Intervention measure acquisition method for high-loss-risk customers and related equipment

Abstract

The embodiment of the invention belongs to the technical field of intelligent decision making, and relates to an intervention measure obtaining method and related equipment for high-loss-risk customers, and the method comprises the steps: extracting feature subsets related to activities from customer attribute features, and dividing the feature subsets into an intervention group and a control group; a random forest Uplift model is trained for each feature subset; according to the trained random forest Uplift model, predicting and evaluating the increment effect of each intervention measure to obtain the increment effect; according to the comprehensive score of each user, constructing a personalized Uplift score matrix; and according to the personalized Uplift scoring matrix, performing priority ranking on the intervention measures of each customer, and determining an optimal intervention measure. According to the method, the high-loss risk group can be identified more accurately, and targeted intervention measures are taken to retain the customers. The customer satisfaction and loyalty can be improved, the customer wastage rate can be reduced, and the overall competitiveness of an enterprise is improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent decision-making technology, and in particular to a method for obtaining intervention measures for customers with high churn risk and related equipment. Background Art

[0002] In potential customer churn prediction, we face a key technical challenge: how to accurately evaluate the impact of different interventions on customer churn rate.

[0003] Traditional customer churn prediction models usually only focus on the customer's historical behavior and attribute characteristics, while ignoring the effect of the activity. This makes it impossible to effectively distinguish which customers are more sensitive to intervention measures, making it difficult to develop targeted retention strategies. Summary of the invention

[0004] The purpose of the embodiments of the present application is to propose a method and related equipment for obtaining intervention measures for customers with high churn risk, so as to solve the problem that it is difficult for traditional customer churn prediction models to formulate targeted retention strategies.

[0005] In order to solve the above technical problems, the embodiment of the present application provides a method for obtaining intervention measures for customers with high churn risk, which adopts the following technical solution:

[0006] Obtain customer attribute characteristics of high churn risk groups;

[0007] Extracting a feature subset related to the activity from the customer attribute features, wherein each feature subset corresponds to an intervention measure;

[0008] Each characteristic subset was divided into an intervention and a control group;

[0009] A random forest uplift model is trained for each feature subset to obtain several trained random forest uplift models;

[0010] According to the trained random forest Uplift model, the incremental effect of each intervention measure is predicted and evaluated to obtain the incremental effect;

[0011] For each user in the high churn risk group, the incremental effects of all the random forest Uplift models are integrated, and the comprehensive score of the current customer for each intervention measure is calculated by weighted average;

[0012] Build a personalized Uplift rating matrix based on each user's comprehensive rating;

[0013] Prioritize interventions for each client based on the personalized Uplift Scoring Matrix to determine the optimal intervention.

[0014] Furthermore, before the step of obtaining the customer attribute characteristics of the high churn risk group, the following steps are also included:

[0015] Obtain all customer historical behavior data and attribute feature data;

[0016] Constructing a customer churn prediction model based on a decision tree algorithm and the attribute feature data;

[0017] Perform prediction operations on all user data according to the churn prediction model to obtain churn probability information for each user;

[0018] The users corresponding to the churn probability information that meets the preset churn threshold are divided into the high churn risk group.

[0019] Furthermore, after the step of obtaining all the historical behavior data and attribute feature data of customers, the following steps are also included:

[0020] Preprocessing operations are performed on the historical behavior data and the attribute feature data.

[0021] Furthermore, the step of prioritizing intervention measures for each customer according to the personalized Uplift scoring matrix and determining the optimal intervention measure specifically includes the following steps:

[0022] Acquiring actual response data of users to the optimal intervention measures;

[0023] A model optimization operation is performed on the random forest Uplift model according to the actual response data to improve the accuracy of the personalized Uplift scoring matrix.

[0024] Furthermore, the step of prioritizing intervention measures for each customer according to the personalized Uplift scoring matrix and determining the optimal intervention measure specifically includes the following steps:

[0025] Clustering the users in the high churn risk group according to the K-means clustering algorithm and each user's personalized Uplift score matrix, and dividing customers with similar responses into the same customer subgroup;

[0026] For each customer subgroup, the Uplift score distribution of different intervention combinations is statistically analyzed to identify the optimal intervention measures that can maximize the positive response.

