User dynamic probability prediction method, device, computer equipment and storage medium

Abstract

The present application relates to a method, device, computer equipment and storage medium for predicting the probability of user's spending. The method comprises: obtaining user characteristic data of a target user; obtaining the current number of credit limit adjustments of the target user; predicting the probability of the target user's spending after the current number of credit limit adjustments according to the user characteristic data of the target user, the current number of credit limit adjustments and a pre-established probability of spending prediction model; wherein the probability of spending prediction model is a model constructed with the user characteristic data and survival time of sample users as model features, and is used to predict the probability of a user's spending after any one or more credit limit adjustments, and the survival time is related to the number of credit limit adjustments of the sample user. The use of this method can improve the accuracy of the prediction of the probability of user spending.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device, computer equipment and storage medium for predicting user spending probability. Background Art

[0002] In the process of credit business data processing, the credit limit will be adjusted for the user to meet the user's needs based on the user's characteristics and spending behavior, for example, increasing the user's loan limit. However, the total amount of funds is limited. If the credit limit adjustment strategy is not deployed reasonably, the user will not make full use of the credit limit after the credit limit is increased, which will cause a waste of credit limit resources. Therefore, the user's spending probability can be analyzed in advance, and the strategy can be deployed based on the analysis results.

[0003] However, in the traditional method, it is assumed that the probability of users making withdrawals after each increase in their credit limit is the same. Moreover, based on the traditional method, some samples that have just been increased a few times will be discarded because these samples do not have enough data to reflect whether the user has made withdrawals. As a result, there will be a deviation between the sample and the population, and it cannot reflect the user's actual situation well, resulting in low accuracy in the prediction of the user's withdrawal probability. Summary of the invention

[0004] Based on this, it is necessary to provide a method, device, computer equipment and storage medium for predicting the probability of user spending, which can improve the accuracy of user spending probability prediction, in response to the above technical problems.

[0005] A method for predicting user's probability of spending, the method comprising: obtaining user characteristic data of a target user; obtaining the current number of credit limit adjustments of the target user; predicting the probability of spending of the target user after the current number of credit limit adjustments based on the user characteristic data of the target user, the current number of credit limit adjustments and a pre-established probability of spending prediction model; wherein the probability of spending prediction model is a model constructed with user characteristic data and survival time of sample users as model features, and is used to predict the probability of spending of the user after any one or more credit limit adjustments, and the survival time is related to the number of credit limit adjustments of the sample users.

[0006] In one embodiment, a method for constructing a dynamic spending probability prediction model includes: obtaining user characteristic data of sample users; analyzing the survival time of sample users; determining target variables corresponding to sample users; and constructing a dynamic spending probability prediction model based on the user characteristic data of sample users, the survival time of sample users, and the target variables corresponding to sample users.

[0007] In one embodiment, analyzing the survival time of sample users includes: obtaining the spending behavior information of the sample users after each credit limit adjustment, and analyzing whether the sample users have any unspent spending behavior based on the spending behavior information; if so, the number of credit limit adjustments corresponding to the first occurrence of unspent spending behavior is used as the survival time of the sample users; if not, the number of credit limit adjustments when the sample is collected is determined as the survival time of the sample user.

[0008] In one embodiment, a spending probability prediction model is constructed based on user characteristic data of sample users, survival time of sample users and target variables corresponding to sample users, including: obtaining a risk function of any sample user after any credit limit adjustment based on machine learning algorithm and survival analysis method according to user characteristic data of sample users, survival time of sample users and target variables corresponding to sample users; and obtaining a spending probability prediction model for calculating the spending probability of any user after any one or more credit limit adjustments according to each risk function.

[0009] In one embodiment, the user characteristic data includes at least one of user embedding data, user external credit data, and user lending behavior data.

[0010] In one embodiment, after acquiring the user characteristic data of the sample users, the method further includes: performing at least one of data cleaning processing, data fusion processing and data derivation processing on the user characteristic data of the sample users.

[0011] In one embodiment, the method further includes: generating an evaluation report according to the predicted spending probability of the target user, and sending the evaluation report to the terminal.

[0012] In one embodiment, the method further includes: obtaining a preset probability threshold, and when the predicted spending probability of the target user is less than the preset probability threshold, generating a control instruction to stop the next credit limit adjustment operation for the target user.

