Method and device for determining overdue condition of customer group

Abstract

The invention discloses a method and a device for determining an overdue condition of a customer group. The method comprises the following steps: acquiring respective transaction characteristics of a plurality of target customers belonging to the same customer group in a kth repayment period; for at least one target customer, respectively inputting the transaction characteristics of the target customer in the kth repayment period into the first model and the second model, and determining a first overdue transition probability and a second overdue transition probability; according to the first overdue transition probability and the second overdue transition probability, determining a state transition probability of the target customer in the kth repayment period; according to the state transition probability of the target customer in the kth repayment period and the overdue probability of the target customer in the (k-1) th repayment period, determining the overdue probability of the target customer in the kth repayment period; and for a plurality of target customers belonging to the same customer group, according to the overdue probability of the plurality of target customers in the kth repayment period, determining the overdue condition of the customer group to which the plurality of target customers belong in the kth repayment period. By adopting the method, the accuracy of determining the overdue condition of the customer group can be improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method and device for determining the overdue status of a customer group. Background Art

[0002] Installment payment is a common credit method in the financial industry, which means that consumers purchase a product by paying it back in installments. With the popularity of installment payment, the overdue status of each repayment period has become an increasingly important indicator.

[0003] The common calculation of the overdue status of the target customer group in a certain repayment period or several repayment periods is usually obtained based on the overdue status of other customer groups in history. For example, if the overdue status of the target customer group in the first repayment period is calculated, another customer group similar to the target customer group can be determined, and the overdue status of the first repayment period of the other customer group can be used as the overdue status of the target customer group in the first repayment period; or multiple other customer groups similar to the target customer group can be determined, and the overdue status of the first repayment period of multiple similar other customer groups can be weighted to obtain the overdue status of the target customer group in the first repayment period. However, this estimation method is often not accurate enough.

[0004] Therefore, there is an urgent need for a more accurate method to determine the overdue status of customer groups. Summary of the invention

[0005] The present application provides a method and device for determining the overdue status of a customer group, which are used to improve the accuracy of determining the overdue status of a customer group.

[0006] In a first aspect, an embodiment of the present application provides a method for determining the overdue status of a customer group, which can be executed by a device for determining the overdue status of a customer group, and the method includes: obtaining transaction characteristics of multiple target customers belonging to the same customer group in the kth repayment period; k is a positive integer; for at least one target customer, inputting the transaction characteristics of the target customer in the kth repayment period into a first model and a second model respectively, and determining a first overdue transition probability output by the first model and a second overdue transition probability output by the second model, wherein the first overdue transition probability represents the probability that the target customer transfers from non-overdue to overdue, and the second overdue transition probability represents the probability that the target customer remains overdue from overdue; the first model and the second model are based on multiple samples. The target customer's transaction characteristics are trained; the state transition probability of the target customer in the kth repayment period is determined according to the first overdue transition probability and the second overdue transition probability, and the state transition probability represents the probability of the first state transitioning to the second state, and the first state and the second state are either overdue or not overdue; the overdue probability of the target customer in the kth repayment period is determined according to the state transition probability of the target customer in the kth repayment period and the overdue probability of the k-1th repayment period, and the overdue probability represents the probability of not repaying on time; for multiple target customers belonging to the same customer group, the overdue status of the customer group to which the multiple target customers belong in the kth repayment period is determined according to the overdue probabilities of the multiple target customers in the kth repayment period.

[0007] By adopting the above method, the overdue transfer probability is predicted by the trained first model and the second model to obtain the state transfer probability, and the overdue probability of the kth repayment period is determined according to the state transfer probability of the kth repayment period and the overdue probability of the k-1th repayment period. In this way, since the first model and the second model are obtained through training with a large number of training samples, the accuracy of the overdue transfer probability output by the first model and the second model is relatively high, which can improve the accuracy of determining the overdue status of the customer group. In addition, the operation of determining the overdue status of the customer group by combining the model is relatively simple, which saves manpower.

[0008] In a possible implementation, transaction features of the kth repayment period of multiple target customers belonging to the same customer group are obtained, including: obtaining transaction information of transactions that have occurred when the multiple target customers belonging to the same customer group are currently in the Mth repayment period; M is a positive integer, and the Mth repayment period is earlier than the kth repayment period; and using the transaction information of the transactions that have occurred and the kth repayment period as transaction features of the kth repayment period.

[0009] By adopting the above method, the transaction information of the transaction is obtained at the time of transaction completion. The transaction characteristics of each repayment period can be determined based on the transaction information and the number of repayment periods. In this way, the overdue status of each repayment period in the complete life cycle of the transaction can be obtained at the time of transaction completion, which can provide early warning and accurately estimate the losses within a certain period of time in the future, which is of great significance for determining the necessary loss reserves and maintaining financial stability. It is also possible to monitor the fluctuations of the economic cycle and changes in the market environment by estimating the overdue status of the complete life cycle, so as to better adapt to the overall economic environment.

