A training method for overdue repayment prediction model and overdue repayment prediction method

Abstract

The present invention discloses a training method for an overdue repayment prediction model and an overdue repayment prediction method. The method comprises obtaining sample data of a first overdue stage and sample data of a second overdue stage, wherein the sample feature types in the sample data of the second overdue stage are more than the sample feature types in the sample data of the first overdue stage; creating an initialization model, training the initialization model in the first stage based on a first sample data set, obtaining an intermediate prediction model, wherein the first sample data set at least includes sample data of the second overdue stage; training the intermediate prediction model in the second stage based on a second sample data set, obtaining a target prediction model, wherein the second sample data includes sample data of the first overdue stage, and the target prediction model is used to predict overdue repayments in the first overdue stage. The present invention solves the problem of being unable to accurately predict the early overdue stage, and improves the accuracy of the target prediction model's prediction of the first overdue stage.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a training method for an overdue repayment prediction model and an overdue repayment prediction method. Background Art

[0002] In the post-loan risk control field, the main reliance is on the Collection Scorecard (C-card), which predicts the response probability of the collection strategy for overdue accounts during the account management cycle, and thus adopts corresponding collection strategies.

[0003] At present, in order to further improve the risk measurement ability of C-card and predict the possibility of future account repayment, some end-to-end model strategies are adopted as the upstream features of C-card. Commonly used C-card models include conventional regression models, tree models, etc.

[0004] However, it is difficult to measure the improvement of repayment willingness by some intervention behaviors only by relying on the overall score and customer segmentation of C-card. In order to extract this part of ability, the C-card modeling process needs to be split into two parts. The first part is the score of intervention behavior to improve repayment ability, and the second part is to apply the sub-score to the downstream scorecard. Since the distribution of intervention features in different overdue stages is quite different, there are fewer intervention features in the early overdue stage, which makes it impossible to accurately predict the early overdue stage. Summary of the invention

[0005] The present invention provides a training method for an overdue repayment prediction model and an overdue repayment prediction method to solve the problem that it is impossible to accurately predict the early overdue stage.

[0006] According to one aspect of the present invention, a method for training an overdue repayment prediction model is provided, comprising:

[0007] Acquire sample data of a first overdue stage and sample data of a second overdue stage, wherein the sample data of the second overdue stage has more sample feature types than the sample data of the first overdue stage;

[0008] Creating an initialization model, and performing a first-stage training on the initialization model based on a first sample data set to obtain an intermediate prediction model, wherein the first sample data set at least includes sample data of the second overdue stage;

[0009] The intermediate prediction model is trained in the second stage based on the second sample data set to obtain a target prediction model, wherein the second sample data includes sample data of the first overdue stage, and the target prediction model is used to predict overdue repayments in the first overdue stage.

[0010] According to another aspect of the present invention, there is provided a method for predicting overdue repayments, comprising:

[0011] Obtain the pending data of the target object in the first overdue stage;

[0012] The data to be processed is predicted based on a target prediction model to obtain an overdue repayment prediction result of the target object, wherein the target prediction model is trained using the overdue repayment prediction model training method described in any embodiment of the present invention.

[0013] According to another aspect of the present invention, there is provided a training device for an overdue repayment prediction model, comprising:

[0014] A sample data acquisition module, used to acquire sample data of a first overdue stage and sample data of a second overdue stage, wherein the sample data of the second overdue stage has more sample feature types than the sample data of the first overdue stage;

[0015] An intermediate prediction model training module, used to create an initialization model, and perform a first phase of training on the initialization model based on a first sample data set to obtain an intermediate prediction model, wherein the first sample data set at least includes sample data of the second overdue phase;

[0016] The target prediction model training module is used to perform a second stage of training on the intermediate prediction model based on a second sample data set to obtain a target prediction model, wherein the second sample data includes sample data of the first overdue stage, and the target prediction model is used to perform overdue repayment prediction in the first overdue stage.

[0017] According to another aspect of the present invention, there is provided an overdue repayment prediction device, comprising:

[0018] A to-be-processed data acquisition module, used to acquire the to-be-processed data of the target object in the first overdue stage;

[0019] An overdue repayment prediction module is used to perform prediction processing on the data to be processed based on a target prediction model to obtain an overdue repayment prediction result of the target object, wherein the target prediction model is trained based on the training method of the overdue repayment prediction model described in any embodiment of the present invention.