[0013] A user spending probability prediction device, the device comprising:

[0014] A feature data acquisition module is used to acquire user feature data of a target user;

[0015] An adjustment times determination module is used to obtain the current credit limit adjustment times of the target user;

[0016] A payment probability prediction module is used to predict the payment probability of the target user after the current number of credit limit adjustments based on the user characteristic data of the target user, the current number of credit limit adjustments and a pre-established payment probability prediction model; wherein the payment probability prediction model is a model constructed with the user characteristic data and survival time of sample users as model features, and is used to predict the user's payment probability after any one or more credit limit adjustments, and the survival time is related to the number of credit limit adjustments of the sample users.

[0017] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of any one of the above-mentioned user dynamic support probability prediction methods when executing the computer program.

[0018] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any of the above-mentioned user dynamic support probability prediction methods.

[0019] The above-mentioned user's probability of spending can be predicted by using the user characteristic data of the target user combined with the current number of credit limit adjustments as the input of the pre-established probability of spending prediction model, and can obtain the probability of spending generated by the target user after the current number of credit limit adjustments, which is output by the probability of spending prediction model. Since the probability of spending prediction model is a model constructed in advance based on the user characteristic data of the sample user and combined with the survival time determined based on the number of credit limit adjustments, that is, the probability of spending prediction model adds the user's survival time as one of the model features to the model construction, the probability of spending prediction model can convert the number of credit limit adjustments of the user into the corresponding survival time for prediction. In the traditional method, the probability of spending after each credit limit adjustment is regarded as the same, but this does not conform to the actual objective law. In the actual process of credit limit adjustment, the probability of spending corresponding to the first few adjustments will be higher than the next few adjustments. Therefore, accurate prediction cannot be achieved without considering the impact of the number of credit limit adjustments on the probability of spending. This solution associates the number of credit limit adjustments with the user's survival time, and adds the survival time as a model feature to the model construction. Therefore, the payment probability prediction model of this solution is used to predict the user's payment probability. It can predict the user's payment probability after any one or more credit limit adjustments, thereby improving the accuracy of the user's payment probability prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 1 is a flow chart of a method for predicting user spending probability in one embodiment;

[0021] Figure 2 1 is a flow chart of a method for constructing a dynamic support probability prediction model in one embodiment;

[0022] Figure 3 is a structural block diagram of a device for predicting user movement probability in one embodiment;

[0023] Figure 4 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0025] In one embodiment, the user's spending probability prediction method provided by the present application can be applied to a server. Specifically, the server obtains the user characteristic data of the target user; obtains the current credit limit adjustment times of the target user; predicts the spending probability of the target user after the current credit limit adjustment times based on the user characteristic data of the target user, the current credit limit adjustment times and a pre-established spending probability prediction model; wherein the spending probability prediction model is a model constructed with the user characteristic data and survival time of sample users as model features for predicting the user's spending probability after any one or more credit limit adjustments, and the survival time is related to the credit limit adjustment times of the sample users. Among them, the server can be a cloud server, a terminal server, a virtual server, etc., and the server can be implemented as an independent server or a server cluster consisting of multiple servers.

[0026] In one embodiment, Figure 1 As shown, a method for predicting the probability of user spending is provided, and the method is applied to a server as an example for explanation, including the following steps:

[0027] Step S102: Acquire user characteristic data of the target user.

[0028] The target user refers to the user who is determined to be subject to a probability prediction for loan withdrawal, and can be any one or more credit users. User feature data refers to data used to reflect user attribute characteristics in credit business, for example, it can include but is not limited to: user's embedded point data, external credit data, lending behavior data, etc.

[0029] Specifically, in this step, the server may determine the target user for whom the probability of dynamic expenditure is to be predicted, and may lock and read the user feature data corresponding to the target user by calling a cache, accessing a database, and the like.

[0030] Step S104: Obtain the current credit limit adjustment times of the target user.

[0031] Among them, credit limit adjustment refers to the operation of adjusting the credit limit data that a user can borrow, for example, increasing the upper limit of the credit limit that a user can borrow, etc. The current credit limit adjustment number refers to the total number of credit limit adjustment operations that have been performed on the user at the current time point or time period. The number of credit limit adjustment operations can be one or more times. For example, if the user has been adjusted three times at the current time point or time period, the current credit limit adjustment number is 3.