[0010] In a possible implementation, the first model and the second model are trained based on transaction characteristics of multiple sample customers, including: for the number of repayment periods for each transaction of each sample customer, obtaining a training sample corresponding to each repayment period, each training sample including a transaction feature and an overdue status corresponding to the transaction feature; dividing the multiple training samples into a first category of training samples and a second category of training samples based on the overdue status of each training sample in the previous repayment period; the overdue status corresponding to the sample customer in the first training sample in the previous repayment period is not overdue; the overdue status corresponding to the sample customer in the second training sample in the previous repayment period is overdue; the first model is obtained by training based on the first category of training samples, and the second model is obtained by training based on the second category of training samples.

[0011] Using the above approach, the accuracy of both models can be improved by training them separately.

[0012] In one possible implementation, the transaction features of each training sample include transaction information of transactions that have occurred up to the last repayment period corresponding to the training sample and the repayment period corresponding to the training sample; or the transaction features of each training sample include transaction information of transactions that have occurred when the loan corresponding to the training sample takes effect and the repayment period corresponding to the training sample.

[0013] In a possible implementation, the state transition probability is a 2*2 state transition matrix; the state transition matrix is

[0014]

[0015] in, represents the first non-overdue transfer probability of the i-th target customer in the k-th repayment period, represents the first overdue transfer probability of the i-th target customer in the k-th repayment period, represents the second non-overdue transfer probability of the i-th target customer in the k-th repayment period, represents the second overdue transfer probability of the i-th target customer in the k-th repayment period.

[0016] In a possible implementation, the overdue probability of the target customer in the kth repayment period is determined based on the state transition probability of the target customer in the kth repayment period and the overdue probability in the k-1th repayment period, including: for the target customer, determining a first product based on the non-overdue probability in the k-1th repayment period and the first overdue transition probability; determining a second product based on the overdue probability in the k-1th repayment period and the second overdue transition probability; determining the overdue probability of the kth repayment period based on the first product and the second product, the sum of the non-overdue probability in the k-1th repayment period and the overdue probability in the k-1th repayment period is 1.

[0017] In one possible implementation, the overdue status is an overdue rate; based on the overdue probabilities of the multiple target customers in the kth repayment period, the overdue status of the customer group to which the multiple target customers belong in the kth repayment period is determined, including: based on the overdue probabilities of the multiple target customers in the kth repayment period and the loan amounts of the multiple target customers, the overdue rate of the customer group to which the multiple target customers belong in the kth repayment period is determined.

[0018] In a second aspect, an embodiment of the present application provides a device for determining the overdue status of a customer group, the device comprising: an acquisition module for acquiring transaction characteristics of multiple target customers belonging to the same customer group in the kth repayment period; k is a positive integer; a processing module for inputting the transaction characteristics of the target customer in the kth repayment period into a first model and a second model respectively for at least one target customer, and determining a first overdue transition probability output by the first model and a second overdue transition probability output by the second model, the first overdue transition probability indicating the probability that the target customer transfers from non-overdue to overdue, and the second overdue transition probability indicating the probability that the target customer remains overdue from overdue; the first model and the second model are trained based on the transaction characteristics of multiple sample customers; A determination module is used to determine the state transition probability of the target customer in the kth repayment period according to the first overdue transition probability and the second overdue transition probability, the state transition probability indicating the probability of the first state transitioning to the second state, the first state and the second state being either overdue or not overdue; and, based on the state transition probability of the target customer in the kth repayment period and the overdue probability of the k-1th repayment period, determine the overdue probability of the target customer in the kth repayment period, the overdue probability indicating the probability of not repaying on time; the determination module is also used to determine, for multiple target customers belonging to the same customer group, the overdue status of the customer group to which the multiple target customers belong in the kth repayment period according to the overdue probabilities of the multiple target customers in the kth repayment period.

[0019] In a possible implementation, the acquisition module is specifically used to obtain transaction information of transactions that have occurred when multiple target customers belonging to the same customer group are currently in the Mth repayment period; M is a positive integer, and the Mth repayment period is earlier than the kth repayment period; the transaction information of the transactions that have occurred and the kth repayment period are used as transaction features of the kth repayment period.

[0020] In a possible implementation, the acquisition module is also used to obtain, for each repayment period of each transaction in each sample customer, a training sample corresponding to each repayment period, each training sample including a transaction feature and an overdue status corresponding to the transaction feature; the processing module is also used to divide the multiple training samples into a first category of training samples and a second category of training samples according to the overdue status of each training sample in the previous repayment period; the overdue status corresponding to the sample customer in the first training sample in the previous repayment period is not overdue; the overdue status corresponding to the sample customer in the second training sample in the previous repayment period is overdue; and, the first model is obtained by training based on the first category of training samples, and the second model is obtained by training based on the second category of training samples.

