Customer churn prediction method, device, electronic device and computer storage medium

Abstract

The present application provides a method, device, electronic device and computer storage medium for predicting customer churn, the method comprising: obtaining the transaction record of at least one customer in the current quarter; determining the customer's churn risk value according to the customer's transaction record and risk tolerance; inputting the first historical transaction record of the target customer into the target customer's analysis model to obtain the predicted transaction value of the target customer's next transaction; the target customer is a customer whose churn risk value is greater than a threshold; judging whether the target customer is a churn risk customer according to the predicted transaction value of the target customer's next transaction and the target customer's accumulated transaction value; if the target customer is a churn risk customer, providing a corresponding prevention plan for the churn risk customer according to the customer level of the churn risk customer and the risk tolerance of the churn risk customer. In order to achieve the purpose of accurately predicting the customers who may churn, and providing corresponding prevention plans to prevent the churn of the customers.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, device, electronic device and computer storage medium for predicting customer churn. Background Art

[0002] Currently, market competition is becoming increasingly fierce, and the differences in products or services among banks are becoming smaller and smaller. More and more banks are shifting from being "product-centric" to being "customer-centric".

[0003] Relevant research has found that the cost of winning a new customer is 5 to 6 times that of retaining an old customer. Therefore, retaining old customers, predicting potential lost customers in advance, and preventing business crises caused by customer loss are of strategic significance for improving the competitiveness of enterprises. Summary of the invention

[0004] In view of this, the present application provides a customer churn prediction method, device, electronic device and computer storage medium for accurately predicting customers who may churn and providing corresponding prevention plans to prevent the customer churn.

[0005] The first aspect of the present application provides a method for predicting customer churn, comprising:

[0006] Acquire transaction records of the current quarter; wherein the transaction records of the current quarter include transaction records of at least one customer in the current quarter;

[0007] For each of the customers, determining a transaction sentiment analysis result of the customer according to the customer's transaction record;

[0008] Combining the transaction sentiment analysis result and the risk tolerance, the customer's churn risk value is obtained; wherein the customer's churn risk value indicates the probability of the customer churn; the risk tolerance is determined based on the customer's account opening situation and transaction records;

[0009] Determine the customers whose churn risk value is greater than a threshold as target customers;

[0010] For each target customer, the first historical transaction record of the target customer is input into the analysis model of the target customer to obtain the predicted transaction value of the next transaction of the target customer; wherein the first historical transaction record of the target customer is all transaction records of the target customer up to the present; the analysis model is obtained by training a statistical regression model with the second historical transaction record of the target customer; the second historical transaction record of the target customer is the transaction record up to the time of training the statistical regression model;

[0011] Determining whether the target customer is a customer at risk of churn based on the predicted transaction value of the next transaction of the target customer and the accumulated transaction value of the target customer;

[0012] If it is determined that the target customer is a churn risk customer, a corresponding prevention plan is provided for the churn risk customer according to the customer level of the churn risk customer and the risk tolerance of the churn risk customer.

[0013] Optionally, the training method of the analysis model includes:

[0014] For each of the target customers, inputting the transaction information of the first N transactions in the second historical transaction record of the target customer into the statistical regression model to obtain the predicted transaction value of the N+1th transaction;

[0015] The statistical regression model is continuously adjusted by utilizing the error between the predicted transaction value of the N+1th transaction and the actual transaction value of the N+1th transaction, until the error between the predicted transaction value of the N+1th transaction output by the adjusted statistical regression model and the actual transaction value of the N+1th transaction meets a preset convergence condition, and the adjusted statistical regression model is determined as the analysis model for the target customer.

[0016] Optionally, after determining the adjusted statistical regression model as the analysis model of the target customer, the method further includes:

[0017] At a preset time point, determining the latest second historical transaction record of the target customer;

[0018] The analysis model of the target customer is retrained according to the latest second historical transaction record of the target customer.

[0019] Optionally, the risk tolerance determination method includes:

[0020] Obtaining the account opening information of the customer;

[0021] Determine the initial risk tolerance value based on the account opening information;

[0022] The sum of the weight value corresponding to the product information of each historical transaction of the customer and the initial risk tolerance value is used as the risk tolerance value of the customer;

[0023] The risk tolerance of the customer is determined according to a preset correspondence between the risk tolerance value and the risk tolerance; wherein, the higher the risk tolerance value, the stronger the risk tolerance.

[0024] Optionally, for each of the customers, determining a transaction sentiment analysis result of the customer according to the customer's transaction record includes:

[0025] Calculate the rate of return of each transaction of the customer and the average rate of return of the customer's historical transaction records;

[0026] If the rate of return of the customer shows a downward trend, the sentiment analysis result of the customer is determined according to the downward trend of the rate of return of the customer and the average rate of return of the customer's historical transaction records.