[0027] Furthermore, after the step of statistically analyzing the Uplift score distribution of different intervention combinations for each customer subgroup and identifying the optimal intervention measure that can maximize the positive response, the following step is also included:

[0028] Constructing an intervention combination strategy selection model based on the customer attribute characteristics of the high churn risk group and a decision tree algorithm;

[0029] When a new user arrives, the customer attribute characteristics of the new user are analyzed according to the intervention combination strategy selection model to obtain the optimal intervention combination and implement the intervention.

[0030] In order to solve the above technical problems, the embodiment of the present application further provides a device for obtaining intervention measures for customers with high churn risk, which adopts the following technical solution:

[0031] Customer attribute acquisition module, used to obtain customer attribute characteristics of high churn risk groups;

[0032] A feature subset acquisition module, used to extract a feature subset related to the activity from the customer attribute features, wherein each feature subset corresponds to an intervention measure;

[0033] The feature subset partitioning module is used to divide each feature subset into an intervention group and a control group;

[0034] The Uplift model training module is used to train a random forest Uplift model for each feature subset to obtain several trained random forest Uplift models;

[0035] A prediction and evaluation module, used to predict and evaluate the incremental effect of each intervention measure according to the trained random forest Uplift model to obtain the incremental effect;

[0036] A comprehensive score calculation module is used to calculate the comprehensive score of the current customer for each intervention measure by weighted average based on the incremental effects of all the random forest Uplift models for each user in the high churn risk group;

[0037] Rating matrix building module, used to build a personalized Uplift rating matrix based on each user's comprehensive rating;

[0038] The optimal intervention measure acquisition module is used to prioritize the intervention measures for each customer according to the personalized Uplift scoring matrix and determine the optimal intervention measure.

[0039] Furthermore, the device also includes:

[0040] The full customer attribute acquisition module is used to obtain the full customer historical behavior data and attribute feature data;

[0041] A churn prediction model building module, used to build a customer churn prediction model based on a decision tree algorithm and the attribute feature data;

[0042] A churn prediction module, used to perform prediction operations on all user data according to the churn prediction model to obtain churn probability information for each user;

[0043] The high churn group classification module is used to classify users corresponding to the churn probability information that meets a preset churn threshold into the high churn risk group.

[0044] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0045] The method comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for obtaining intervention measures for customers with high churn risks as described above are implemented.

[0046] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0047] The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the method for obtaining intervention measures for customers with high churn risks as described above are implemented.

[0048] The present application provides a method for obtaining intervention measures for customers with high churn risk, including: obtaining customer attribute characteristics of a high churn risk group; extracting a feature subset related to the activity from the customer attribute characteristics, wherein each feature subset corresponds to an intervention measure; dividing each feature subset into an intervention group and a control group; training a random forest Uplift model for each feature subset to obtain several trained random forest Uplift models; predicting and evaluating the incremental effect of each intervention measure according to the trained random forest Uplift model to obtain the incremental effect; for each user in the high churn risk group, combining the incremental effects of all the random forest Uplift models, calculating the current customer's comprehensive score for each intervention measure by weighted average; constructing a personalized Uplift score matrix according to the comprehensive score of each user; and prioritizing the intervention measures for each customer according to the personalized Uplift score matrix to determine the optimal intervention measure. Compared with the prior art, the present application can more accurately identify high churn risk groups and take targeted intervention measures to retain these customers. This can not only improve customer satisfaction and loyalty, but also reduce customer churn rate and enhance the overall competitiveness of the enterprise. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0050] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0051] Figure 2 is a flowchart of a method for obtaining intervention measures for customers with high churn risk provided in an embodiment of the present application;

[0052] Figure 3 It is a structural schematic diagram of a device for obtaining intervention measures for customers with high churn risk provided in an embodiment of the present application;

[0053] Figure 4 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0054] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0055] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0056] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0057] like Figure 1As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0058] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0059] The terminal device 101 can be any electronic device with a display screen and supporting web browsing. In addition to the laptop computer 1011, tablet computer 1012 or mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV), a laptop computer, a desktop computer, etc.

