal to 2 and less than or equal to 6.

[0106] Step 203, screen out loans with bad debt dates within M months after loan disbursement, thereby screening out sample loans.

[0107] Step 204: for each sample loan, randomly select the end of S months after the loan is issued as the warning time point; where S is greater than or equal to 2.

[0108] Step 205: for each warning time point, generate a derivative sample loan corresponding to the warning time point.

[0109] Step 206: For the sample loans of bad debts, if the warning time point of the sample loans is after the bad debt date of the sample loans, the derivative sample loans corresponding to the warning time point are screened out.

[0110] Step 207, labeling each of the derivative sample loans; wherein the label includes positive samples and negative samples.

[0111] Step 208: training a model based on the derivative sample loans and their corresponding labels, thereby obtaining a post-loan risk warning model.

[0112] Step 209, calculating the probability of the target loan becoming a bad debt loan at a future warning point through the post-loan risk warning model.

[0113] In addition, the specific implementation content of the post-loan risk warning method in a reference embodiment of the present invention has been described in detail in the post-loan risk warning method described above, so the repeated content will not be described again here.

[0114] Figure 3 Schematic diagram of the main process of the post-loan risk early warning method according to another reference embodiment of the present invention. As another embodiment of the present invention, Figure 3 As shown, the post-loan risk early warning method may include:

[0115] Step 301, performing sample derivation on each sample loan based on the early warning time point to obtain a derived sample loan.

[0116] Step 302, for the sample loans that are bad debts, determine whether the time interval between the warning time point of the sample loans and the bad debt date of the sample loans is greater than T months; wherein T is greater than or equal to 2 and less than or equal to 6; if so, execute step 303; if not, execute step 304.

[0117] Step 303: Mark the derivative sample loans corresponding to the warning time point as positive samples.

[0118] Step 304: Mark the derivative sample loans corresponding to the warning time point as negative samples.

[0119] Step 305: training a model based on the derivative sample loans and their corresponding labels, thereby obtaining a post-loan risk warning model.

[0120] Step 306, calculating the probability of the target loan becoming a bad debt loan at a future warning point through the post-loan risk warning model.

[0121] In addition, the specific implementation content of the post-loan risk warning method in another reference embodiment of the present invention has been described in detail in the post-loan risk warning method described above, so the repeated content will not be described again here.

[0122] Figure 4 Schematic diagram of the main process of a post-loan risk early warning method according to a reference embodiment of the present invention. As another embodiment of the present invention, Figure 4 As shown, the post-loan risk early warning method may include:

[0123] Step 401, performing sample derivation on each sample loan based on the early warning time point to obtain a derived sample loan.

[0124] Step 402, labeling each of the derivative sample loans; wherein the label includes positive samples and negative samples.

[0125] Step 403 : randomly extracting derivative sample loans with positive labels and derivative sample loans with negative labels, so that the number ratio of the derivative sample loans with positive labels to the derivative sample loans with negative labels is a preset ratio.

[0126] Step 404, using the extracted derivative sample loans and their corresponding labels to train the model, thereby obtaining a post-loan risk warning model.

[0127] Step 405, calculating the probability of the target loan becoming a bad debt loan at the end of each future month through the post-loan risk warning model.

[0128] In addition, the specific implementation content of the post-loan risk warning method in another reference embodiment of the present invention has been described in detail in the post-loan risk warning method described above, so the repeated content will not be described again here.

[0129] Figure 5 Schematic diagram of the main modules of the post-loan risk early warning device according to an embodiment of the present invention. Figure 5 As shown, the post-loan risk warning device 500 includes a derivation module 501, a marking module 502, a training module 503 and a warning module 504; wherein the derivation module 501 is used to perform sample derivation on each sample loan based on the warning time point to obtain a derived sample loan; the marking module 502 is used to mark the label of each derived sample loan; wherein the label includes a positive sample and a negative sample; the training module 503 is used to train a model according to each derived sample loan and its corresponding label, so as to obtain a post-loan risk warning model; the warning module 504 is used to calculate the probability of the target loan becoming a bad debt loan at the future warning time point through the post-loan risk warning model.