[0021] In a possible implementation method, the determination module is specifically used to, for the target customer, determine a first product based on the non-overdue probability of the k-1th repayment period and the first overdue transfer probability; determine a second product based on the overdue probability of the k-1th repayment period and the second overdue transfer probability; determine the overdue probability of the kth repayment period based on the first product and the second product, and the sum of the non-overdue probability of the k-1th repayment period and the overdue probability of the k-1th repayment period is 1.

[0022] In a possible implementation method, the determination module is further used to determine the overdue rate of the customer group to which the multiple target customers belong in the kth repayment period based on the overdue probability of the multiple target customers in the kth repayment period and the loan amount of the multiple target customers.

[0023] In a third aspect, an embodiment of the present application further provides a device for determining the overdue status of a customer group, the device comprising a memory and a processor, the memory being used to store computer programs or instructions; the processor being used to call the computer programs or instructions stored in the memory to execute a method as in any possible implementation of the first aspect.

[0024] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which instructions are stored. When a computer reads and executes the instructions, the computer executes a method in any possible implementation of the first aspect.

[0025] In a fifth aspect, an embodiment of the present application provides a computer program product, wherein instructions are stored in the computer program product. When a computer reads and executes the instructions, the computer executes the method in any possible implementation of the first aspect above. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0027] Figure 1 A flowchart corresponding to a method for determining the overdue status of a customer group provided in an embodiment of the present application;

[0028] Figure 2 A flowchart corresponding to a method for training a first model and a second model provided in an embodiment of the present application;

[0029] Figure 3 A schematic diagram of constructing a training sample by sliding a variable-length window provided in an embodiment of the present application;

[0030] Figure 4 A process framework diagram for training a first model and a second model provided in an embodiment of the present application;

[0031] Figure 5 A schematic diagram of internal modules of a device 5000 for determining the overdue status of a customer group provided in an embodiment of the present application;

[0032] Figure 6 A structural diagram of a device 6000 for determining the overdue status of a customer group provided in an embodiment of the present application. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0034] With the development of the financial industry, installment payment has gradually become an increasingly common payment method. Users can obtain goods in advance through installment payment, but at the same time, they may be overdue when repaying in each repayment period. The existing method of determining the overdue status of the target customer group is usually calculated based on the overdue status of other similar customer groups, but this method often lacks accuracy.

[0035] Based on this, an embodiment of the present application provides a method for determining the overdue status of a customer group, which is used to improve the accuracy of determining the overdue status of a customer group.

[0036] Figure 1 The flowchart of the method for determining the overdue status of a customer group provided in the embodiment of the present application can be executed by a device for determining the overdue status of a customer group. Figure 1 As shown, the process includes the following steps:

[0037] Step 101, obtaining transaction characteristics of multiple target customers belonging to the same customer group in the kth repayment period.

[0038] For example, according to actual business needs, it is often necessary to obtain the overdue status of a customer group, so multiple target customers are classified into the same customer group. The transaction characteristics of multiple target customers in the same customer group in the kth repayment period are obtained, where k is a positive integer greater than or equal to 1 and less than or equal to the total number of repayment periods. Optionally, after the transaction time, every first time period is the next repayment period, and the first time period can be one month or other time lengths.

[0039] Obtaining transaction features of the kth repayment period of multiple target customers belonging to the same customer group, including: obtaining transaction information of transactions that have occurred when the multiple target customers belonging to the same customer group are currently in the Mth repayment period, M is a positive integer, and the Mth repayment period is earlier than the kth repayment period; using the transaction information of the transactions that have occurred and the kth repayment period as transaction features of the kth repayment period.

[0040] For example, multiple target customers of the same customer group have a transaction with a repayment period of five periods. Multiple target customers are currently in the second repayment period. To obtain the transaction characteristics of the third repayment period of each of the multiple target customers, the transaction information of the transactions that have occurred when the multiple target customers are currently in the second repayment period can be obtained, and the transaction information of the transactions that have occurred in the second repayment period and the repayment period three can be used as the transaction characteristics of the third repayment period. For another example, multiple target customers can obtain the transaction characteristics of the first repayment period when the transaction is completed. In order to estimate the overdue status of each repayment period of multiple target customers from the first repayment period to the last repayment period, it is necessary to obtain the transaction characteristics of each repayment period. The transaction information of the first repayment period and the number of repayment periods to be estimated can be used as the transaction characteristics of the number of repayment periods to be estimated. In this way, the overdue status of multiple target customers in the entire life cycle can be estimated at the time of the transaction. The complete life cycle is the transaction time to the last repayment period of the transaction.