[0032] Exemplarily, after each credit limit adjustment operation on a user, the relevant information of the credit limit adjustment can be recorded and stored accordingly. The server can determine the number of credit limit adjustments for the target user by reading the relevant information of the credit limit adjustment of the target user recorded previously, thereby determining the current number of credit limit adjustments.

[0033] Step S106: Based on the user characteristic data of the target user, the current number of credit limit adjustments and a pre-established payment probability prediction model, predict the payment probability of the target user after the current number of credit limit adjustments; wherein the payment probability prediction model is a model constructed with the user characteristic data and survival time of sample users as model features, and is used to predict the user's payment probability after any one or more credit limit adjustments, and the survival time is related to the number of credit limit adjustments of the sample users.

[0034] In this step, the probability of expenditure prediction model can be a model that is pre-built and trained using the data of the sample set, and can be used to predict the probability of expenditure for any user after any one or more credit limit adjustments. The sample set can be composed of the data of sample users, and all or part of the features can be selected from the user feature data of the sample users as model features as needed, and the survival time of the sample users can be combined, and the survival time can also be added as one of the model features to the construction and training of the probability of expenditure prediction model. Among them, survival time refers to the concept in the survival analysis method, and its definition can be: the time span experienced by the target object from the starting event to the end event is called the survival time. Here, the number of credit limit adjustments of the sample user can be associated with the survival time, that is, the number of credit limit adjustments of the sample user can be used as a discrete survival time, and the survival time of the user can be determined by the number of credit limit adjustments of the user. For example, the number of times the user's current credit limit is adjusted when the sample is collected can be used as the user's survival time; the first credit limit adjustment can also be regarded as the starting point of the event, and the first credit limit adjustment operation that occurs after the credit limit adjustment but the user does not spend the money can be regarded as the end point of the event, and the number of credit limit adjustments experienced by the user between the event start point and the event end point can be regarded as the user's survival time.

[0035] For example, the server can use the user feature data of the target user and the current number of credit limit adjustments as input variables of a pre-established credit limit probability prediction model. After prediction and calculation based on the credit limit probability prediction model, the credit limit probability prediction model can output the credit limit probability corresponding to the target user after the current number of credit limit adjustments. For example, if the current number of credit limit adjustments is 3, the credit limit probability prediction model can output the credit limit probability of the target user after 3 adjustments.

[0036] The above-mentioned user's probability of moving the payment is predicted by taking the user characteristic data of the target user combined with the current number of credit limit adjustments as the input of the pre-established probability of moving the payment prediction model, and can obtain the output of the probability of moving the payment prediction model, which characterizes the probability of moving the payment generated by the target user after the current number of credit limit adjustments. Since the probability of moving the payment prediction model is a model constructed in advance based on the user characteristic data of the sample user and combined with the survival time determined based on the number of credit limit adjustments, that is, the probability of moving the payment prediction model adds the user's survival time as one of the model features to the model construction, therefore, the probability of moving the payment prediction model can convert the number of credit limit adjustments of the user into the corresponding survival time for prediction. In the traditional method, the probability of moving the payment of the user after each credit limit adjustment is regarded as the same, but this does not conform to the actual objective law. In the actual process of credit limit adjustment, the probability of moving the payment corresponding to the first few adjustments will be higher than the next few adjustments. Therefore, it is impossible to achieve accurate prediction without considering the impact of the number of credit limit adjustments on the probability of moving the payment. This solution associates the number of credit limit adjustments with the user's survival time, and adds the survival time as a model feature to the model construction. Therefore, the payment probability prediction model of this solution is used to predict the user's payment probability. It can predict the user's payment probability after any one or more credit limit adjustments, thereby improving the accuracy of the user's payment probability prediction.

[0037] In one embodiment, the method for constructing the dynamic support probability prediction model includes:

[0038] Step S202: Obtain user characteristic data of sample users.