[0130] Optionally, the derivation module 501 is further used for:

[0131] Filter out loans with repayment periods that reach the performance period;

[0132] Calculate the bad debt date of each loan, and select sample loans according to the bad debt date of each loan;

[0133] For each sample loan, a derivative sample loan is generated according to the early warning time point of the sample loan.

[0134] Optionally, the derivation module 501 is further used for:

[0135] For each loan, the starting date after N days of overdue is taken as the bad debt date of the loan;

[0136] Eliminate loans with bad debt dates within M months after loan disbursement, thereby selecting sample loans;

[0137] Among them, N is greater than or equal to 25 and less than or equal to 45, and M is greater than or equal to 2 and less than or equal to 6.

[0138] Optionally, the derivation module 501 is further used for:

[0139] Randomly select the sample loans at the end of S months after the loan is disbursed as the warning time point; where S is greater than or equal to 2;

[0140] Derivative sample loans are generated according to various warning time points of the sample loans.

[0141] Optionally, the derivation module 501 is further used for:

[0142] For each warning time point, generate a derivative sample loan corresponding to the warning time point;

[0143] For a sample loan of bad debt, if the warning time point of the sample loan is after the bad debt date of the sample loan, the derivative sample loan corresponding to the warning time point is screened out.

[0144] Optionally, the marking module 502 is further used for:

[0145] For a sample loan of bad debt, determine whether the time interval between the warning time point of the sample loan and the bad debt date of the sample loan is greater than T months; wherein T is greater than or equal to 2 and less than or equal to 6;

[0146] If yes, the derivative sample loan corresponding to the warning time point is marked as a positive sample;

[0147] If not, the derivative sample loans corresponding to the warning time point are marked as negative samples.

[0148] Optionally, the training module 503 is further used for:

[0149] Randomly extracting derivative sample loans with positive labels and derivative sample loans with negative labels, so that the number ratio of the derivative sample loans with positive labels to the derivative sample loans with negative labels is a preset ratio;

[0150] The model is trained using the extracted derivative sample loans and their corresponding labels to obtain a post-loan risk warning model.

[0151] Optionally, the early warning module 504 is further configured to:

[0152] The post-loan risk early warning model is used to calculate the probability that the target loan will become a bad debt loan at the end of each future month.

[0153] According to the various embodiments described above, it can be seen that the embodiment of the present invention solves the technical problem of inaccurate post-loan risk warning results in the prior art by deriving samples of each sample loan based on the warning time point to obtain derived sample loans and marking the labels of each derived sample loan. The embodiment of the present invention fully considers the scene characteristics of post-loan risk warning, derives sample loans by warning time point, and screens out suitable samples and defines suitable labels, so that the post-loan risk warning model constructed by the derived samples can accurately calculate the probability of the loan becoming a bad debt at the future warning time point. Compared with the prior art that only predicts that the loan will eventually become a bad debt, the embodiment of the present invention can reduce the warning ratio each month, thereby reducing the pressure of post-loan risk warning processing.

[0154] It should be noted that the specific implementation content of the post-loan risk warning device of the present invention has been described in detail in the above-mentioned post-loan risk warning method, so the repeated content will not be described again here.

[0155] Figure 6 An exemplary system architecture 600 is shown to which the post-loan risk early warning method or post-loan risk early warning device according to the embodiment of the present invention can be applied.

[0156] like Figure 6 As shown, system architecture 600 may include terminal devices 601, 602, 603, network 604 and server 605. Network 604 is used to provide a medium for communication links between terminal devices 601, 602, 603 and server 605. Network 604 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0157] Users can use terminal devices 601, 602, and 603 to interact with server 605 through network 604 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 601, 602, and 603, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only examples).