[0041] Transaction features include transaction information and repayment period. Transaction information includes customer information and transaction information. For example, customer information includes whether the customer is a new or old customer, the customer's name, gender, age, monthly income, historical purchase information, customer's initial rating, historical approval status, etc., which can describe the customer's information in various dimensions. The customer's initial rating is a rating of the customer based on the customer's historical repayment situation. The customer's initial rating can be directly obtained from the database. The historical approval status refers to whether the customer's historical purchase of goods was approved normally. Transaction information includes the number of periods for purchasing products and the amount of the loan.

[0042] Step 102, for at least one target customer, input the transaction characteristics of the target customer in the kth repayment period into the first model and the second model respectively, and determine a first overdue transition probability output by the first model and a second overdue transition probability output by the second model.

[0043] For at least one target customer in the same customer group, the transaction characteristics of the target customer in the kth repayment period are respectively input into the trained first model and the second model, and the first model outputs the first overdue transfer probability M1_Prob, and the second model outputs the second overdue transfer probability M2_Prob. Among them, the first overdue transfer probability represents the probability that the target customer transfers from non-overdue to overdue, and the second overdue transfer probability represents the probability that the target customer remains overdue from overdue. The first model and the second model are obtained by training based on the transaction characteristics of multiple sample customers, and the training process of the first model and the second model is described as follows.

[0044] Step 103, determining the state transition probability of the target customer in the kth repayment period according to the first overdue transition probability and the second overdue transition probability.

[0045] Specifically, the state transition probability of the target customer in the kth repayment period is determined based on the first overdue transition probability output by the first model and the second overdue transition probability output by the second model. The state transition probability represents the probability of the first state transitioning to the second state, and the first state and the second state are either overdue or not overdue. The state transition probability includes the first overdue transition probability and the second overdue transition probability. Optionally, the state transition probability can be a 2*2 state transition matrix, and the state transition matrix is

[0046]

[0047] in, represents the first non-overdue transfer probability of the i-th target customer in the k-th repayment period, represents the first overdue transfer probability of the i-th target customer in the k-th repayment period, represents the second non-overdue transfer probability of the i-th target customer in the k-th repayment period, It represents the second overdue transfer probability of the i-th target customer in the k-th repayment period. The sum of the first non-overdue transfer probability and the first overdue transfer probability is 1, and the sum of the second non-overdue transfer probability and the second overdue transfer probability is 1.

[0048] Step 104, determining the overdue probability of the target customer in the kth repayment period according to the state transition probability of the target customer in the kth repayment period and the overdue probability of the k-1th repayment period.

[0049] Specifically, the overdue probability of the target customer in the kth repayment period is determined based on the state transition probability of the target customer in the kth repayment period and the overdue probability of the k-1th repayment period, combined with a preset calculation method, and the preset calculation method is not repeated here. Optionally, the overdue probability of the target customer in the kth repayment period is determined based on the state transition probability of the target customer in the kth repayment period and the overdue probability of the k-1th repayment period, including: determining a first product based on the non-overdue probability of the k-1th repayment period and the first overdue transition probability; determining a second product based on the overdue probability of the k-1th repayment period and the second overdue transition probability; determining the overdue probability of the kth repayment period based on the first product and the second product, and the sum of the non-overdue probability of the k-1th repayment period and the overdue probability of the k-1th repayment period is 1. It should be noted that the embodiments of the present application do not limit the order of determining the first product and the second product. Similarly, the overdue probability of the target customer in the k-1th repayment period is determined based on the state transition probability of the target customer in the k-1th repayment period and the overdue probability in the k-2th repayment period.

[0050] For example, the state transition matrix of target customer i in the first repayment period is: represents the second non-overdue transfer probability of target customer i in the first repayment period. The second non-overdue transfer probability represents the probability of transferring from overdue to non-overdue. Since there is no overdue payment in the first repayment period, The default value is 0. The default is 1, so

[0051] According to the state transition probability of the target customer in the first repayment period and the overdue probability in the zeroth repayment period, the overdue probability of the target customer in the first repayment period is determined. At the time of the transaction, the target customer has not yet made a repayment. Therefore, at the zeroth repayment period, i.e., the transaction time, the target customer must be in a non-overdue state, and the probability of non-overdue is 1, i.e., P(Y i (0) =0) = 1; the probability of overdue is 0, that is, P(Y i (0) =1) = 0. The overdue probability distribution of target customer i at the transaction time It should be noted that the overdue probability distribution S of the kth repayment period i (k) The probability of overdue payment on the kth repayment period P(Y i (k) =1) and the probability that the kth repayment period is not overdue P(Y i (k) =0), the probability of overdue is P(Y i (k) =1) and the probability of not being overdue P(Y i (k) =0) is 1, and the probability of overdue indicates the probability of not repaying on time. The probability distribution of overdue of target customer i at the transaction time and the state transfer matrix T of the first repayment period i (1) Multiply them together to get the predicted overdue probability distribution S of target customer i in the first repayment period i (1) , the calculation process is:

[0052]

[0053] The overdue probability distribution S of target customer i in the second repayment period i (2) The calculation process is:

[0054]

[0055] Similarly, the overdue probability distribution of target customer i in the kth repayment period is The calculation process is:

[0056]

[0057]

[0058] From this, we can deduce the overdue probability distribution of target customer i in the kth repayment period

[0059]

[0060] about The reasons are:

[0061] The transition probability from not overdue to overdue is: Not overdue->Not overdue transfer probability: Overdue->Overdue transition probability: Overdue->Not overdue transition probability:

[0062]

[0063]

[0064]

[0065]

[0066] The overdue status Y of target customer i in the kth repayment period i (k) The predicted probability expression is converted into matrix form:

[0067]

[0068] Here we define (P(Y i (k) =0), P(Y i (k) =1)) is S i (k) , represents the state probability distribution of target customer i in the kth repayment period. Definition:

[0069] T i (k) , represents the state transition matrix of target customer i in the kth repayment period. Therefore, the above formula is simplified to:

[0070]

[0071] Step 105, for multiple target customers belonging to the same customer group, according to the overdue probabilities of the multiple target customers in the kth repayment period, determine the overdue status of the customer group to which the multiple target customers belong in the kth repayment period.

[0072] Exemplarily, the overdue status may be an overdue rate or an overdue probability distribution. When the overdue status is the overdue rate, the overdue status of the customer group to which the multiple target customers belong in the k repayment period is determined based on the overdue probabilities of the multiple target customers in the k repayment period, including: determining the overdue rate of the customer group to which the multiple target customers belong in the k repayment period based on the overdue probabilities of the multiple target customers in the k repayment period and the loan amounts of the multiple target customers.

[0073] The overdue rate of multiple target customers in the kth repayment period The mathematical expectation expression of is as follows:

[0074]

[0075] In the formula, n represents the number of target customers in the customer group, m i represents the loan amount of target customer i,

[0076] P(Y i (k) =1) represents the probability of overdue payment of target customer i in the kth repayment period, P(Y i (k) =1) can be output through a model Approximately, Represents the features of the model input when predicting the overdue status of target customer i in the kth repayment period.

[0077] By adopting the above method, based on the transaction characteristics obtained by target customer i at the time of transaction, the transaction characteristics of target customer i in each repayment period in the complete life cycle can be obtained; the transaction characteristics of target customer i in each repayment period in the complete life cycle are input into the first model and the second model to obtain the first overdue transfer probability and the second overdue transfer probability of target customer i in each repayment period, and then the state transfer matrix in each repayment period is obtained. According to the overdue probability distribution of target customer i at the time of transaction and the state transfer matrix of each repayment period, the overdue probability distribution of target customer i in each repayment period is obtained, and according to the overdue probability distribution of target customer i in each repayment period, the overdue status of target customer i in each repayment period is determined, and then according to the overdue status of multiple target customers in each repayment period, the overdue status of the customer groups to which multiple target customers belong in each repayment period can be obtained.

[0078] The first model and the second model are trained according to transaction characteristics of multiple sample customers. Optionally, the first model and the second model are both trained according to a lightweight gradient boosting machine (Light Gradient Boosting Machine, LightGbm). Figure 2A flowchart corresponding to a method for training a first model and a second model provided in an embodiment of the present application, the process includes the following steps:

[0079] Step 201, for each repayment period of each transaction in each sample customer, obtain a training sample corresponding to each repayment period, each training sample including a transaction feature and an overdue status corresponding to the transaction feature.

[0080] Exemplarily, if the sample customer's transaction has five repayment periods, five training samples can be obtained based on the five repayment periods of the transaction, and each training sample includes a transaction feature and an overdue status corresponding to the transaction feature. Optionally, the transaction feature of each training sample includes transaction information that has occurred up to the last repayment period corresponding to the training sample and the repayment period corresponding to the training sample. The overdue status corresponding to the transaction feature of each training sample can be an overdue status m days after the repayment period, where m is an integer less than the number of days between two adjacent repayment periods. Figure 3 A schematic diagram of constructing training samples by sliding a variable-length window is provided in an embodiment of the present application. Figure 3 As shown, Figure 3 Take a transaction with two repayment periods as an example. Two training samples can be obtained based on this transaction. The transaction features of the first training sample are the transaction information of transactions that have occurred up to the transaction time and the repayment period of 1. The overdue status corresponding to the transaction features of the first training sample is the overdue status 5 days after the first repayment period. 5 days is just an example. It can also be other days. The selection of days depends on the scenario required for modeling, which means any day between the first repayment period and the second repayment period. The transaction features of the second training sample are the transaction information of transactions that have occurred up to the first repayment period and the repayment period of 2. The information of the target customer may change between the transaction time and the first repayment period. Therefore, the latest transaction features of the target customer are obtained.