[0027] Optionally, the step of combining the transaction sentiment analysis result and risk tolerance to obtain the customer's churn risk value includes:

[0028] In the preset correspondence between risk tolerance - transaction sentiment analysis result - churn risk value, the transaction sentiment analysis result of the customer and the risk tolerance of the customer, and the corresponding churn risk value are found as the churn risk value of the customer.

[0029] Optionally, if it is determined that the target customer is a churn risk customer, after providing a corresponding prevention plan for the churn risk customer according to the customer level of the churn risk customer and the risk tolerance of the churn risk customer, the method further includes:

[0030] receiving feedback information from the at-risk customers regarding the prevention plan;

[0031] If the feedback information indicates that the customer at risk of churn agrees with the prevention plan, the churn risk level of the customer at risk of churn is lowered.

[0032] A second aspect of the present application provides a customer churn prediction device, comprising:

[0033] A first acquisition unit is used to acquire transaction records of the current quarter; wherein the transaction records of the current quarter include transaction records of at least one customer in the current quarter;

[0034] A first determining unit is used to determine, for each of the customers, a transaction sentiment analysis result of the customer according to the customer's transaction record;

[0035] The second determination unit is used to obtain the customer's churn risk value by combining the transaction sentiment analysis result and the risk tolerance; wherein the customer's churn risk value indicates the probability of the customer churn; and the risk tolerance is determined based on the customer's account opening situation and transaction records;

[0036] A third determining unit, configured to determine the customer whose churn risk value is greater than a threshold as a target customer;

[0037] An input unit is used for inputting the first historical transaction record of each target customer into the analysis model of the target customer to obtain a predicted transaction value of the next transaction of the target customer; wherein the first historical transaction record of the target customer is all transaction records of the target customer up to the present; the analysis model is obtained by training a statistical regression model with the second historical transaction record of the target customer; the second historical transaction record of the target customer is the transaction record up to the time of training the statistical regression model;

[0038] a judgment unit, configured to judge whether the target customer is a customer at risk of churn based on the predicted transaction value of the next transaction of the target customer and the accumulated transaction value of the target customer;

[0039] The prediction unit is used to provide a corresponding prevention plan for the target customer according to the customer level of the target customer and the risk tolerance of the target customer if the judgment unit judges that the target customer is a target customer.

[0040] Optionally, the training unit of the analysis model includes:

[0041] A second input unit is used to input, for each target customer, the transaction information of the first N transactions in the second historical transaction record of the target customer into the statistical regression model to obtain a predicted transaction value of the N+1th transaction;

[0042] An adjustment unit is used to continuously adjust the statistical regression model by utilizing the error between the predicted transaction value of the N+1th transaction and the actual transaction value of the N+1th transaction, until the error between the predicted transaction value of the N+1th transaction output by the adjusted statistical regression model and the actual transaction value of the N+1th transaction meets a preset convergence condition, and the adjusted statistical regression model is determined as the analysis model of the target customer.

[0043] Optionally, the customer churn prediction device further includes:

[0044] An updating data unit, used to determine the latest second historical transaction record of the target customer at a preset time point;

[0045] The training subunit of the analysis model is used to retrain the analysis model of the target customer based on the latest second historical transaction record of the target customer.

[0046] Optionally, the risk tolerance determination unit includes:

[0047] A second acquisition unit, used to acquire the account opening information of the customer;

[0048] an initial value determination unit, configured to determine an initial risk tolerance value according to the account opening information;

[0049] a risk tolerance value determination unit, configured to take the sum of the weight value corresponding to the product information of each historical transaction of the customer and the initial risk tolerance value as the risk tolerance value of the customer;

[0050] The risk tolerance determination subunit is used to determine the risk tolerance of the customer according to a preset correspondence relationship between the risk tolerance value and the risk tolerance; wherein, the higher the risk tolerance value, the stronger the risk tolerance.

[0051] Optionally, the first determining unit includes:

[0052] A calculation unit, used to calculate the rate of return of each transaction of the customer and the average rate of return of the customer's historical transaction records;

[0053] The first determination subunit is used to determine the sentiment analysis result of the customer according to the downward trend of the customer's yield rate and the average yield rate of the customer's historical transaction records if the customer's yield rate shows a downward trend.

[0054] Optionally, the second determining unit includes:

[0055] The second determination subunit is used to find the transaction sentiment analysis result of the customer and the risk tolerance of the customer, and the corresponding churn risk value in the preset correspondence relationship between risk tolerance-transaction sentiment analysis result-churn risk value as the churn risk value of the customer.