[0020] According to another aspect of the present invention, there is provided an electronic device, the electronic device comprising:

[0021] at least one processor; and

[0022] a memory communicatively connected to the at least one processor; wherein,

[0023] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the training method of the overdue repayment prediction model and / or the overdue repayment prediction method described in any embodiment of the present invention.

[0024] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the training method of the overdue payment prediction model and / or the overdue payment prediction method described in any embodiment of the present invention when executed.

[0025] The technical solution of the embodiment of the present invention is to obtain sample data of the first overdue stage and sample data of the second overdue stage, wherein the sample feature types in the sample data of the second overdue stage are more than the sample feature types in the sample data of the first overdue stage; create an initialization model, perform the first stage training on the initialization model based on the first sample data set to obtain an intermediate prediction model, wherein the first sample data set at least includes the sample data of the second overdue stage; perform the second stage training on the intermediate prediction model based on the second sample data set to obtain a target prediction model, wherein the second sample data includes the sample data of the first overdue stage, and the target prediction model is used to predict overdue repayments in the first overdue stage. The initialization model is trained in the first stage with a sample data set with more feature types to obtain an intermediate prediction model, and then the intermediate prediction model is trained in the second stage based on the sample data set of the first overdue stage to obtain a target prediction model, which solves the problem of being unable to accurately predict the early overdue stage and improves the accuracy of the target prediction model in predicting the first overdue stage.

[0026] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0028] Figure 1 It is a flowchart of a method for training an overdue repayment prediction model provided by an embodiment of the present invention;

[0029] Figure 2is a flow chart of a method for predicting overdue repayments provided by an embodiment of the present invention;

[0030] Figure 3 It is a structural schematic diagram of a training device for an overdue repayment prediction model provided by an embodiment of the present invention;

[0031] Figure 4 It is a structural schematic diagram of an overdue repayment prediction device provided by an embodiment of the present invention;

[0032] Figure 5 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0034] It should be noted that the terms "first overdue stage", "second overdue stage", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0035] Figure 1 This is a flowchart of a method for training an overdue repayment prediction model provided by an embodiment of the present invention. This embodiment can be applied to training a post-loan overdue repayment prediction model. The method can be executed by a training device for an overdue repayment prediction model. The training device for an overdue repayment prediction model can be implemented in the form of hardware and / or software. The training device for an overdue repayment prediction model can be configured in a device for training an overdue repayment prediction model. Figure 1 As shown, the method includes:

[0036] S110: Acquire sample data of a first overdue stage and sample data of a second overdue stage, wherein the sample data of the second overdue stage has more sample feature types than the sample data of the first overdue stage.

[0037] Among them, the first overdue stage is an early stage compared to the second overdue stage. Specifically, since the overdue objects will have different overdue performances in different overdue stages, targeted collection behaviors need to be adopted in different overdue stages, and the collection behaviors include but are not limited to text messages, robot automatic collection, manual collection, etc.; Under normal circumstances, the first overdue stage will not adopt active collection behaviors such as manual collection, and most of them will occur in the second overdue stage with a longer overdue time. In this embodiment, sample data of the first overdue stage and sample data of the second overdue stage are obtained, wherein the sample features in the sample data mainly include two parts of features, one is the portrait features of the target object, including behavioral accounts, balances, user status, service rates and other features, which are not limited here; the other part is overdue features, including available credit, overdue days, overdue amount and other features, which are not limited here.

[0038] It can be understood that compared with the second overdue stage, the first overdue stage is in the early stage, and there are fewer active collection behaviors. Correspondingly, there are fewer sample feature types in the sample data of the first overdue stage. Therefore, the later the overdue stage, the more sample feature types there are in the sample data of the overdue stage. In other words, there are more sample feature types in the sample data of the second overdue stage than in the sample data of the first overdue stage.

[0039] Based on the above embodiment, optionally, the label type of the sample data in the first overdue stage and the label type of the sample data in the second overdue stage are the same.

[0040] In this embodiment, since the label types of the sample data in the first overdue stage and the sample data in the second overdue stage are the same, the label types of the sample data in both the first stage training and the second stage training are the same, and the same label type can improve the prediction accuracy of the target prediction model.