[0081] Step 202: divide the multiple training samples into a first category of training samples and a second category of training samples according to the overdue status of each training sample in the previous repayment period.

[0082] For example, the overdue status of the sample customer in the first training sample in the previous repayment period is not overdue; the overdue status of the sample customer in the second training sample in the previous repayment period is overdue. Figure 3 Taking the example shown as an example, there are two training samples in total. The overdue status of the first training sample in the previous repayment period, i.e., the transaction time, is not overdue, then the first training sample is divided into the first training sample; if the overdue status of the first repayment period is overdue, that is, the overdue status of the second training sample in the upper repayment period is overdue, then the second training sample is divided into the second training sample.

[0083] Step 203: training a first model based on the first type of training samples, and training a second model based on the second type of training samples.

[0084] Exemplarily, the first training sample is input into the first model, and the second training sample is input into the second model. Figure 4 A flowchart of a training process of a first model and a second model provided in an embodiment of the present application. Figure 4 As shown, the first model and the second model are used to train the training samples to obtain the overdue transition probability of each training sample, and then the four state transition probabilities of not overdue->overdue, not overdue->not overdue, overdue->overdue, and overdue->not overdue are obtained.

[0085] An out-of-sample test (OOS) is used. 20% of the samples of each month are randomly sampled as test samples, and 80% are used as training samples. The reason for using OOS testing here is that the environment, strategies, models, etc. are constantly adjusted every month. Using the latest training samples to train the model can train the model more accurately.

[0086] In the embodiment of the present application, the training samples used in the training of the first model and the second model, the transaction information of each training sample is the transaction information of the transaction that has occurred as of the last repayment period. When the model is used to predict the overdue status of the transaction, the transaction characteristics of each repayment period in the complete life cycle are determined based on the transaction information obtained at the time of the transaction. Optionally, when the transaction information of each training sample used in the training of the first model and the second model is the transaction information of the transaction that has occurred as of the last repayment period, in this training case, when the model is used to predict the overdue status of the transaction, the overdue status of the repayment period to be predicted can be predicted based on the transaction information of the transaction that has occurred as of the last repayment period and the number of repayment periods to be predicted as transaction characteristics. Optionally, the transaction characteristics of each training sample used in the training of the first model and the second model can include the transaction information of the transaction that has occurred when the loan corresponding to the training sample takes effect and the repayment period corresponding to the training sample. In this training case, when the model is used to predict the overdue status of the transaction, the transaction characteristics of each repayment period in the complete life cycle are determined based on the transaction information obtained at the time of the transaction.

[0087] Figure 5 Schematic diagram of the internal modules of a device 5000 for determining the overdue status of a customer group provided in an embodiment of the present application. Figure 5 As shown, the device may include: an acquisition module 501, a processing module 502 and a determination module 503, and optionally, a storage module, the storage module is used to store computer instructions or programs, and the processing module 502 can call the computer instructions or programs in the storage module.

[0088] Among them, the acquisition module 501 is used to obtain the transaction characteristics of multiple target customers belonging to the same customer group in the kth repayment period; k is a positive integer; the processing module 502 is used to input the transaction characteristics of the target customer in the kth repayment period into the first model and the second model respectively for at least one target customer, and determine the first overdue transition probability output by the first model and the second overdue transition probability output by the second model, the first overdue transition probability represents the probability that the target customer transfers from non-overdue to overdue, and the second overdue transition probability represents the probability that the target customer remains overdue from overdue; the first model and the second model are obtained by training based on the transaction characteristics of multiple sample customers; the determination module 503 is used to determine the first overdue transition probability output by the first model and the second overdue transition probability output by the second model according to the first The overdue transfer probability and the second overdue transfer probability are used to determine the state transfer probability of the target customer in the kth repayment period, and the state transfer probability represents the probability of the first state transferring to the second state, and the first state and the second state are either overdue or not overdue; and, based on the state transfer probability of the target customer in the kth repayment period and the overdue probability of the k-1th repayment period, the overdue probability of the target customer in the kth repayment period is determined, and the overdue probability represents the probability of not repaying on time; the determination module 503 is also used to determine, for multiple target customers belonging to the same customer group, the overdue status of the customer group to which the multiple target customers belong in the kth repayment period according to the overdue probabilities of the multiple target customers in the kth repayment period.