[0158] The terminal devices 601 , 602 , and 603 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers, etc.

[0159] The server 605 may be a server that provides various services, such as a backend management server (for example only) that provides support for shopping websites browsed by users using the terminal devices 601, 602, and 603. The backend management server may analyze and process the received data such as the item information query request, and feed back the processing results to the terminal device.

[0160] It should be noted that the post-loan risk warning method provided in the embodiment of the present invention is generally executed by the server 605, and accordingly, the post-loan risk warning device is generally set in the server 605. The post-loan risk warning method provided in the embodiment of the present invention can also be executed by the terminal devices 601, 602, and 603, and accordingly, the post-loan risk warning device can be set in the terminal devices 601, 602, and 603.

[0161] It should be understood that Figure 6 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0162] Reference below Figure 7 , which shows a schematic diagram of the structure of a computer system 700 of a terminal device suitable for implementing an embodiment of the present invention. Figure 7 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0163] like Figure 7 As shown, the computer system 700 includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 702 or a program loaded from a storage part 708 into a random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the system 700 are also stored. The CPU 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0164] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, a mouse, etc.; an output section 707 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN card, a modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 710 as needed, so that a computer program read therefrom is installed into the storage section 708 as needed.

[0165] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 709, and / or installed from the removable medium 711. When the computer program is executed by the central processing unit (CPU) 701, the above-mentioned functions defined in the system of the present invention are executed.

[0166] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0167] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operations of the system, method and computer program according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0168] The modules involved in the embodiments of the present invention may be implemented by software or hardware. The modules described may also be set in a processor, for example, it may be described as: a processor includes a derivative module, a marking module, a training module and an early warning module, wherein the names of these modules do not constitute a limitation on the modules themselves in some cases.

[0169] As another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiment; or it may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by a device, the device implements the following method: performing sample derivation on each sample loan based on the warning time point to obtain a derived sample loan; marking the labels of each derived sample loan; wherein the labels include positive samples and negative samples; training a model based on each derived sample loan and its corresponding labels to obtain a post-loan risk warning model; and calculating the probability of the target loan becoming a bad debt loan at the future warning time point through the post-loan risk warning model.

[0170] According to the technical solution of the embodiment of the present invention, because the technical means of deriving samples of each sample loan based on the warning time point, obtaining derived sample loans, and marking the labels of each derived sample loan are adopted, the technical problem of inaccurate post-loan risk warning results in the prior art is overcome. The embodiment of the present invention fully considers the scene characteristics of post-loan risk warning, derives sample loans through the warning time point, and screens out suitable samples and defines suitable labels, so that the post-loan risk warning model constructed by the derived samples can accurately calculate the probability of the loan becoming a bad debt at the future warning time point. Compared with the prior art that only predicts that the loan will eventually become a bad debt, the embodiment of the present invention can reduce the warning ratio each month, thereby reducing the pressure of post-loan risk warning processing.

[0171] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A post-loan risk early warning method, characterized in that: include: Filter out loans with repayment periods that reach the performance period; Calculate the bad debt date of each loan, and select sample loans according to the bad debt date of each loan; For each sample loan, randomly select the sample loan at the end of S months after the loan is disbursed as the warning time point, and generate a derivative sample loan according to each warning time point of the sample loan; where S is greater than or equal to 2; Labeling each of the derivative sample loans; wherein the labels include positive samples and negative samples; A model is trained based on each of the derivative sample loans and their corresponding labels, thereby obtaining a post-loan risk warning model; Calculate the probability of the target loan becoming a bad debt loan at the future warning time point through the post-loan risk warning model; Generate derivative sample loans based on each warning time point of the sample loans, including: For each warning time point, generate a derivative sample loan corresponding to the warning time point; For a sample loan of bad debt, if the warning time point of the sample loan is after the bad debt date of the sample loan, the derivative sample loan corresponding to the warning time point is screened out.