[0060] The server 103 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal device 101 .

[0061] It should be noted that the method for obtaining intervention measures for high-churn risk customers provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the device for obtaining intervention measures for high-churn risk customers is generally set in the server / terminal device.

[0062] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.

[0063] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for obtaining intervention measures for high-loss risk customers according to the present application. The method for obtaining intervention measures for high-loss risk customers includes: step S201, step S202, step S203, step S204, step S205, step S206, step S207 and step S208.

[0064] In step S201, customer attribute characteristics of a high churn risk group are obtained.

[0065] In step S202, feature subsets related to the activity are extracted from the customer attribute features, wherein each feature subset corresponds to an intervention measure.

[0066] In step S203, each feature subset is divided into an intervention group and a control group.

[0067] In step S204, a random forest Uplift model is trained for each feature subset to obtain a plurality of trained random forest Uplift models.

[0068] In step S205, the incremental effect of each intervention measure is predicted and evaluated based on the trained random forest Uplift model to obtain the incremental effect.

[0069] In step S206, for each user in the high churn risk group, the incremental effects of all random forest Uplift models are integrated, and the current customer's comprehensive score for each intervention measure is calculated by weighted average.

[0070] In step S207, a personalized Uplift score matrix is ​​constructed according to the comprehensive score of each user.

[0071] In step S208, the intervention measures for each customer are prioritized according to the personalized Uplift scoring matrix to determine the optimal intervention measure.

[0072] In the embodiment of the present application, it is necessary to identify which customer groups have a higher risk of churn through data analysis, which is usually based on historical data, such as the customer's purchase history, service usage, account activity records, etc.

[0073] In an embodiment of the present application, a churn risk prediction model can be constructed through machine learning algorithms (such as logistic regression, decision trees, etc.) or statistical methods to identify groups with high churn risk.

[0074] In the embodiment of the present application, once these high-risk groups are identified, it is necessary to collect their customer attribute characteristics, such as age, gender, income level, geographic location, usage habits, etc.

[0075] In the embodiment of the present application, among the collected customer attribute features, features directly related to the intervention activities need to be screened out. These features should be able to reflect the customer's potential response to different intervention measures. For example, if the intervention measure is to provide coupons, then the customer's purchase frequency, average consumption amount and other features may be directly related to the response rate.

[0076] In the embodiment of the present application, each feature subset corresponds to an intervention measure, so that the effects of different intervention measures can be evaluated based on different feature subsets.

[0077] In the embodiment of the present application, for each feature subset, the customers therein need to be randomly divided into an intervention group and a control group. The intervention group will receive specific intervention measures, while the control group will not receive any intervention as a baseline comparison.

[0078] In the present application embodiment, for each feature subset (i.e., each intervention measure), an Uplift model is trained using a random forest algorithm. The Uplift model is a special machine learning model that is used to evaluate the incremental effect of intervention measures on the target variable (i.e., the difference between the intervention group and the control group). By training these models, the potential impact of different intervention measures on customers in each feature subset can be predicted.

[0079] In the present application embodiment, the trained random forest Uplift model is used to predict and evaluate the incremental effect of each intervention measure. This usually involves calculating the difference between the intervention group and the control group in the target variable (such as retention rate, purchase amount, etc.).

[0080] In the embodiment of the present application, for each user in the high churn risk group, the incremental effects of all random forest Uplift models are combined. The current customer's comprehensive score for each intervention measure is calculated by weighted average. This score reflects the customer's potential response to different intervention measures.

[0081] In the embodiment of the present application, a personalized Uplift score matrix is ​​constructed based on the comprehensive score of each user. This matrix shows each customer's score and priority for different intervention measures.

[0082] In the embodiment of the present application, the intervention measures for each customer are prioritized according to the personalized Uplift scoring matrix. The intervention measure with the highest score or the most significant incremental effect is selected as the optimal intervention measure. This helps companies develop personalized retention strategies and improve the effectiveness and efficiency of intervention measures.