[0056] Optionally, the customer churn prediction device further includes:

[0057] A receiving unit, configured to receive feedback information from the at-risk customer regarding the prevention plan;

[0058] The lowering unit is configured to lower the churn risk level of the churn risk customer if the feedback information indicates that the churn risk customer agrees with the prevention plan.

[0059] A third aspect of the present application provides an electronic device, including:

[0060] one or more processors;

[0061] a storage device having one or more programs stored thereon;

[0062] When the one or more programs are executed by the one or more processors, the one or more processors implement the customer churn prediction method as described in any one of the first aspects.

[0063] A fourth aspect of the present application provides a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the customer churn prediction method as described in any one of the first aspects is implemented.

[0064] It can be seen from the above scheme that in a customer churn prediction method, device, electronic device and computer storage medium provided by the present application, the customer churn prediction method includes: obtaining the current quarter transaction record; wherein the current quarter transaction record includes the transaction record of at least one customer in the current quarter; then, for each of the customers, determining the transaction sentiment analysis result of the customer based on the customer's transaction record; and combining the transaction sentiment analysis result and risk tolerance to obtain the customer's churn risk value; wherein the customer's churn risk value indicates the probability of customer churn; the risk tolerance is determined based on the customer's account opening situation and transaction record; determining the customer whose churn risk value is greater than a threshold as a target customer; then, for each of the target customers, the first historical The transaction record is input into the analysis model of the target customer to obtain the predicted transaction value of the next transaction of the target customer; wherein, the first historical transaction record of the target customer is all the transaction records of the target customer up to the present; the analysis model is obtained by training the statistical regression model with the second historical transaction record of the target customer; the second historical transaction record of the target customer is the transaction record up to the time of training the statistical regression model; then, according to the predicted transaction value of the next transaction of the target customer and the cumulative transaction value of the target customer, it is judged whether the target customer is a customer at risk of churn; finally, if the target customer is judged to be a customer at risk of churn, a corresponding prevention plan is provided for the customer at risk of churn according to the customer level of the customer at risk of churn and the risk tolerance of the customer at risk of churn. In order to achieve the purpose of accurately predicting the customers who may be lost and providing corresponding prevention plans to prevent the loss of the customers. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0066] Figure 1 A specific flow chart of a method for predicting customer churn provided in an embodiment of the present application;

[0067] Figure 2A flowchart of a customer sentiment analysis provided in another embodiment of the present application;

[0068] Figure 3 A flow chart of a method for determining risk tolerance provided in another embodiment of the present application;

[0069] Figure 4 A flowchart of a training method for an analysis model provided in another embodiment of the present application;

[0070] Figure 5 A schematic diagram of a customer churn prediction method provided by another embodiment of the present application;

[0071] Figure 6 A schematic diagram of an electronic device for implementing a method for predicting customer churn provided in another embodiment of the present application. DETAILED DESCRIPTION

[0072] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 are within the scope of protection of the present invention.

[0073] It should be noted that the concepts of "first", "second", etc. mentioned in this application are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units, and the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0074] The present application embodiment provides a method for predicting customer churn, such as Figure 1 As shown, the specific steps include:

[0075] S101. Obtain transaction records for the current quarter.

[0076] The transaction records of the current quarter include the transaction records of at least one customer in the current quarter. The transaction records include the transaction identification of each transaction of the customer in the current quarter, the transaction record of each transaction, the profit and loss of each transaction, the transaction category of each transaction, the customer situation analysis results at each transaction, the proportion of the transaction funds of each transaction to the total amount, the transaction channel of each transaction, etc.

[0077] S102: For each customer, determine the customer's transaction sentiment analysis result based on the customer's transaction record.

[0078] Specifically, for each customer, based on the customer's transaction records, the sentiment analysis results of the customer during each transaction are determined.

[0079] Optionally, in another embodiment of the present application, an implementation of step S102 is as follows: Figure 2 As shown, including:

[0080] S201. Calculate the rate of return of each transaction of the customer and the average rate of return of the customer's historical transaction records.

[0081] S202. If the customer's rate of return is on a downward trend, determine the customer's sentiment analysis result based on the downward trend of the customer's rate of return and the average rate of return of the customer's historical transaction records.

[0082] It can be, but is not limited to, generating a coordinate graph based on the calculated rate of return of each transaction of the customer according to the transaction time. For example, the horizontal axis of the coordinate graph is set to the transaction time, and the vertical axis is set to the rate of return of the transaction. In this way, the slope of the customer's rate of return when it is on a downward trend can be obtained. There is no limitation here.