[0041] For example, in the collection scenario, the monthly repayment period is used, and the tag type is defined as:

[0042]

[0043] S120, creating an initialization model, and performing a first stage of training on the initialization model based on a first sample data set to obtain an intermediate prediction model, wherein the first sample data set at least includes sample data of the second overdue stage.

[0044] In this embodiment, before model training is performed, an initialization model is created, wherein the initialization model includes but is not limited to a deep neural network model, a convolutional neural network model, a residual neural network model, etc., which are not limited here. After the initialization model is created, the initialization model can be trained in the first stage according to the first sample data set to obtain an intermediate prediction model. Among them, the first stage is the pre-training stage. It can be understood that the target prediction model trained in this embodiment is to predict overdue repayments in the first overdue stage. In the pre-training stage, the initialization model needs to be trained with sample data of more sample types to obtain an intermediate prediction model. Therefore, the first sample data set at least includes sample data of the second overdue stage.

[0045] On the basis of the above embodiment, optionally, the first stage of training the initialization model is performed based on the first sample data set to obtain an intermediate prediction model, including: iteratively executing the following training process to obtain the intermediate prediction model: extracting sample data of local label type from the first sample data set, and locally extracting data features of the extracted sample data to form a third sample data set; training the initialization model based on the third sample data set to update the model parameters in the initialization model.

[0046] Among them, the third sample data set is a sample data set formed after two local extractions of the first sample data set. Specifically, the third sample data set includes a support set and a query set, wherein the support set is formed by the sample data after extraction, and the query set is formed by the sample data remaining after extraction from the first sample data set. In this embodiment, the iterative training initialization model, before each round of iterative training, can first randomly extract local label type sample data from the first sample data set, and then perform local extraction of data features in the extracted sample data to form the third sample data, and then train the initialization model according to the third sample data to update the model parameters in the initialization model. Specifically, the formation process of the third sample data set is: randomly extract N label type sample data from the first sample data set, and then extract K sample data from the sample data of each label type to form the third sample data set; exemplary, it can be expressed as N-way K-shot, wherein N is less than the number of label types of the first sample data set, way represents the label type, K is less than the total number of sample data under the label type, shot represents the sample data, and N and K are set by those skilled in the art according to needs and experience, which is not limited.

[0047] On the basis of the above embodiment, optionally, the training of the initialization model based on the third sample data set includes: creating multiple training tasks, the training task including an initialization model; training the initialization model in the training task based on the sample data in the third sample data set to update the model parameters of the initialization model, and using the model parameters of the initialization model as the initial parameters of the initialization model in the next training task; obtaining the target initial parameters of the initialization model in the current iteration training based on the model parameters of the updated initialization model of the multiple training tasks, and using the target initial parameters of the initialization model in the current iteration training as the initial parameters of the initialization model in the next iteration.

[0048] Among them, the training task refers to the training task of each round of iterative training, and each round of iterative training includes multiple training tasks, and each training task includes an initialization model. In this embodiment, in each round of iterative training, multiple training tasks are created, and multiple training tasks form a batch training task; for any training task in the batch training task, the initialization model of the training task is trained based on the sample data in the third sample data set, the model parameters of the initialization model are updated based on the first gradient descent model parameter update formula, and the model parameters of the updated initialization model are used as the initial parameters of the initialization model in the next training task, until the initialization model of all training tasks in the batch training task is trained. Before the end of the current iteration, the model parameters of the updated initialization model of each training task in the batch training task are tested based on the query set in the third sample data set, and the initial parameters are updated based on the second gradient descent model parameter update formula to obtain the target initial parameters of the initialization model in the current iterative training, and the target initial parameters of the initialization model in the current iterative training are used as the initial parameters of the initialization model in the next iteration.

[0049] It should be noted that when the model is initialized for the first time, the initial parameters are obtained by randomly initializing the model parameters.

[0050] Exemplarily, the first gradient descent model parameter update formula is:

[0051]

[0052] Among them, θ ‘ i The model parameters of the initialized model after updating for the i-th training task in the batch training task, θ i is the model parameters of the initialization model before updating the i-th training task in the batch training task, α is the learning rate, is the gradient of the loss function L, is the loss value of the regression loss function in the i-th training task in the batch training task; regression loss functions include but are not limited to mean square error loss function, mean absolute error loss function, etc., which are not limited here.