[0083] In an embodiment of the present application, a method for obtaining intervention measures for customers with high churn risk is provided, including: obtaining customer attribute characteristics of a high churn risk group; extracting a feature subset related to the activity from the customer attribute characteristics, wherein each feature subset corresponds to an intervention measure; dividing each feature subset into an intervention group and a control group; training a random forest Uplift model for each feature subset to obtain several trained random forest Uplift models; predicting and evaluating the incremental effect of each intervention measure according to the trained random forest Uplift model to obtain the incremental effect; for each user in the high churn risk group, combining the incremental effects of all random forest Uplift models, calculating the current customer's comprehensive score for each intervention measure by weighted average; constructing a personalized Uplift score matrix according to the comprehensive score of each user; and prioritizing the intervention measures for each customer according to the personalized Uplift score matrix to determine the optimal intervention measure. Compared with the prior art, the present application can more accurately identify high churn risk groups and take targeted intervention measures to retain these customers. This can not only improve customer satisfaction and loyalty, but also reduce customer churn rate and enhance the overall competitiveness of the enterprise.

[0084] In some optional implementations of the embodiments of the present application, before the step of obtaining the customer attribute characteristics of the high churn risk group, the following steps are also included:

[0085] Obtain all customer historical behavior data and attribute feature data;

[0086] Build a customer churn prediction model based on the decision tree algorithm and attribute feature data;

[0087] Perform prediction operations on all user data based on the churn prediction model to obtain the churn probability information of each user;

[0088] The users corresponding to the churn probability information that meets the preset churn threshold are divided into a high churn risk group.

[0089] In the embodiment of the present application, the full amount of customer historical behavior data refers to all interaction records between customers and the enterprise in the past period of time, such as purchase records, browsing records, customer service consultation records, complaint records, etc. These data can reflect the customer's consumption habits, preferences, and satisfaction with the enterprise's products or services.

[0090] In the embodiment of the present application, the attribute feature data includes static information of the customer, such as age, gender, occupation, income level, geographic location, etc. This information helps to understand the background and socioeconomic characteristics of the customer, thereby more comprehensively understanding the customer's behavior and churn risk.

[0091] In the embodiment of the present application, these data can be obtained through internal databases, customer relationship management systems (CRM), data warehouses, etc. At the same time, it is also possible to consider purchasing relevant data from third-party data providers to enrich and improve customer portraits.

[0092] In the embodiment of the present application, the decision tree constructs a model by recursively segmenting the data set. In customer churn prediction, the decision tree can predict whether a customer will churn based on the customer's attribute feature data.

[0093] In the embodiment of the present application, the prediction operation can input the full amount of user data into the trained churn prediction model, and the model will calculate the churn probability of each user based on the user's attribute feature data, where the churn probability information is a value between 0 and 1, indicating the possibility of user churn in the future. The closer the value is to 1, the higher the risk of user churn; the closer the value is to 0, the lower the risk of user churn.

[0094] In the embodiment of the present application, the preset churn threshold is mainly used to determine whether the churn risk of the user is high enough to classify it into a high churn risk group, wherein the classification operation refers to comparing the churn probability of each user with the preset churn threshold. If the churn probability of the user is greater than or equal to the threshold, it is classified into a high churn risk group; if the churn probability of the user is less than the threshold, it is regarded as a low-risk or normal user.

[0095] In the embodiment of the present application, it is possible to accurately identify which users have a higher risk of churn, so as to take targeted retention measures, improve customer satisfaction and loyalty, and reduce customer churn rate. At the same time, this also provides enterprises with valuable market insights and decision-making support, helping them to formulate more accurate and effective marketing strategies.

[0096] In some optional implementations of the embodiments of the present application, after the step of obtaining the full amount of customer historical behavior data and attribute feature data, the following steps are also included:

[0097] Perform preprocessing operations on historical behavior data and attribute feature data.

[0098] In the embodiment of the present application, through preprocessing operations, the quality of historical behavior data and attribute feature data can be ensured, providing a solid foundation for subsequent data analysis and machine learning model training.