[0039] Specifically, first, a certain number of users can be selected as sample users. The sample users can be randomly selected from all borrowing users, and there is no need to abandon users with fewer credit limit adjustments. In traditional methods, the samples required by traditional models must be samples with a longer performance period. Therefore, when building a model, traditional methods need to discard some samples that have only had their credit limits adjusted a few times. In other words, traditional methods need to discard the data of recent loan users, which will cause the model constructed by traditional methods to be unable to reflect the performance of recent loan users, resulting in inaccurate predictions of the probability of withdrawal. The method of the present application does not impose any special restrictions on the selection of samples. Samples can be randomly selected from borrowing users in different periods, and users in the entire loan life cycle can be used as sample users. Thus, the comprehensiveness of the model sample can be improved, thereby improving the accuracy of model construction and prediction.

[0040] Furthermore, after selecting sample users, user characteristic data of these sample users can be obtained. User characteristic data refers to data used to reflect user attribute characteristics in credit business. For example, user characteristic data may include at least one of user's embedded data, external credit data, lending behavior data, and user credit limit adjustment data.

[0041] Step S204: Analyze the survival time of sample users.

[0042] Specifically, the server can determine the survival time of the sample user by analyzing the number of credit limit adjustments of the sample user. For example, the first credit limit adjustment can be regarded as the event starting point, the first credit limit adjustment that occurs after the credit limit adjustment when the user does not spend the money can be regarded as the event end point, and the total number of credit limit adjustments experienced by the user between the event starting point and the event end point can be regarded as the user's survival time.

[0043] Step S206: Determine the target variable corresponding to the sample user.

[0044] Exemplarily, the target variable refers to the result variable of the machine learning model. For example, the target variable can be divided into two categories: good samples and bad samples. The method for determining the target variable may include: when a user engages in borrowing behavior, the first time the user uses the money within a preset time (for example, thirty days) after the credit limit is increased is regarded as a good sample, and the first time the user uses the money is regarded as a bad sample.

[0045] Step S208: construct a dynamic probability prediction model based on the user characteristic data of the sample user, the survival time of the sample user, and the target variable corresponding to the sample user.

[0046] In this embodiment, a survival analysis method can be used, and the survival analysis method can be combined with a machine learning algorithm to add the survival time reflecting the number of times the user's credit limit is adjusted as a feature variable to the modeling of the probability of activation prediction model. Specifically, survival analysis is a statistical analysis method. This application regards the number of credit limit increases as the user's survival time to predict the user's acceptance of the adjusted credit limit. The number of credit limit increases is regarded as a discrete survival time. For example, the value can be 1, 2, 3, ... N, and then the user's survival analysis is performed based on the XGBoost (eXtreme Gradient Boosting, extreme gradient boosting) algorithm, and the user's probability of activation is determined based on the user's survival rate after the credit limit adjustment.

[0047] In one embodiment, after obtaining the user characteristic data of the sample users, the above method may further include: performing at least one of data cleaning processing, data fusion processing and data derivation processing on the user characteristic data of the sample users.

[0048] In this embodiment, after obtaining user feature data such as user embedding data, user external credit data, and user lending behavior data, further data processing operations such as data cleaning, data fusion, and / or data derivation can be performed on these data. After data processing, model features suitable for model construction can be screened based on the processed user feature data and combined with business needs, etc., so as to improve the diversity, flexibility, and accuracy of model feature selection.

[0049] In one embodiment, analyzing the survival time of sample users may include: obtaining the spending behavior information of the sample users after each credit limit adjustment, and analyzing whether the sample users have any unspent spending behavior based on the spending behavior information; if so, the number of credit limit adjustments corresponding to the first occurrence of unspent spending behavior is used as the survival time of the sample users; if not, the number of credit limit adjustments when the sample is collected is determined as the survival time of the sample user.

[0050] In this embodiment, the server can read the spending behavior information of the sample user and analyze whether the sample user has made spending behavior within a preset time range after each credit limit adjustment based on the spending behavior information of the sample user. For example, the time node can be customized according to business needs, such as thirty days after the credit limit adjustment can be used as the time node for judging whether the user has made spending behavior.

[0051] In one embodiment, a spending probability prediction model is constructed based on user characteristic data of sample users, survival time of sample users and target variables corresponding to sample users, including: obtaining a risk function of any sample user after any credit limit adjustment based on machine learning algorithm and survival analysis method according to user characteristic data of sample users, survival time of sample users and target variables corresponding to sample users; and obtaining a spending probability prediction model for calculating the spending probability of any user after any one or more credit limit adjustments according to each risk function.