[0053] It should be noted that, i∈[1,bitch size], bit size is the number of training tasks in a batch training task, which is set by those skilled in the art and is not limited here. Exemplarily, the bit size can be 10.

[0054] Exemplarily, the second gradient descent model parameter update formula is:

[0055]

[0056] Among them, θ 目标 is the target initial parameter, θ 初 is the initial parameter of random initialization, β is the learning rate, and other symbols are the same as the above formula and will not be repeated here.

[0057] S130. Perform a second stage of training on the intermediate prediction model based on a second sample data set to obtain a target prediction model, wherein the second sample data includes sample data of the first overdue stage, and the target prediction model is used to perform overdue repayment prediction in the first overdue stage.

[0058] Among them, the second sample data set is the sample data of the first overdue stage. Specifically, the second sample data set includes a support set and a query set. Exemplarily, the ratio of the support set to the query set can be 1:4. The ratio of the support set to the query set in the second sample data set is set by technicians in this field based on experience and needs, and is not limited here. In this embodiment, the intermediate prediction model can be trained and tested in the second stage according to the support set and query set in the second sample data set to obtain a target prediction model; wherein the target prediction model is used to predict overdue repayments based on the feature data of the target object in the first overdue stage.

[0059] On the basis of the above embodiment, optionally, the intermediate prediction model is trained in a second stage based on the second sample data set to obtain a target prediction model, including: creating multiple models to be optimized based on the model parameters of the intermediate prediction model; training the multiple models to be optimized based on the second sample data set to obtain multiple optional models; and determining the target prediction model based on the multiple optional models.

[0060] Among them, the model to be optimized is a model whose model parameters and model structure are the same as those of the intermediate prediction model. In this embodiment, multiple models to be optimized can be created based on the model parameters of the intermediate prediction model, and then the multiple models to be optimized are trained and tested respectively based on the support set and query set in the second sample data set to obtain multiple optional models; further, the target prediction model is determined based on the multiple optional models. Among them, the optional model is the model after the second stage of training, and the model structure of the optional model is the same as the above-mentioned model to be optimized, the initialization model, etc., and the difference lies in the model parameters.

[0061] On the basis of the above embodiment, optionally, determining the target prediction model based on the multiple optional models includes: determining a parameter mean of model parameters of the multiple optional models, and creating a target prediction model based on the parameter mean; or, screening the multiple optional models based on preset indicators to determine the target prediction model.

[0062] In this embodiment, among multiple optional models, the model parameters of each optional model can be obtained, the parameter mean can be calculated according to the model parameters of each optional model, and the parameter mean can be used as the model parameter to create a target prediction model; in this embodiment, the target prediction model is created by the model parameter mean of multiple optional models, which can avoid extreme situations and improve the stability of the target prediction model.

[0063] In this embodiment, among multiple optional models, the preset index of each optional model can also be calculated, and the optional model that meets the screening conditions can be selected as the target prediction model, or the optional model with the best preset index can be selected as the target prediction model. Among them, the preset index is the screening evaluation index of the optional model, and the preset index includes but is not limited to the mean square error (MSE), the area under the curve (Area Under Curve, AUC) of the receiver operating characteristic curve (receiveroperating characteristic curve, referred to as ROC curve), etc., which are not limited here.

[0064] The technical solution of this embodiment is to obtain sample data of the first overdue stage and sample data of the second overdue stage, wherein the sample feature types in the sample data of the second overdue stage are more than the sample feature types in the sample data of the first overdue stage; create an initialization model, perform the first stage training on the initialization model based on the first sample data set to obtain an intermediate prediction model, wherein the first sample data set at least includes sample data of the second overdue stage; perform the second stage training on the intermediate prediction model based on the second sample data set to obtain a target prediction model, wherein the second sample data includes sample data of the first overdue stage, and the target prediction model is used to predict overdue repayments in the first overdue stage. The initialization model is trained in the first stage with a sample data set with more feature types to obtain an intermediate prediction model, and then the intermediate prediction model is trained in the second stage based on the sample data set of the first overdue stage to obtain a target prediction model, which solves the problem of being unable to accurately predict the early overdue stage and improves the accuracy of the target prediction model in the first overdue stage.

[0065] Based on the above embodiment, optionally, the method further includes: determining a verification index of the target prediction model, and performing accuracy verification on the target prediction model based on the verification index.