2. The method according to claim 1, characterized in that Calculate the bad debt date of each loan, and select sample loans according to the bad debt date of each loan, including: For each loan, the starting date after N days of overdue is taken as the bad debt date of the loan; Eliminate loans with bad debt dates within M months after loan disbursement, thereby selecting sample loans; Among them, N is greater than or equal to 25 and less than or equal to 45, and M is greater than or equal to 2 and less than or equal to 6.

3. The method according to claim 1, characterized in that Labels for marking each of the sample derivative loans include: For a sample loan of bad debt, determine whether the time interval between the warning time point of the sample loan and the bad debt date of the sample loan is greater than T months; wherein T is greater than or equal to 2 and less than or equal to 6; If yes, the derivative sample loan corresponding to the warning time point is marked as a positive sample; If not, the derivative sample loans corresponding to the warning time point are marked as negative samples.

4. The method according to claim 1, characterized in that: The model is trained based on the derivative sample loans and their corresponding labels, thereby obtaining a post-loan risk warning model, including: Randomly extracting derivative sample loans with positive labels and derivative sample loans with negative labels, so that the number ratio of the derivative sample loans with positive labels to the derivative sample loans with negative labels is a preset ratio; The model is trained using the extracted derivative sample loans and their corresponding labels to obtain a post-loan risk warning model.

5. The method according to claim 1, characterized in that The probability of the target loan becoming a bad debt loan at the future warning time point is calculated by the post-loan risk warning model, including: The post-loan risk early warning model is used to calculate the probability that the target loan will become a bad debt loan at the end of each future month.

6. A post-loan risk early warning device, characterized in that: include: The derivative module is used to filter out loans whose repayment periods have reached the performance period; Calculate the bad debt date of each loan, and select sample loans according to the bad debt date of each loan; for each sample loan, randomly select the end of S months after the loan is disbursed as the warning time point, and generate derivative sample loans according to each warning time point of the sample loan; where S is greater than or equal to 2; A labeling module, used to label each of the derivative sample loans; wherein the label includes positive samples and negative samples; A training module, used for training a model based on each of the derivative sample loans and their corresponding labels, so as to obtain a post-loan risk warning model; An early warning module, used to calculate the probability of the target loan becoming a bad debt loan at a future early warning time point through the post-loan risk early warning model; The derivative module is also used to generate, for each warning time point, a derivative sample loan corresponding to the warning time point; for a sample loan of a bad debt, if the warning time point of the sample loan is after the bad debt date of the sample loan, the derivative sample loan corresponding to the warning time point is screened out.

7. The device according to claim 6, characterized in that The derivative module is also used to: For each loan, the starting date after N days of overdue is taken as the bad debt date of the loan; Eliminate loans with bad debt dates within M months after loan disbursement, thereby selecting sample loans; Among them, N is greater than or equal to 25 and less than or equal to 45, and M is greater than or equal to 2 and less than or equal to 6.

8. The device according to claim 6, characterized in that The marking module is also used for: For a sample loan of bad debt, determine whether the time interval between the warning time point of the sample loan and the bad debt date of the sample loan is greater than T months; wherein T is greater than or equal to 2 and less than or equal to 6; If yes, the derivative sample loan corresponding to the warning time point is marked as a positive sample; If not, the derivative sample loans corresponding to the warning time point are marked as negative samples.

9. The device according to claim 6, characterized in that The training module is also used to: Randomly extracting derivative sample loans with positive labels and derivative sample loans with negative labels, so that the number ratio of the derivative sample loans with positive labels to the derivative sample loans with negative labels is a preset ratio; The model is trained using the extracted derivative sample loans and their corresponding labels to obtain a post-loan risk warning model.

10. The device according to claim 6, characterized in that The early warning module is also used for: The post-loan risk early warning model is used to calculate the probability that the target loan will become a bad debt loan at the end of each future month.

11. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 5.

12. A computer readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images