[0052] Specifically, the present embodiment is further described in detail below in conjunction with an actual application scenario, wherein the specific process of combining the survival analysis method with the machine learning algorithm to construct the dynamic support probability prediction model of the present embodiment can be referred to as follows:

[0053] 1. Relevant concepts of survival analysis.

[0054] The survival function is the probability that the survival time T of an observed object is greater than or equal to the set threshold t, which is the number of people alive at time t divided by the number of people at the beginning of the period, that is,

[0055]

[0056] The death function is the probability that the survival time of a target object is less than the set threshold t, that is,

[0057] F(t)=1-S(t)

[0058] The death density function is the death rate of a target object at time t, that is,

[0059] f(t)=dF(t) / dt=-dS(t) / dt

[0060] The hazard function is the mortality rate of a target object that survives at time t, but at the moment after time t, that is, it can be expressed by the following formula:

[0061] h(t)=f(t) / S(t)=-d(ln(S(t))) / dt

[0062] The cumulative hazard function is the cumulative hazard function at time t, that is,

[0063]

[0064] 2. The number of credit limit adjustments can be regarded as discrete survival time, and the value of survival time can be 1, 2, 3, ..., N. When building the model, the first non-use behavior within 30 days after the credit limit adjustment is taken as the endpoint event, and the survival time of all sample users can be obtained. Combined with other feature dimensions of sample users, machine learning algorithms such as Python and / or XGboost algorithm are used for modeling. The model can derive the hazard function H(t,s) of any user, that is, for any user instance s i and the number of times any amount is adjusted t j , we can get the risk function H(t j ,s i ), where t j The value can be 1, 2, 3, …, N.

[0065] 3. According to the above risk function of each credit limit adjustment, it can be concluded that after N credit limit adjustments, user s i The calculation formula of the probability of dynamic support is:

[0066]

[0067] Based on the above, after N times of quota adjustment, user s i The calculation formula of the probability of dynamic support can be used to obtain the user's dynamic support probability prediction model based on survival analysis.

[0068] In one embodiment, after the step of predicting the target user's spending probability after the current credit limit adjustment times, the method may further include the following steps: generating an evaluation report based on the predicted spending probability of the target user, and sending the evaluation report to the terminal.

[0069] In this embodiment, the server can obtain the target user's spending probability based on the user spending probability prediction model involved in this application, and generate an evaluation report for evaluating whether the target user is suitable for continuing to adjust the credit limit in combination with other relevant information of the target user. The generated evaluation report can further be displayed in the display interface, or it can also be sent to the terminal corresponding to the manager, so that the manager can clearly and accurately perform strategy analysis based on the evaluation report, make reasonable credit limit adjustment strategies, and improve the accuracy and efficiency of strategy formulation.

[0070] In one embodiment, after the step of predicting the target user's spending probability after the current number of credit limit adjustments, the method may further include the following steps: obtaining a preset probability threshold, and when the predicted target user's spending probability is less than the preset probability threshold, generating a control instruction to stop the next credit limit adjustment operation for the target user.

[0071] In this embodiment, the server may also obtain a preset probability threshold, which refers to a critical value used to determine whether the target user's probability of spending meets business requirements. When the target user's probability of spending is less than the preset probability threshold, the next credit limit adjustment operation for the target user may be stopped. For example, the server may generate a stop instruction to control the parameters of the credit limit adjustment service or program through the instruction, so that the credit limit adjustment service or program stops continuing to adjust the credit limit for the target user; or, further, the server may control the relevant program to stop sending a reminder message to the target user that the credit limit has been adjusted, etc.

[0072] The user's spending probability prediction method involved in the above embodiment can predict the user's spending probability after any one or more credit limit adjustments. Compared with the traditional method, the user's spending probability prediction method involved in this application has at least the following advantages:

[0073] 1. The number of credit limit adjustment operations performed on users is added as survival time into the model construction, so that the model's prediction effect is better, more in line with real scenarios and actual needs, and can more accurately reflect the user's probability of spending after each increase in credit limit.