[0083] Specifically, after obtaining the slope of the customer's rate of return when it is on a downward trend, it is also necessary to combine the average rate of return of the customer's historical transaction records to determine the customer's sentiment analysis results. For example: when the customer's rate of return is higher than the average rate of return of the customer's historical transaction records, although the customer's rate of return is on a downward trend, the customer's emotions may still be able to accept the current situation and wait for the possibility of recovery, so the customer's sentiment analysis result is stable; when the customer's rate of return is lower than the average rate of return of the customer's historical transaction records, and the customer's rate of return is on a downward trend, then the customer's sentiment analysis result is low.

[0084] S103. Combining the transaction sentiment analysis results and risk tolerance, obtain the customer's churn risk value.

[0085] Among them, the customer's churn risk value indicates the probability of customer churn; the risk tolerance is determined based on the customer's account opening situation and transaction records.

[0086] Optionally, in another embodiment of the present application, an implementation of the risk tolerance determination method is as follows: Figure 3 As shown, including:

[0087] S301. Obtain the customer's account opening information.

[0088] S302. Determine the initial risk tolerance value based on the account opening information.

[0089] S303: The sum of the weight value corresponding to the product information of each historical transaction of the customer and the initial risk tolerance value is used as the risk tolerance value of the customer.

[0090] Among them, the product information may be deposit products, fund products, etc. It is understandable that the weight values ​​corresponding to different products may be the same or different, for example: the weight value of fund A is 7, the weight value of fund B is 8, the weight value of fund C is 9, the weight value of deposit A ​​is 2, the weight value of deposit B is 8, etc., which is not limited here.

[0091] It should be noted that the customer's risk tolerance value needs to be recalculated after each transaction.

[0092] S304. Determine the risk tolerance of the client according to the preset correspondence between the risk tolerance value and the risk tolerance.

[0093] It is understandable that the higher the risk tolerance value, the greater the risk tolerance.

[0094] Optionally, in another embodiment of the present application, an implementation of step S103 specifically includes:

[0095] In the preset correspondence between risk tolerance, transaction sentiment analysis result and churn risk value, the transaction sentiment analysis result of the customer and the risk tolerance of the customer are found, and the corresponding churn risk value is used as the churn risk value of the customer.

[0096] S104: Determine the customers whose churn risk value is greater than a threshold as target customers.

[0097] The threshold is set by the technical personnel and can be changed according to the actual situation, and is not limited here.

[0098] S105 . For each target customer, input the first historical transaction record of the target customer into the analysis model of the target customer to obtain a predicted transaction value of the next transaction of the target customer.

[0099] Among them, the first historical transaction record of the target customer is all transaction records of the target customer up to the present; the analysis model is obtained by training the statistical regression model with the second historical transaction record of the target customer; the second historical transaction record of the target customer is the transaction record up to the time of training the statistical regression model.

[0100] Optionally, in another embodiment of the present application, an implementation of the training method of the analysis model is as follows: Figure 4 As shown, the specific steps include:

[0101] S401. For each target customer, input the transaction information of the first N transactions in the second historical transaction record of the target customer into a statistical regression model to obtain a predicted transaction value of the N+1th transaction.

[0102] It should be noted that in the statistical regression model, the transaction information of the first N transactions in the second historical transaction record of the target customer will be preprocessed to generate corresponding feature values. The methods for preprocessing the transaction information of the first N transactions in the second historical transaction record of the target customer to generate corresponding feature values ​​are very diverse and mature. For example: encoding the transaction information of the first N transactions in the second historical transaction record of the target customer, eliminating outliers, filling the mean, expanding features, etc., and finally generating feature values, which will not be discussed here.

[0103] It is understandable that the statistical regression model can be trained for multiple rounds, for example, N is selected as 3 for the first time and N is selected as 7 for the second time. The more training rounds there are, the more accurate the analysis model obtained in the end is.

[0104] S402: Determine whether the error between the predicted transaction value of the N+1th transaction and the actual transaction value of the N+1th transaction satisfies a preset convergence condition.

[0105] Specifically, if the error between the predicted transaction value of the N+1th transaction and the actual transaction value of the N+1th transaction meets the preset convergence condition, step S404 is executed; if the error between the predicted transaction value of the N+1th transaction and the actual transaction value of the N+1th transaction does not meet the preset convergence condition, step S403 is executed.

[0106] S403: Using the error between the predicted transaction value of the N+1th transaction and the actual transaction value of the N+1th transaction, adjust the statistical regression model.

[0107] S404: Determine the statistical regression model as the analysis model for the target customers.