[0089] In a possible implementation, the acquisition module 501 is specifically used to obtain transaction information of transactions that have occurred when multiple target customers belonging to the same customer group are currently in the Mth repayment period; M is a positive integer, and the Mth repayment period is earlier than the kth repayment period; the transaction information of the transactions that have occurred and the kth repayment period are used as transaction features of the kth repayment period.

[0090] In a possible implementation, the acquisition module 501 is also used to obtain, for each repayment period of each transaction in each sample customer, a training sample corresponding to each repayment period, each training sample including a transaction feature and an overdue status corresponding to the transaction feature; the processing module 502 is also used to divide the multiple training samples into a first category of training samples and a second category of training samples according to the overdue status of each training sample in the previous repayment period; the overdue status corresponding to the sample customer in the first training sample in the previous repayment period is not overdue; the overdue status corresponding to the sample customer in the second training sample in the previous repayment period is overdue; and, the first model is obtained by training based on the first category of training samples, and the second model is obtained by training based on the second category of training samples.

[0091] In a possible implementation method, the determination module 503 is specifically used to, for the target customer, determine a first product based on the non-overdue probability of the k-1th repayment period and the first overdue transfer probability; determine a second product based on the overdue probability of the k-1th repayment period and the second overdue transfer probability; determine the overdue probability of the kth repayment period based on the first product and the second product, and the sum of the non-overdue probability of the k-1th repayment period and the overdue probability of the k-1th repayment period is 1.

[0092] In a possible implementation method, the determination module 503 is further used to determine the overdue rate of the customer group to which the multiple target customers belong in the kth repayment period based on the overdue probability of the multiple target customers in the kth repayment period and the loan amounts of the multiple target customers.

[0093] Figure 6 The structure diagram of a device 6000 for determining the overdue status of a customer group provided in an embodiment of the present application is shown in FIG. Figure 6 As shown, it includes at least one processor 601 and a memory 602 connected to the at least one processor 601. The specific connection medium between the processor 601 and the memory 602 is not limited in the embodiment of the present application. Figure 6 For example, the processor 601 and the memory 602 are connected via a bus. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0094] In the embodiment of the present application, the memory 602 stores instructions that can be executed by at least one processor 601. The at least one processor 601 can implement the steps of the above-mentioned method for determining the overdue status of a customer group by executing the instructions stored in the memory 602.

[0095] The processor 601 is the control center of the computer device, and can use various interfaces and lines to connect various parts of the computer device, and perform resource settings by running or executing instructions stored in the memory 602 and calling data stored in the memory 602. Optionally, the processor 601 may include one or more processing units, and the processor 601 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 601. In some embodiments, the processor 601 and the memory 602 may be implemented on the same chip, and in some embodiments, they may also be implemented separately on independent chips.

[0096] Processor 601 can be a general-purpose processor, such as a central processing unit (CPU), a digital signal processor, an application-specific integrated circuit (ASIC), a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, and can implement or execute the methods, steps and logic block diagrams disclosed in the embodiments of the present application. A general-purpose processor can be a microprocessor or any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as a hardware processor for execution, or can be executed by a combination of hardware and software modules in the processor.

[0097] The memory 602 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory 602 may include at least one type of storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory, a random access memory (Random Access Memory, RAM), a static random access memory (Static Random Access Memory, SRAM), a programmable read-only memory (Programmable Read Only Memory, PROM), a read-only memory (Read Only Memory, ROM), an electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, EEPROM), a magnetic memory, a disk, an optical disk, etc. The memory 602 is any other medium that can be used to carry or store a desired program code in the form of an instruction or data structure and can be accessed by a computer, but is not limited thereto. The memory 602 in the embodiment of the present application can also be a circuit or any other device that can realize a storage function, for storing program instructions and / or data.

[0098] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0099] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0100] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0101] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0102] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for determining the overdue status of a customer group, characterized in that: The method comprises: Obtain the transaction characteristics of multiple target customers belonging to the same customer group in the kth repayment period; k is a positive integer; For at least one target customer, the transaction characteristics of the target customer in the kth repayment period are input into a first model and a second model respectively, and a first overdue transition probability output by the first model and a second overdue transition probability output by the second model are determined, wherein the first overdue transition probability represents the probability that the target customer transfers from non-overdue to overdue, and the second overdue transition probability represents the probability that the target customer remains overdue from overdue; the first model and the second model are obtained by training based on the transaction characteristics of multiple sample customers; Determine the state transition probability of the target customer in the kth repayment period according to the first overdue transition probability and the second overdue transition probability, wherein the state transition probability represents the probability of the first state transitioning to the second state, and the first state and the second state are either overdue or not overdue; Determine the overdue probability of the target customer in the kth repayment period according to the state transition probability of the target customer in the kth repayment period and the overdue probability of the k-1th repayment period, wherein the overdue probability indicates the probability of not repaying on time; For multiple target customers belonging to the same customer group, the overdue status of the customer group to which the multiple target customers belong in the kth repayment period is determined according to the overdue probabilities of the multiple target customers in the kth repayment period.