[0099] In some optional implementations of the embodiments of the present application, the above-mentioned step of prioritizing the intervention measures for each customer according to the personalized Uplift scoring matrix and determining the optimal intervention measure specifically includes the following steps:

[0100] Obtain actual user response data for optimal intervention measures;

[0101] The random forest Uplift model is optimized based on the actual response data to improve the accuracy of the personalized Uplift score matrix.

[0102] In the embodiment of the present application, based on the previous analysis, each user is assigned the best intervention measures (such as coupons, customized recommendations, customer service follow-up, etc.). These measures are communicated to users through the company's marketing channels (such as emails, text messages, APP push, etc.).

[0103] In the embodiment of the present application, once the intervention measures are implemented, the enterprise needs to track the actual response of the user to these measures, including whether the user opens the pushed message, clicks on the link, uses the coupon, makes a purchase or other related behaviors.

[0104] In the embodiment of the present application, the user's response behavior will be recorded in the enterprise's data system, such as CRM system, order system, log system, etc. These data include response time, response type (such as click, purchase), response amount, etc.

[0105] In the embodiment of the present application, the model optimization operation of the random forest uplift model according to the actual response data may be:

[0106] ① Data integration: Integrate the collected actual response data with the previous user attribute feature data, historical behavior data, and the feature subset previously used to train the Uplift model. This ensures that the data used in the model optimization process is comprehensive and consistent.

[0107] ②Model evaluation: Use actual response data to evaluate the performance of the previously trained random forest Uplift model. This includes calculating the model's accuracy, recall, F1 score and other evaluation indicators, as well as comparing the differences between the intervention group and the control group through methods such as A / B testing.

[0108] ③ Feature importance analysis: Analyze which features are most important in predicting user responses. This helps companies understand which user attributes or behavior patterns have a significant impact on the effectiveness of interventions, thereby better designing interventions in the future.

[0109] ④Model Adjustment: Based on the evaluation results and feature importance analysis, adjust the Random Forest Uplift model. This may include adding or removing features, adjusting model parameters (such as the number of trees, maximum depth, etc.), using different data preprocessing techniques, etc.

[0110] ⑤ Retraining: Retrain the Random Forest Uplift model using the adjusted features and parameters. This step aims to improve the model’s predictive power, making it more accurate in predicting users’ responses to different interventions.

[0111] ⑥ Model validation: After retraining the model, use a portion of the actual response data that did not participate in the model training for validation to ensure that the performance of the model has improved after optimization.

[0112] In the embodiment of the present application, the random forest Uplift model can be continuously optimized to improve the accuracy of the personalized Uplift scoring matrix, thereby formulating more effective personalized marketing strategies, reducing customer churn rates, and improving customer satisfaction and loyalty.

[0113] In some optional implementations of the embodiments of the present application, the above-mentioned step of prioritizing the intervention measures for each customer according to the personalized Uplift scoring matrix and determining the optimal intervention measure specifically includes the following steps:

[0114] Cluster users in high churn risk groups based on the K-means clustering algorithm and each user's personalized Uplift score matrix, and divide customers with similar responses into the same customer subgroup;

[0115] For each client subgroup, the Uplift score distribution of different intervention combinations was statistically analyzed to identify the optimal intervention measures that can maximize the positive response.

[0116] In the embodiment of the present application, the clustering operation can be: first extract the user data of the high churn risk group, including their personalized Uplift score matrix. This matrix contains the user's scores for a variety of different interventions, reflecting the user's potential response to different interventions; determine the K value, which can be determined by business logic, data characteristics or experimental methods; use the K-means algorithm to cluster users, and the algorithm will assign users to K cluster centers (i.e., customer subgroups) according to the user's personalized Uplift score matrix, so that users in the same subgroup have similar Uplift score patterns; use indicators such as the Silhouette Coefficient and the Calinski-Harabasz Index to evaluate the quality of clustering to ensure that users are correctly divided into different subgroups.