[0066] Among them, the verification index is a related index for verifying the accuracy of the target prediction model. Specifically, the verification index includes but is not limited to MSE, AUC value, etc., which are not limited here. In this embodiment, each verification index of the target prediction model can be determined, and then the accuracy of the prediction model can be verified for the target prediction model according to the verification index. If the verification result does not meet the accuracy requirement of the target prediction model, the target prediction model can be further trained.

[0067] Figure 2 This is a flow chart of an overdue repayment prediction method provided by an embodiment of the present invention. This embodiment is applicable to the case of predicting the overdue repayment risk in the early overdue stage after the loan. The method can be executed by an overdue repayment prediction device, which can be implemented in the form of hardware and / or software. The overdue repayment prediction device can be configured in an overdue repayment prediction device. Figure 2 As shown, the method includes:

[0068] S210: Obtain the to-be-processed data of the target object in the first overdue stage.

[0069] Among them, the target object refers to the object that is overdue after the loan. Specifically, the target object includes but is not limited to the loan unit, loan company, bandwidth individual, etc., which is limited here. The data to be processed refers to the feature data of the target object in the first overdue stage. Specifically, the data to be processed includes the portrait features and overdue features of the target object. The specific feature types are the same as the feature types in the sample data in the above embodiment, and will not be repeated here.

[0070] S220. Predictively process the data to be processed based on a target prediction model to obtain an overdue repayment prediction result of the target object, wherein the target prediction model is trained using the overdue repayment prediction model training method described in any embodiment of the present invention.

[0071] In this embodiment, the target object's pending data in the first overdue stage is input into the training method of the overdue repayment prediction model described in any embodiment of the present invention to obtain a target prediction model for training and prediction processing to obtain an overdue repayment prediction result of the target object; wherein the overdue repayment prediction result includes but is not limited to the target object's willingness to repay, the target object's arrears date distribution and the amount of arrears, etc., which are not limited here.

[0072] The technical solution of this embodiment obtains the to-be-processed data of the target object in the first overdue stage. The to-be-processed data is predicted and processed based on the target prediction model to obtain the overdue repayment prediction result of the target object. The overdue repayment prediction is performed in the early stage of overdue by the target prediction model to obtain the overdue repayment prediction result, and the overdue repayment prediction result is used as the basis for measuring the repayment willingness and risk of the target object, so as to arrange the relevant resources of post-loan management more reasonably.

[0073] Figure 3 Schematic diagram of the structure of a training device for an overdue repayment prediction model provided by an embodiment of the present invention. Figure 3 As shown, the device comprises:

[0074] The sample data acquisition module 310 is used to acquire sample data of a first overdue stage and sample data of a second overdue stage, wherein the sample feature types in the sample data of the second overdue stage are more than the sample feature types in the sample data of the first overdue stage;

[0075] An intermediate prediction model training module 320 is used to create an initialization model, and perform a first stage of training on the initialization model based on a first sample data set to obtain an intermediate prediction model, wherein the first sample data set at least includes sample data of the second overdue stage;

[0076] The target prediction model training module 330 is used to perform a second stage of training on the intermediate prediction model based on a second sample data set to obtain a target prediction model, wherein the second sample data includes sample data of the first overdue stage, and the target prediction model is used to perform overdue repayment prediction in the first overdue stage.

[0077] The technical solution of this embodiment is to obtain sample data of the first overdue stage and sample data of the second overdue stage, wherein the sample feature types in the sample data of the second overdue stage are more than the sample feature types in the sample data of the first overdue stage; create an initialization model, perform the first stage training on the initialization model based on the first sample data set to obtain an intermediate prediction model, wherein the first sample data set at least includes sample data of the second overdue stage; perform the second stage training on the intermediate prediction model based on the second sample data set to obtain a target prediction model, wherein the second sample data includes sample data of the first overdue stage, and the target prediction model is used to predict overdue repayments in the first overdue stage. The initialization model is trained in the first stage with a sample data set with more feature types to obtain an intermediate prediction model, and then the intermediate prediction model is trained in the second stage based on the sample data set of the first overdue stage to obtain a target prediction model, which solves the problem of being unable to accurately predict the early overdue stage and improves the accuracy of the target prediction model in the first overdue stage.