[0108] It is understandable that different time points and abnormal transactions may affect the customer's transaction situation. For example, on a wedding anniversary, a customer spends a lot of money to prepare gifts for his or her significant other. The transaction value may not match the predicted transaction value output by the analysis model, thereby causing inaccurate prediction analysis of lost customers. Therefore, in another embodiment of the present application, an implementation method after step S403 further includes:

[0109] At a preset time point, determine the target customer's latest second historical transaction record.

[0110] The preset time point may be, but is not limited to, holidays, anniversaries, etc., which are not limited here.

[0111] Retrain the analysis model of the target customer based on the target customer's latest second historical transaction record.

[0112] This will help us analyze lost customers more accurately.

[0113] S106: Determine whether the target customer is a customer at risk of churn based on the predicted transaction value of the next transaction of the target customer and the accumulated transaction value of the target customer.

[0114] Specifically, the absolute value of the difference between the predicted transaction value of the target customer's next transaction and the target customer's cumulative transaction value is compared with the preset risk transaction value. If it is greater than the preset risk transaction value, the target customer is judged to be a customer at risk of churn, and step S107 is executed; if it is greater than the preset risk transaction value, the target customer is judged to have no risk of churn, and no processing is performed; wherein, the preset risk transaction value is pre-set by technical personnel and can be changed according to actual conditions, and is not limited here.

[0115] S107. Provide corresponding prevention plans for the churn risk customers based on their customer levels and their risk tolerance.

[0116] Among them, customer level is obtained by the bank's rating of customers.

[0117] Specifically, the customer level of the customer at risk of churn and the corresponding prevention plan of the customer's risk tolerance can be queried in the corresponding relationship between customer level-risk tolerance-prevention plan, and then recommended to the customer at risk of churn. The prevention plan can be, but is not limited to, recommending specific financial products, timely informing of special offers, etc., so as to facilitate the customer's next normal investment transaction, thereby achieving the purpose of retaining the customer.

[0118] Optionally, in another embodiment of the present application, an implementation after step S107 further includes:

[0119] Receive feedback from customers at risk of churn regarding prevention plans.

[0120] If the feedback information indicates that the customer at risk of churn agrees to the prevention plan, the churn risk level of the customer at risk of churn will be lowered. If the feedback information indicates that the customer at risk of churn does not agree to the prevention plan, you can call the customer at risk of churn to inquire about the specific situation, and use communication to gain a deeper understanding of the customer at risk of churn so as to make a more correct strategy to achieve the purpose of customer retention. After successful communication, the churn risk level of the customer at risk of churn will be lowered.

[0121] It can be seen from the above scheme that a customer churn prediction method provided by the present application includes: obtaining the transaction records of the current quarter; wherein the transaction records of the current quarter include the transaction records of at least one customer in the current quarter; then, for each customer, determining the transaction sentiment analysis result of the customer based on the customer's transaction records; and combining the transaction sentiment analysis result and the risk tolerance to obtain the customer's churn risk value; wherein the customer's churn risk value indicates the probability of the customer churn; the risk tolerance is determined based on the customer's account opening situation and transaction records; determining the customer whose churn risk value is greater than the threshold as the target customer; then, for each target customer, inputting the first historical transaction record of the target customer into the target In the customer analysis model, the predicted transaction value of the target customer's next transaction is obtained; wherein, the first historical transaction record of the target customer is all the transaction records of the target customer up to now; the analysis model is obtained by training the statistical regression model with the second historical transaction record of the target customer; the second historical transaction record of the target customer is the transaction record up to the time of training the statistical regression model; then, according to the predicted transaction value of the target customer's next transaction and the cumulative transaction value of the target customer, it is judged whether the target customer is a customer at risk of churn; finally, if the target customer is judged to be a customer at risk of churn, a corresponding prevention plan is provided for the customer at risk of churn according to the customer level of the customer at risk of churn and the risk tolerance of the customer at risk of churn. In order to achieve the purpose of accurately predicting the customers who may churn and providing corresponding prevention plans to prevent the churn of the customer.

[0122] Another embodiment of the present application provides a device for predicting customer churn, such as Figure 5 As shown, specifically including:

[0123] The first acquisition unit 501 is used to acquire the transaction records of the current quarter.

[0124] The current quarter transaction records include transaction records of at least one customer in the current quarter.

[0125] The first determining unit 502 is used to determine, for each customer, a transaction sentiment analysis result of the customer according to the customer's transaction record.

[0126] Optionally, in another embodiment of the present application, an implementation of the first determining unit 502 includes:

[0127] The calculation unit is used to calculate the rate of return of each transaction of the customer and the average rate of return of the customer's historical transaction records.