[0117] In the embodiment of the present application, for each customer subgroup, their Uplift scores for different intervention combinations are collected. These scores reflect the potential response of users in the subgroup to specific intervention measures; for each intervention combination, the Uplift score distribution within the subgroup is calculated. This includes statistics such as the mean, median, and variance of the score to understand the performance of different intervention measures in the subgroup.

[0118] In the present application example, the effects of different interventions in subgroups are compared based on the distribution of Uplift scores. Generally, interventions with high mean and low variance are considered to be more effective because they can induce more stable and significant positive responses.

[0119] In the embodiment of the present application, for each customer subgroup, the intervention measure with the best Uplift score distribution is selected as the optimal intervention measure for the subgroup. This means that the measure can maximize the positive response in the subgroup and reduce the risk of customer churn.

[0120] In the embodiment of the present application, a personalized marketing strategy is formulated based on the optimal intervention measures for each customer subgroup. This includes determining the marketing channels, message content, discount strength, etc., to ensure that the intervention measures can accurately reach the target users and trigger a positive response.

[0121] In the embodiment of the present application, users in the high churn risk group can be more accurately identified and divided into different customer subgroups. For each subgroup, the enterprise can formulate personalized marketing strategies and select the best intervention measures to improve customer satisfaction and loyalty and reduce customer churn rate.

[0122] In some optional implementations of the embodiments of the present application, after the above step of statistically analyzing the Uplift score distribution of different intervention combinations for each customer subgroup and identifying the optimal intervention measure that can maximize the positive response, the following steps are also included:

[0123] Construct an intervention combination strategy selection model based on the customer attribute characteristics of the high churn risk group and the decision tree algorithm;

[0124] When a new user arrives, the customer attribute characteristics of the new user are analyzed according to the intervention combination strategy selection model, the optimal intervention combination is obtained and the intervention is implemented.

[0125] In the embodiment of the present application, an accurate intervention combination strategy selection model can be constructed, and personalized intervention measures can be provided for new users. This helps to reduce the risk of customer churn, improve customer satisfaction and loyalty, and thus enhance the overall competitiveness of the enterprise.

[0126] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0127] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0128] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0129] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0130] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of a device for obtaining intervention measures for customers with high churn risk. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0131] like Figure 3 As shown, the intervention measure acquisition device 200 for high churn risk customers in the embodiment of the present application includes:

[0132] A customer attribute acquisition module 210, used to acquire customer attribute characteristics of a high churn risk group;

[0133] A feature subset acquisition module 220 is used to extract a feature subset related to the activity from the customer attribute features, wherein each feature subset corresponds to an intervention measure;

[0134] A feature subset division module 230, for dividing each feature subset into an intervention group and a control group;

[0135] An Uplift model training module 240 is used to train a random forest Uplift model for each feature subset to obtain a plurality of trained random forest Uplift models;

[0136] The prediction and evaluation module 250 is used to predict and evaluate the incremental effect of each intervention measure according to the trained random forest Uplift model to obtain the incremental effect;

[0137] A comprehensive score calculation module 260 is used to calculate the comprehensive score of the current customer for each intervention measure by weighted average based on the incremental effects of all random forest Uplift models for each user in the high churn risk group;

[0138] A rating matrix construction module 270, for constructing a personalized Uplift rating matrix according to the comprehensive rating of each user;

[0139] The optimal intervention measure acquisition module 280 is used to prioritize the intervention measures for each customer according to the personalized Uplift scoring matrix and determine the optimal intervention measure.