[0078] Based on the above embodiment, optionally, the label type of the sample data in the first overdue stage and the label type of the sample data in the second overdue stage are the same.

[0079] On the basis of the above embodiment, optionally, the intermediate prediction model training module 320 is specifically used to iteratively execute the following training process to obtain the intermediate prediction model: extracting sample data of local label type from the first sample data set, and performing local extraction of data features on the extracted sample data to form a third sample data set; training the initialization model based on the third sample data set to update the model parameters in the initialization model.

[0080] Based on the above embodiment, optionally, the intermediate prediction model training module 320 includes an initialization model training unit, which is used to create multiple training tasks, wherein the training tasks include an initialization model; based on the sample data in the third sample data set, the initialization model in the training task is trained to update the model parameters of the initialization model, and the model parameters of the initialization model are used as the initial parameters of the initialization model in the next training task; based on the model parameters of the updated initialization model of the multiple training tasks, the target initial parameters of the initialization model in the current iteration training are obtained, and the target initial parameters of the initialization model in the current iteration training are used as the initial parameters of the initialization model in the next iteration.

[0081] Based on the above embodiment, optionally, the target prediction model training module 330 is specifically used to create multiple models to be optimized based on the model parameters of the intermediate prediction model; train the multiple models to be optimized based on the second sample data set to obtain multiple optional models; and determine the target prediction model based on the multiple optional models.

[0082] On the basis of the above embodiment, optionally, the target prediction model training module 330 target prediction model determination unit is used to determine the parameter mean of the model parameters of multiple optional models, and create a target prediction model based on the parameter mean; or, screen the multiple optional models based on preset indicators to determine the target prediction model.

[0083] Based on the above embodiment, optionally, the device further includes an accuracy verification module, which is used to determine a verification index of the target prediction model, and perform accuracy verification on the target prediction model based on the verification index.

[0084] The training device for the overdue repayment prediction model provided in the embodiment of the present invention can execute the training method for the overdue repayment prediction model provided in any embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method.

[0085] Figure 4 Schematic diagram of the structure of an overdue repayment prediction device provided by an embodiment of the present invention. Figure 4 As shown, the device comprises:

[0086] The to-be-processed data acquisition module 410 is used to acquire the to-be-processed data of the target object in the first overdue stage;

[0087] The overdue repayment prediction module 420 is used to perform prediction processing on the data to be processed based on the target prediction model to obtain the overdue repayment prediction result of the target object, wherein the target prediction model is trained based on the training method of the overdue repayment prediction model described in any embodiment of the present invention.

[0088] The overdue repayment prediction device provided in the embodiment of the present invention can execute the overdue repayment prediction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0089] Figure 5 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0090] like Figure 5 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0091] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0092] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the training method of the overdue repayment prediction model and / or the overdue repayment prediction method.

[0093] In some embodiments, the training method of the overdue repayment prediction model and / or the overdue repayment prediction method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the training method of the overdue repayment prediction model and / or the overdue repayment prediction method described above may be executed. Alternatively, in other embodiments, the processor 11 may be configured to execute the training method of the overdue repayment prediction model and / or the overdue repayment prediction method in any other appropriate manner (e.g., by means of firmware).

[0094] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0095] The computer program for implementing the training method of the overdue repayment prediction model and / or the overdue repayment prediction method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or the block diagram are implemented. The computer program can be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0096] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to cause a processor to execute a training method for an overdue repayment prediction model, the method comprising:

[0097] Obtain sample data of a first overdue stage and sample data of a second overdue stage, wherein the sample data of the second overdue stage has more sample feature types than the sample data of the first overdue stage; create an initialization model, perform a first stage of training on the initialization model based on the first sample data set to obtain an intermediate prediction model, wherein the first sample data set at least includes the sample data of the second overdue stage; perform a second stage of training on the intermediate prediction model based on the second sample data set to obtain a target prediction model, wherein the second sample data includes the sample data of the first overdue stage, and the target prediction model is used to predict overdue repayments in the first overdue stage.

[0098] And / or, performing a method for predicting overdue payments, the method comprising:

[0099] Obtain the data to be processed of the target object in the first overdue stage; perform prediction processing on the data to be processed based on a target prediction model to obtain an overdue repayment prediction result of the target object, wherein the target prediction model is trained based on the training method of the overdue repayment prediction model described in any embodiment of the present invention.