[0074] 2. Traditional models require a long sample performance period, and some samples need to be discarded for modeling. The payment probability prediction model involved in this application can use users throughout the entire loan life cycle as samples to build a model without discarding users with fewer credit limit adjustments. Since users with fewer credit limit adjustments are mostly recent loan users, the model constructed after discarding these samples cannot reflect the performance of recent loan users. This application does not require special restrictions on sample selection, and sample selection is more comprehensive and can better reflect the overall characteristics. Therefore, the payment probability prediction model involved in this application can improve the accuracy of the prediction.

[0075] 3. The traditional payment probability model assumes that the payment probability is the same after each credit limit adjustment, while the payment probability prediction model involved in this application does not make such unreasonable limitations, that is, it allows differences in the payment probability after each credit limit adjustment, which is more in line with the actual situation, that is, the user's payment probability is not the same after any one or more credit limit adjustments, and can be predicted. Therefore, it is of great significance for subsequent applications such as generating evaluation reports, analyzing the upper limit of credit limit adjustments, and analyzing credit limit adjustment strategies.

[0076] It should be understood that although Figures 1-2 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figures 1-2 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0077] In one embodiment, Figure 3 As shown, a user dynamic support probability prediction device is provided, comprising: a feature data acquisition module 310, an adjustment number determination module 320 and a dynamic support probability prediction module 330, wherein:

[0078] The feature data acquisition module 310 is used to acquire user feature data of the target user;

[0079] The adjustment number determination module 320 is used to obtain the current credit limit adjustment number of the target user;

[0080] The payment probability prediction module 330 is used to predict the payment probability of the target user after the current number of credit limit adjustments based on the user characteristic data of the target user, the current number of credit limit adjustments and a pre-established payment probability prediction model; wherein the payment probability prediction model is a model constructed with the user characteristic data and survival time of sample users as model features, and is used to predict the user's payment probability after any one or more credit limit adjustments, and the survival time is related to the number of credit limit adjustments of the sample users.

[0081] In one embodiment, the device also includes a dynamic support model construction module 340, which is used to construct a dynamic support probability prediction model, which is specifically used to obtain user characteristic data of sample users; analyze the survival time of sample users; determine the target variables corresponding to the sample users; and construct a dynamic support probability prediction model based on the user characteristic data of sample users, the survival time of sample users, and the target variables corresponding to the sample users.

[0082] In one embodiment, the spending model construction module 340 obtains the spending behavior information of the sample user after each credit limit adjustment, and analyzes whether the sample user has any unspent spending behavior based on the spending behavior information; if so, the number of credit limit adjustments corresponding to the first occurrence of unspent spending behavior is used as the survival time of the sample user; if not, the number of credit limit adjustments when the sample is collected is determined as the survival time of the sample user.

[0083] In one embodiment, the expenditure model construction module 340 obtains the risk function of any sample user after any credit limit adjustment based on the user characteristic data of the sample users, the survival time of the sample users and the target variables corresponding to the sample users, based on the machine learning algorithm and the survival analysis method; based on each risk function, a expenditure probability prediction model is obtained for calculating the expenditure probability of any user after any one or more credit limit adjustments.

[0084] In one embodiment, the dynamic support model building module 340 is further used to perform at least one of data cleaning processing, data fusion processing and data derivation processing on the user feature data of the sample users.

[0085] In one embodiment, the spending probability prediction module 330 is further used to generate an evaluation report according to the predicted spending probability of the target user, and send the evaluation report to the terminal.

[0086] In one embodiment, the spending probability prediction module 330 is also used to obtain a preset probability threshold, and when the predicted spending probability of the target user is less than the preset probability threshold, generate a control instruction to stop the next credit limit adjustment operation for the target user.

[0087] For the specific limitations of the user's dynamic probability prediction device, please refer to the limitations of the user's dynamic probability prediction method above, which will not be repeated here. Each module in the above-mentioned user dynamic probability prediction device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0088] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a user dynamic support probability prediction method is implemented.

[0089] Those skilled in the art will understand that Figure 4The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0090] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program: obtaining user characteristic data of a target user; obtaining the current number of credit limit adjustments of the target user; predicting the credit limit probability of the target user after the current number of credit limit adjustments based on the user characteristic data of the target user, the current number of credit limit adjustments, and a pre-established credit limit probability prediction model; wherein the credit limit probability prediction model is a model constructed with user characteristic data and survival time of sample users as model features, and is used to predict the user's credit limit probability after any one or more credit limit adjustments, and the survival time is related to the number of credit limit adjustments of the sample users.