[0140] In an embodiment of the present application, an intervention measure acquisition device 200 for high churn risk customers is provided, including: a customer attribute acquisition module 210, used to acquire customer attribute characteristics of a high churn risk group; a feature subset acquisition module 220, used to extract feature subsets related to activities from customer attribute characteristics, wherein each feature subset corresponds to an intervention measure; a feature subset division module 230, used to divide each feature subset into an intervention group and a control group; an Uplift model training module 240, used to train a random forest Uplift model for each feature subset, respectively, to obtain a plurality of trained random forest Uplift models; a prediction evaluation module 250; 0, used to predict and evaluate the incremental effect of each intervention measure based on the trained random forest Uplift model to obtain the incremental effect; comprehensive score calculation module 260, used to calculate the current customer's comprehensive score for each intervention measure by weighted average based on the incremental effects of all random forest Uplift models for each user in the high churn risk group; scoring matrix construction module 270, used to construct a personalized Uplift scoring matrix based on the comprehensive score of each user; optimal intervention measure acquisition module 280, used to prioritize the intervention measures for each customer based on the personalized Uplift scoring matrix, and determine the optimal intervention measure. Compared with the prior art, the present application can more accurately identify high churn risk groups and take targeted intervention measures to retain these customers. This can not only improve customer satisfaction and loyalty, but also reduce customer churn rate and enhance the overall competitiveness of the enterprise.

[0141] In some optional implementations of the embodiments of the present application, the above-mentioned intervention measure acquisition device 200 for high churn risk customers further includes:

[0142] The full customer attribute acquisition module is used to obtain the full customer historical behavior data and attribute feature data;

[0143] A churn prediction model building module is used to build a customer churn prediction model based on a decision tree algorithm and attribute feature data;

[0144] The churn prediction module is used to perform prediction operations on all user data according to the churn prediction model to obtain the churn probability information of each user;

[0145] The high churn group classification module is used to classify users corresponding to the churn probability information that meets a preset churn threshold into a high churn risk group.

[0146] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of a computer device according to an embodiment of the present application.

[0147] The computer device 300 includes a memory 310, a processor 320, and a network interface 330 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 300 having components 310-330, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0148] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0149] The memory 310 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 310 can be an internal storage unit of the computer device 300, such as a hard disk or memory of the computer device 300. In other embodiments, the memory 310 can also be an external storage device of the computer device 300, such as a plug-in hard disk equipped on the computer device 300, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card, etc. Of course, the memory 310 can also include both the internal storage unit of the computer device 300 and its external storage device. In the embodiment of the present application, the memory 310 is generally used to store the operating system and various application software installed on the computer device 300, such as computer-readable instructions for obtaining intervention measures for high-risk customers. In addition, the memory 310 can also be used to temporarily store various data that have been output or are to be output.

[0150] The processor 320 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 320 is generally used to control the overall operation of the computer device 300. In the embodiment of the present application, the processor 320 is used to run the computer-readable instructions stored in the memory 310 or process data, such as running the computer-readable instructions of the method for obtaining intervention measures for high-risk customers.

[0151] The network interface 330 may include a wireless network interface or a wired network interface. The network interface 330 is generally used to establish a communication connection between the computer device 300 and other electronic devices.

[0152] The computer device provided by this application can more accurately identify high-risk groups for churn and take targeted intervention measures to retain these customers. This can not only improve customer satisfaction and loyalty, but also reduce customer churn rate and enhance the overall competitiveness of the enterprise.

[0153] The present application also provides another embodiment, namely, providing a computer-readable storage medium, which stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the method for obtaining intervention measures for customers with high churn risk as described above.

[0154] The computer-readable storage medium provided by the present application can more accurately identify high-risk groups for churn and take targeted intervention measures to retain these customers. This can not only improve customer satisfaction and loyalty, but also reduce customer churn rate and enhance the overall competitiveness of the enterprise.

[0155] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0156] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for obtaining intervention measures for customers with high churn risk, characterized in that: The steps include: Obtain customer attribute characteristics of high churn risk groups; Extracting a feature subset related to the activity from the customer attribute features, wherein each feature subset corresponds to an intervention measure; Each characteristic subset was divided into an intervention and a control group; A random forest uplift model is trained for each feature subset to obtain several trained random forest uplift models; According to the trained random forest Uplift model, the incremental effect of each intervention measure is predicted and evaluated to obtain the incremental effect; For each user in the high churn risk group, the incremental effects of all the random forest Uplift models are integrated, and the comprehensive score of the current customer for each intervention measure is calculated by weighted average; Build a personalized Uplift rating matrix based on each user's comprehensive rating; Prioritize interventions for each client based on the personalized Uplift Scoring Matrix to determine the optimal intervention.