[0100] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, 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 foregoing.

[0101] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0102] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0103] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0104] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0105] 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 can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A training method for an overdue repayment prediction model, characterized in that: include: Acquire sample data of a first overdue stage and sample data of a second overdue stage, wherein the sample data of the second overdue stage has more sample feature types than the sample data of the first overdue stage; Creating an initialization model, and performing a first-stage training on the initialization model based on a first sample data set to obtain an intermediate prediction model, wherein the first sample data set at least includes sample data of the second overdue stage; The intermediate prediction model is trained in the second stage based on the second sample data set to obtain a target prediction model, wherein the second sample data includes sample data of the first overdue stage, and the target prediction model is used to predict overdue repayments in the first overdue stage.

2. The method according to claim 1, characterized in that The sample data in the first overdue stage and the sample data in the second overdue stage have the same label type.

3. The method according to claim 1, characterized in that The first stage of training the initialization model based on the first sample data set to obtain an intermediate prediction model includes: Iterate the following training process to obtain the intermediate prediction model: Extracting sample data of a local label type from the first sample data set, and performing local extraction of data features on the extracted sample data to form a third sample data set; The initialization model is trained based on the third sample data set to update model parameters in the initialization model.

4. The method according to claim 3, characterized in that The training of the initialization model based on the third sample data set includes: Creating a plurality of training tasks, wherein the training tasks include an initialization model; Training an initialization model in one of the training tasks based on the sample data in the third sample data set to update model parameters of the initialization model, and using the model parameters of the initialization model as initial parameters of the initialization model in the next training task; The target initial parameters of the initialized model in the current iteration training are obtained based on the model parameters of the updated initialized model of the multiple training tasks, and the target initial parameters of the initialized model in the current iteration training are used as the initial parameters of the initialized model in the next iteration.

5. The method according to claim 1, characterized in that The second stage of training the intermediate prediction model based on the second sample data set to obtain a target prediction model includes: Creating a plurality of models to be optimized based on the model parameters of the intermediate prediction model; Based on the second sample data set, the plurality of models to be optimized are trained respectively to obtain a plurality of optional models; The target prediction model is determined based on the multiple optional models.

6. The method according to claim 5, characterized in that The determining the target prediction model based on the multiple optional models includes: Determine the parameter means of the model parameters of the plurality of optional models, and create a target prediction model based on the parameter means; or, Based on preset indicators, the multiple optional models are screened to determine the target prediction model.

7. The method according to claim 1, characterized in that The method further comprises: Determine a verification indicator for the target prediction model, and perform accuracy verification on the target prediction model based on the verification indicator.

8. A method for predicting overdue repayments, characterized in that: include: Obtain the pending data of the target object in the first overdue stage; The data to be processed is predicted based on a target prediction model to obtain an overdue repayment prediction result of the target object, wherein the target prediction model is trained based on the training method of the overdue repayment prediction model as described in any one of claims 1-7.

9. A training device for an overdue repayment prediction model, characterized in that: include: A sample data acquisition module, used to acquire sample data of a first overdue stage and sample data of a second overdue stage, wherein the sample data of the second overdue stage has more sample feature types than the sample data of the first overdue stage; An intermediate prediction model training module, used to create an initialization model, and perform a first phase of training on the initialization model based on a first sample data set to obtain an intermediate prediction model, wherein the first sample data set at least includes sample data of the second overdue phase; The target prediction model training module is used to perform a second stage of training on the intermediate prediction model based on a second sample data set to obtain a target prediction model, wherein the second sample data includes sample data of the first overdue stage, and the target prediction model is used to perform overdue repayment prediction in the first overdue stage.

10. An overdue repayment prediction device, characterized in that: include: A to-be-processed data acquisition module, used to acquire the to-be-processed data of the target object in the first overdue stage; An overdue repayment prediction module is used to perform prediction processing on the data to be processed based on a target prediction model to obtain an overdue repayment prediction result of the target object, wherein the target prediction model is trained based on the training method of the overdue repayment prediction model as described in any one of claims 1-7.

11. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the training method of the overdue repayment prediction model described in any one of claims 1-7 and / or the overdue repayment prediction method described in claim 8.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the training method of the overdue repayment prediction model described in any one of claims 1 to 7 and / or the overdue repayment prediction method described in claim 8 when executed.

Images