[0091] In one embodiment, when the processor executes the computer program, it also implements the step of constructing a dynamic branch probability prediction model, specifically, obtaining user feature data of sample users; analyzing the survival time of sample users; determining the target variable corresponding to the sample users; and constructing a dynamic branch probability prediction model based on the user feature data of sample users, the survival time of sample users, and the target variable corresponding to the sample users.

[0092] In one embodiment, when the processor executes a computer program to implement analysis of the survival time of a sample user, the following steps are specifically implemented: obtaining the sample user's spending behavior information after each credit limit adjustment, and analyzing whether the sample user has any unspent spending behavior based on the spending behavior information; if so, the number of credit limit adjustments corresponding to the first occurrence of unspent spending behavior is used as the survival time of the sample user; if not, the number of credit limit adjustments when the sample is collected is determined as the survival time of the sample user.

[0093] In one embodiment, when the processor executes a computer program to construct a spending probability prediction model based on user characteristic data of sample users, survival time of sample users, and target variables corresponding to sample users, the following steps are specifically implemented: based on the user characteristic data of sample users, survival time of sample users, and target variables corresponding to sample users, a risk function of any sample user after any credit limit adjustment is obtained based on a machine learning algorithm and a survival analysis method; based on each risk function, a spending probability prediction model is obtained for calculating the spending probability of any user after any one or more credit limit adjustments.

[0094] In one embodiment, when the processor executes the computer program, the following steps are also implemented: performing at least one of data cleaning processing, data fusion processing and data derivation processing on the user feature data of the sample users.

[0095] In one embodiment, when the processor executes the computer program, the processor further implements the following steps: generating an evaluation report according to the predicted spending probability of the target user, and sending the evaluation report to the terminal.

[0096] In one embodiment, when the processor executes the computer program, the following steps are also implemented: obtaining a preset probability threshold, and when the predicted probability of the target user's spending is less than the preset probability threshold, generating a control instruction to stop the next credit limit adjustment operation for the target user.

[0097] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: obtaining user characteristic data of a target user; obtaining the current number of credit limit adjustments of the target user; predicting the probability of credit limit of the target user after the current number of credit limit adjustments based on the user characteristic data of the target user, the current number of credit limit adjustments and a pre-established credit limit probability prediction model; wherein the credit limit probability prediction model is a model constructed with user characteristic data and survival time of sample users as model features, and is used to predict the probability of credit limit of the user after any one or more credit limit adjustments, and the survival time is related to the number of credit limit adjustments of the sample users.

[0098] In one embodiment, when the computer program is executed by the processor, it also implements the steps of constructing a dynamic support probability prediction model, specifically, obtaining user feature data of sample users; analyzing the survival time of sample users; determining the target variables corresponding to the sample users; and constructing a dynamic support probability prediction model based on the user feature data of sample users, the survival time of sample users, and the target variables corresponding to the sample users.

[0099] In one embodiment, when a computer program is executed by a processor to implement analysis of the survival time of a sample user, the following steps are specifically implemented: obtaining the sample user's spending behavior information after each credit limit adjustment, and analyzing whether the sample user has any unspent spending behavior based on the spending behavior information; if so, the number of credit limit adjustments corresponding to the first occurrence of unspent spending behavior is used as the survival time of the sample user; if not, the number of credit limit adjustments when the sample is collected is determined as the survival time of the sample user.

[0100] In one embodiment, when a computer program is executed by a processor to construct a spending probability prediction model based on user characteristic data of sample users, survival time of sample users, and target variables corresponding to sample users, the following steps are specifically implemented: based on the user characteristic data of sample users, survival time of sample users, and target variables corresponding to sample users, a risk function of any sample user after any credit limit adjustment is obtained based on a machine learning algorithm and a survival analysis method; based on each risk function, a spending probability prediction model is obtained for calculating the spending probability of any user after any one or more credit limit adjustments.

[0101] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: performing at least one of data cleaning processing, data fusion processing and data derivation processing on the user feature data of the sample users.

[0102] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: generating an evaluation report according to the predicted spending probability of the target user, and sending the evaluation report to the terminal.