[0128] The first determination subunit is used to determine the customer's sentiment analysis result according to the downward trend of the customer's yield rate and the average yield rate of the customer's historical transaction records if the customer's yield rate shows a downward trend.

[0129] For the specific working process of the units disclosed in the above embodiments of the present application, please refer to the corresponding method embodiments, such as Figure 2 As shown, no further description is given here.

[0130] The second determining unit 503 is used to obtain the customer's churn risk value by combining the transaction sentiment analysis result and the risk tolerance.

[0131] Among them, the customer's churn risk value indicates the probability of customer churn; the risk tolerance is determined based on the customer's account opening situation and transaction records.

[0132] Optionally, in another embodiment of the present application, an implementation of the risk tolerance determination unit includes:

[0133] The second acquisition unit is used to acquire the customer's account opening information.

[0134] The initial value determination unit is used to determine the initial risk tolerance value according to the account opening information.

[0135] The risk tolerance value determination unit is used to take the sum of the weight value corresponding to the product information of each historical transaction of the customer and the initial risk tolerance value as the risk tolerance value of the customer.

[0136] The risk tolerance determination subunit is used to determine the risk tolerance of the customer based on the preset correspondence between the risk tolerance value and the risk tolerance.

[0137] Among them, the higher the risk tolerance value, the stronger the risk tolerance.

[0138] For the specific working process of the units disclosed in the above embodiments of the present application, please refer to the corresponding method embodiments, such as Figure 3 As shown, no further description is given here.

[0139] Optionally, in another embodiment of the present application, an implementation of the second determining unit 503 includes:

[0140] The second determination subunit is used to find the customer's transaction sentiment analysis result and the customer's risk tolerance, and the corresponding churn risk value in the preset correspondence relationship between risk tolerance-transaction sentiment analysis result-churn risk value as the customer's churn risk value.

[0141] The specific working process of the units disclosed in the above embodiments of the present application can be found in the corresponding method embodiments, which will not be repeated here.

[0142] The third determining unit 504 is configured to determine customers whose churn risk value is greater than a threshold as target customers.

[0143] The input unit 505 is used to input the first historical transaction record of each target customer into the analysis model of the target customer to obtain the predicted transaction value of the next transaction of the target customer.

[0144] Among them, the first historical transaction record of the target customer is all transaction records of the target customer up to the present; the analysis model is obtained by training the statistical regression model with the second historical transaction record of the target customer; the second historical transaction record of the target customer is the transaction record up to the time of training the statistical regression model.

[0145] Optionally, in another embodiment of the present application, an implementation of a training unit of the analysis model includes:

[0146] The second input unit is used to input the transaction information of the first N transactions in the second historical transaction record of each target customer into the statistical regression model to obtain the predicted transaction value of the N+1th transaction.

[0147] The adjustment unit is used to continuously adjust the statistical regression model by using the error between the predicted transaction value of the N+1th transaction and the actual transaction value of the N+1th transaction, until the error between the predicted transaction value of the N+1th transaction output by the adjusted statistical regression model and the actual transaction value of the N+1th transaction meets a preset convergence condition, and the adjusted statistical regression model is determined as the analysis model for the target customer.

[0148] For the specific working process of the units disclosed in the above embodiments of the present application, please refer to the corresponding method embodiments, such as Figure 4 As shown, no further description is given here.

[0149] Optionally, in another embodiment of the present application, an implementation of the training unit of the analysis model further includes:

[0150] The update data unit is used to determine the latest second historical transaction record of the target customer at a preset time point.

[0151] The training subunit of the analysis model is used to retrain the analysis model of the target customer based on the latest second historical transaction record of the target customer.

[0152] The specific working process of the units disclosed in the above embodiments of the present application can be found in the corresponding method embodiments, which will not be repeated here.

[0153] The judgment unit 506 is used to judge whether the target customer is a customer at risk of churn based on the predicted transaction value of the next transaction of the target customer and the accumulated transaction value of the target customer.

[0154] The prediction unit 507 is used to provide a corresponding prevention plan for the churn risk customer according to the customer level of the churn risk customer and the risk tolerance of the churn risk customer if the judgment unit 506 judges that the target customer is a churn risk customer.

[0155] For the specific working process of the units disclosed in the above embodiments of the present application, please refer to the corresponding method embodiments, such as Figure 1 As shown, no further description is given here.

[0156] Optionally, in another embodiment of the present application, an implementation of the customer churn prediction device further includes:

[0157] The receiving unit is used to receive feedback information from customers at risk of churn regarding the prevention plan.