2. The method for obtaining intervention measures for high-risk customers according to claim 1, characterized in that: Before the step of obtaining the customer attribute characteristics of the high churn risk group, the following steps are also included: Obtain all customer historical behavior data and attribute feature data; Constructing a customer churn prediction model based on a decision tree algorithm and the attribute feature data; Perform prediction operations on all user data according to the churn prediction model to obtain churn probability information for each user; The users corresponding to the churn probability information that meets the preset churn threshold are divided into the high churn risk group.

3. The method for obtaining intervention measures for high-risk customers according to claim 2, characterized in that: After the step of obtaining all the customer historical behavior data and attribute feature data, the following steps are also included: Preprocessing operations are performed on the historical behavior data and the attribute feature data.

4. The method for obtaining intervention measures for high-risk customers according to claim 1, characterized in that: The step of prioritizing intervention measures for each customer according to the personalized Uplift scoring matrix and determining the optimal intervention measure specifically includes the following steps: Acquiring actual response data of users to the optimal intervention measures; A model optimization operation is performed on the random forest Uplift model according to the actual response data to improve the accuracy of the personalized Uplift scoring matrix.

5. The method for obtaining intervention measures for high-risk customers according to claim 1, characterized in that: The step of prioritizing intervention measures for each customer according to the personalized Uplift scoring matrix and determining the optimal intervention measure specifically includes the following steps: Clustering the users in the high churn risk group according to the K-means clustering algorithm and each user's personalized Uplift score matrix, and dividing customers with similar responses into the same customer subgroup; For each customer subgroup, the Uplift score distribution of different intervention combinations is statistically analyzed to identify the optimal intervention measures that can maximize the positive response.

6. The method for obtaining intervention measures for high-risk customers according to claim 5, characterized in that: After the step of calculating the Uplift score distribution of different intervention combinations for each customer subgroup and identifying the optimal intervention measure that can maximize the positive response, the following steps are also included: Constructing an intervention combination strategy selection model based on the customer attribute characteristics of the high churn risk group and a decision tree algorithm; When a new user arrives, the customer attribute characteristics of the new user are analyzed according to the intervention combination strategy selection model to obtain the optimal intervention combination and implement the intervention.

7. A device for obtaining intervention measures for customers with high churn risk, characterized in that: include: Customer attribute acquisition module, used to obtain customer attribute characteristics of high churn risk groups; A feature subset acquisition module, used to extract a feature subset related to the activity from the customer attribute features, wherein each feature subset corresponds to an intervention measure; The feature subset partitioning module is used to divide each feature subset into an intervention group and a control group; The Uplift model training module is used to train a random forest Uplift model for each feature subset to obtain several trained random forest Uplift models; A prediction and evaluation module, used to predict and evaluate the incremental effect of each intervention measure according to the trained random forest Uplift model to obtain the incremental effect; A comprehensive score calculation module is used to calculate the comprehensive score of the current customer for each intervention measure by weighted average based on the incremental effects of all the random forest Uplift models for each user in the high churn risk group; Rating matrix building module, used to build a personalized Uplift rating matrix based on each user's comprehensive rating; The optimal intervention measure acquisition module is used to prioritize the intervention measures for each customer according to the personalized Uplift scoring matrix and determine the optimal intervention measure.

8. The device for obtaining intervention measures for high-loss risk customers according to claim 7, characterized in that: The device also includes: The full customer attribute acquisition module is used to obtain the full customer historical behavior data and attribute feature data; A churn prediction model building module, used to build a customer churn prediction model based on a decision tree algorithm and the attribute feature data; A churn prediction module, used to perform prediction operations on all user data according to the churn prediction model to obtain churn probability information for each user; The high churn group classification module is used to classify users corresponding to the churn probability information that meets a preset churn threshold into the high churn risk group.

9. A computer device comprising a memory and a processor, characterized in that: The memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for obtaining intervention measures for customers with high churn risks according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the method for obtaining intervention measures for customers with high churn risks according to any one of claims 1 to 6.

Images