[0103] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining a preset probability threshold, and when the predicted probability of the target user's spending is less than the preset probability threshold, generating a control instruction to stop the next credit limit adjustment operation for the target user.

[0104] Those skilled in the art can understand that all or part of the processes in the methods for implementing the above embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0105] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for predicting user spending probability, the method comprising: Obtain user characteristic data of target users; Obtain the current credit limit adjustment times of the target user; According to the user characteristic data of the target user, the current number of credit limit adjustments and a pre-established credit limit probability prediction model, the credit limit probability of the target user after the current number of credit limit adjustments is predicted; wherein the credit limit probability prediction model is a model constructed with the user characteristic data and survival time of sample users as model features for predicting the user's credit limit probability after any one or more credit limit adjustments, and the survival time is related to the number of credit limit adjustments of the sample users; wherein, The method for constructing the dynamic branch probability prediction model comprises: Obtain user characteristic data of sample users; Analyze the survival time of the sample user; wherein, analyzing the survival time of the sample user includes: obtaining the spending behavior information of the sample user after each credit limit adjustment, and analyzing whether the sample user has unspent spending behavior based on the spending behavior information; if so, the number of credit limit adjustments corresponding to the first occurrence of unspent spending behavior is used as the survival time of the sample user; if not, the number of credit limit adjustments when the sample is collected is determined as the survival time of the sample user; Determine the target variable corresponding to the sample user; Constructing the dynamic support probability prediction model according to the user characteristic data of the sample user, the survival time of the sample user and the target variable corresponding to the sample user; A preset probability threshold is obtained, and when the predicted spending probability of the target user is less than the preset probability threshold, a control instruction is generated to stop the next credit limit adjustment operation for the target user.

2. The method according to claim 1, characterized in that The constructing the dynamic support probability prediction model according to the user feature data of the sample user, the survival time of the sample user and the target variable corresponding to the sample user includes: According to the user characteristic data of the sample users, the survival time of the sample users and the target variable corresponding to the sample users, based on the machine learning algorithm and the survival analysis method, the hazard function of any one of the sample users after any credit limit adjustment is obtained; According to each of the risk functions, a spending probability prediction model is obtained for calculating the spending probability of any user after any one or more credit limit adjustments.

3. The method according to claim 1, characterized in that The user characteristic data includes at least one of user buried point data, user external credit data, and user lending behavior data.

4. The method according to claim 1, characterized in that: After obtaining the user characteristic data of the sample user, the method further includes: At least one of data cleaning, data fusion and data derivation processing is performed on the user feature data of the sample users.

5. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: An evaluation report is generated according to the predicted spending probability of the target user, and the evaluation report is sent to the terminal.

6. A user's spending probability prediction device, characterized in that: The device comprises: A feature data acquisition module is used to acquire user feature data of a target user; An adjustment times determination module is used to obtain the current credit limit adjustment times of the target user; A payment probability prediction module is used to predict the payment probability of the target user after the current number of credit limit adjustments based on the user characteristic data of the target user, the current number of credit limit adjustments and a pre-established payment probability prediction model; wherein the payment probability prediction model is a model constructed with the user characteristic data and survival time of sample users as model features, and is used to predict the payment probability of the user after any one or more credit limit adjustments, and the survival time is related to the number of credit limit adjustments of the sample users; wherein, The device also includes a spending model construction module, which is used to obtain user feature data of sample users; analyze the survival time of the sample users; determine the target variable corresponding to the sample users; and construct the spending probability prediction model according to the user feature data of the sample users, the survival time of the sample users, and the target variable corresponding to the sample users; wherein the analysis of the survival time of the sample users includes: obtaining the spending behavior information of the sample users after each credit limit adjustment, and analyzing whether the sample users have unspent spending behavior according to the spending behavior information; if so, the number of credit limit adjustments corresponding to the first occurrence of unspent spending behavior is used as the survival time of the sample user; if not, the number of credit limit adjustments during sample collection is determined as the survival time of the sample user; The spending probability prediction module is also used to obtain a preset probability threshold, and when the predicted spending probability of the target user is less than the preset probability threshold, generate a control instruction to stop the next credit limit adjustment operation for the target user.

7. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

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

Images