[0158] The lowering unit is used to lower the churn risk level of the churn risk customer if the feedback information shows that the churn risk customer agrees to the prevention plan.

[0159] The specific working process of the units disclosed in the above embodiments of the present application can be found in the corresponding method embodiments, which will not be repeated here.

[0160] As can be seen from the above scheme, a customer churn prediction device provided by the present application includes: a first acquisition unit 501 acquires the current quarter transaction record; wherein the current quarter transaction record includes the transaction record of at least one customer in the current quarter; then, the first determination unit 502 determines the customer's transaction sentiment analysis result for each customer based on the customer's transaction record; the second determination unit 503 combines the transaction sentiment analysis result and the risk tolerance to obtain the customer's churn risk value; wherein the customer's churn risk value indicates the probability of the customer churn; the risk tolerance is determined based on the customer's account opening situation and transaction record; the third determination unit 504 determines the customer whose churn risk value is greater than the threshold as the target customer; then, the input unit 505 determines the target customer's first A historical transaction record is input into the analysis model of the target customer to obtain the predicted transaction value of the next transaction of the target customer; wherein, the first historical transaction record of the target customer is all the transaction records of the target customer up to the present; the analysis model is obtained by training the statistical regression model with the second historical transaction record of the target customer; the second historical transaction record of the target customer is the transaction record up to the time of training the statistical regression model; the judgment unit 506 judges whether the target customer is a customer at risk of churn according to the predicted transaction value of the next transaction of the target customer and the cumulative transaction value of the target customer; finally, if the judgment unit 506 judges that the target customer is a customer at risk of churn, the prediction unit 507 provides a corresponding prevention plan for the customer at risk of churn according to the customer level of the customer at risk of churn and the risk tolerance of the customer at risk of churn. In order to achieve the purpose of accurately predicting the customers who may be lost and providing corresponding prevention plans to prevent the loss of the customers.

[0161] Another embodiment of the present application provides an electronic device, such as Figure 6 As shown, including:

[0162] One or more processors 601 .

[0163] The storage device 602 stores one or more programs.

[0164] When the one or more programs are executed by the one or more processors 601, the one or more processors 601 implement the customer churn prediction method as described in any one of the above embodiments.

[0165] Another embodiment of the present application provides a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the customer churn prediction method as described in any one of the above embodiments is implemented.

[0166] In the above embodiments disclosed in the present application, it should be understood that the disclosed apparatus and method can also be implemented in other ways. The apparatus and method embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the apparatus, method and computer program product according to multiple embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0167] In addition, each functional module in each embodiment of the present disclosure can be integrated together to form an independent part, or each module can exist separately, or two or more modules can be integrated to form an independent part. If the function is implemented in the form of a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present disclosure is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a live broadcast device, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present disclosure. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code.

[0168] The present application can be implemented or used by professionals and technicians. Various modifications to these embodiments will be apparent to professionals and technicians in the field, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting customer churn, characterized in that: include: Acquire transaction records of the current quarter; wherein the transaction records of the current quarter include transaction records of at least one customer in the current quarter; For each of the customers, calculate the rate of return of each transaction of the customer and the average rate of return of the customer's historical transaction records; If the rate of return of the customer shows a downward trend, then according to the downward trend of the rate of return of the customer and the average rate of return of the customer's historical transaction records, the transaction sentiment analysis result of the customer is determined; Combining the transaction sentiment analysis result and the risk tolerance, the churn risk value of the customer is obtained; wherein the churn risk value of the customer indicates the probability of the customer churn; the risk tolerance is obtained by obtaining the account opening information of the customer; determining the initial risk tolerance value according to the account opening information; taking the sum of the weight value corresponding to the product information of each historical transaction of the customer and the initial risk tolerance value as the risk tolerance value of the customer; determining the risk tolerance of the customer according to the preset corresponding relationship between the risk tolerance value and the risk tolerance; wherein the higher the risk tolerance value, the stronger the risk tolerance; Determine the customers whose churn risk value is greater than a threshold as target customers; For each target customer, the first historical transaction record of the target customer is input into the analysis model of the target customer to obtain the predicted transaction value of the next transaction of the target customer; wherein the first historical transaction record of the target customer is all transaction records of the target customer up to the present; the analysis model is obtained by training a statistical regression model with the second historical transaction record of the target customer; the second historical transaction record of the target customer is the transaction record up to the time of training the statistical regression model; The predicted transaction value of the next transaction of the target customer and the absolute value of the difference between the cumulative transaction value of the target customer are compared with the preset risk transaction value to determine whether the target customer is a customer at risk of churn; If it is determined that the absolute value of the difference between the predicted transaction value of the target customer's next transaction and the target customer's cumulative transaction value is greater than the preset risk transaction value, the target customer is determined to be a churn risk customer. If the target customer is a churn risk customer, a corresponding prevention plan is provided for the churn risk customer based on the customer level of the churn risk customer and the risk tolerance of the churn risk customer.