2. The method according to claim 1, characterized in that Obtain the transaction features of multiple target customers belonging to the same customer group during their k-th repayment period, including: Obtain transaction information of transactions that have occurred when multiple target customers belonging to the same customer group are currently in the Mth repayment period; M is a positive integer, and the Mth repayment period is earlier than the kth repayment period; The transaction information of the transaction that has occurred and the kth repayment period are used as transaction features of the kth repayment period.

3. The method according to claim 1 or 2, characterized in that: The first model and the second model are trained based on transaction characteristics of multiple sample customers, including: For each repayment period of each transaction of each sample customer, obtain the training sample corresponding to each repayment period, each training sample includes transaction features and the overdue status corresponding to the transaction features; According to the overdue status of each training sample in the previous repayment period, the multiple training samples are divided into a first type of training samples and a second type of training samples; the overdue status of the sample customers in the first training sample in the previous repayment period is not overdue; the overdue status of the sample customers in the second training sample in the previous repayment period is overdue; The first model is obtained by training according to the first type of training samples, and the second model is obtained by training according to the second type of training samples.

4. The method according to claim 3, characterized in that The transaction features of each training sample include the transaction information of transactions that have occurred up to the last repayment period corresponding to the training sample and the repayment period corresponding to the training sample; or The transaction features of each training sample include the transaction information of the transactions that have occurred when the loan corresponding to the training sample takes effect and the repayment period corresponding to the training sample.

5. The method according to claim 1 or 2, characterized in that: The state transition probability is a 2*2 state transition matrix; The state transfer matrix is in, represents the first non-overdue transfer probability of the i-th target customer in the k-th repayment period, represents the first overdue transfer probability of the i-th target customer in the k-th repayment period, represents the second non-overdue transfer probability of the i-th target customer in the k-th repayment period, represents the second overdue transfer probability of the i-th target customer in the k-th repayment period.

6. The method according to claim 5, characterized in that Determining the overdue probability of the target customer in the kth repayment period according to the state transition probability of the target customer in the kth repayment period and the overdue probability of the k-1th repayment period includes: For the target customer, determining a first product according to the non-overdue probability of the k-1th repayment period and the first overdue transfer probability; Determining a second product according to the overdue probability of the k-1th repayment period and the second overdue transfer probability; The overdue probability of the kth repayment period is determined based on the first product and the second product, and the sum of the non-overdue probability of the k-1th repayment period and the overdue probability of the k-1th repayment period is 1.

7. The method according to claim 1, characterized in that The overdue status is the overdue rate; Determining the overdue status of the customer group to which the multiple target customers belong in the kth repayment period according to the overdue probabilities of the multiple target customers in the kth repayment period includes: The overdue rate of the customer group to which the multiple target customers belong in the kth repayment period is determined according to the overdue probabilities of the multiple target customers in the kth repayment period and the loan amounts of the multiple target customers.

8. A device for determining the overdue status of a customer group, characterized in that: include: An acquisition module is used to acquire transaction characteristics of multiple target customers belonging to the same customer group in the kth repayment period; k is a positive integer; a processing module, for at least one target customer, inputting the transaction characteristics of the target customer in the kth repayment period into a first model and a second model respectively, and determining a first overdue transition probability output by the first model and a second overdue transition probability output by the second model, wherein the first overdue transition probability represents the probability that the target customer transitions from non-overdue to overdue, and the second overdue transition probability represents the probability that the target customer remains overdue from overdue; the first model and the second model are obtained by training based on the transaction characteristics of a plurality of sample customers; a determination module, configured to determine the state transition probability of the target customer in the kth repayment period according to the first overdue transition probability and the second overdue transition probability, wherein the state transition probability indicates the probability of the first state transitioning to the second state, wherein the first state and the second state are either overdue or not overdue; and, based on the state transition probability of the target customer in the kth repayment period and the overdue probability of the k-1th repayment period, determine the overdue probability of the target customer in the kth repayment period, wherein the overdue probability indicates the probability of not repaying on time; The determination module is also used to determine, for multiple target customers belonging to the same customer group, the overdue status of the customer group to which the multiple target customers belong in the kth repayment period according to the overdue probabilities of the multiple target customers in the kth repayment period.

9. A device for determining the overdue status of a customer group, characterized in that: include: Memory, used to store computer programs or instructions; A processor, configured to call a computer program or instruction stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when a computer reads and executes the instructions, the computer executes the method according to any one of claims 1 to 7.

11. A computer program product, characterized in that The computer program product stores instructions, and when a computer reads and executes the instructions, the computer executes the method according to any one of claims 1 to 7.

Images