2. The prediction method according to claim 1, characterized in that: The training method of the analysis model comprises: For each of the target customers, inputting the transaction information of the first N transactions in the second historical transaction record of the target customer into the statistical regression model to obtain the predicted transaction value of the N+1th transaction; The statistical regression model is continuously adjusted by utilizing the error between the predicted transaction value of the N+1th transaction and the actual transaction value of the N+1th transaction, until the error between the predicted transaction value of the N+1th transaction output by the adjusted statistical regression model and the actual transaction value of the N+1th transaction meets a preset convergence condition, and the adjusted statistical regression model is determined as the analysis model for the target customer.

3. The prediction method according to claim 2, characterized in that: After determining the adjusted statistical regression model as the analysis model of the target customer, the method further includes: At a preset time point, determining the latest second historical transaction record of the target customer; The analysis model of the target customer is retrained according to the latest second historical transaction record of the target customer.

4. The prediction method according to claim 1, characterized in that: Combining the transaction sentiment analysis result and the risk tolerance to obtain the customer's churn risk value includes: In the preset correspondence between risk tolerance - transaction sentiment analysis result - churn risk value, the transaction sentiment analysis result of the customer and the risk tolerance of the customer, and the corresponding churn risk value are found as the churn risk value of the customer.

5. The prediction method according to claim 1, characterized in that: If it is determined that the target customer is a churn risk customer, after providing a corresponding prevention plan for the churn risk customer according to the customer level of the churn risk customer and the risk tolerance of the churn risk customer, the method further includes: receiving feedback information from the at-risk customers regarding the prevention plan; If the feedback information indicates that the customer at risk of churn agrees with the prevention plan, the churn risk level of the customer at risk of churn is lowered.

6. A customer churn prediction device, characterized in that: include: A first acquisition unit is used to acquire transaction records of the current quarter; wherein the transaction records of the current quarter include transaction records of at least one customer in the current quarter; A calculation unit, used for calculating, for each of the customers, the rate of return of each transaction of the customer and the average rate of return of the customer's historical transaction records; A first determining unit is configured to determine a transaction sentiment analysis result of the customer according to the downward trend of the customer's rate of return and the average rate of return of the customer's historical transaction records if the customer's rate of return is on a downward trend; The second determination unit is used to obtain the churn risk value of the customer by combining the transaction sentiment analysis result and the risk tolerance; wherein the churn risk value of the customer indicates the probability of the customer churn; the risk tolerance is determined according to the customer's account opening situation and transaction records; the risk tolerance is obtained by obtaining the customer's account opening information; determining the initial risk tolerance value according to the account opening information; taking the sum of the weight value corresponding to the product information of each historical transaction of the customer and the initial risk tolerance value as the risk tolerance value of the customer; determining the risk tolerance of the customer according to the preset correspondence relationship between the risk tolerance value and the risk tolerance; wherein the higher the risk tolerance value, the stronger the risk tolerance; A third determining unit, configured to determine the customer whose churn risk value is greater than a threshold as a target customer; An input unit is used for inputting the first historical transaction record of each target customer into the analysis model of the target customer to obtain a predicted transaction value of the next transaction of the target customer; wherein the first historical transaction record of the target customer is all transaction records of the target customer up to the present; the analysis model is obtained by training a statistical regression model with the second historical transaction record of the target customer; the second historical transaction record of the target customer is the transaction record up to the time of training the statistical regression model; A judgment unit is used to compare the absolute value of the difference between the predicted transaction value of the next transaction of the target customer and the accumulated transaction value of the target customer with a preset risk transaction value to judge whether the target customer is a customer at risk of churn; wherein, if the judgment unit judges that the absolute value of the difference between the predicted transaction value of the next transaction of the target customer and the accumulated transaction value of the target customer is greater than the preset risk transaction value, the target customer is determined to be a customer at risk of churn; The prediction unit is used to provide a corresponding prevention plan for the churn risk customer according to the customer level of the churn risk customer and the risk tolerance of the churn risk customer if the target customer is a churn risk customer.

7. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors implement the customer churn prediction method according to any one of claims 1 to 5.

8. A computer storage medium, characterized in that A computer program is stored thereon, wherein when the computer program is executed by a processor, the method for predicting customer churn as claimed in any one of claims 1 to 5 is implemented.

Images