Customer loss risk assessment method and device based on neural network model

Abstract

The invention discloses a customer loss risk assessment method and device based on a neural network model, and relates to the field of artificial intelligence or other related technical fields, and the method comprises the steps: obtaining the assessment data of a target customer, the assessment data at least comprises a user portrait, a journey map and a behavior data set, the journey map is used for describing the experience feeling of the target customer in the processes of purchasing financial products and using financial services; performing feature extraction on the user portrait and the journey map of the target customer to obtain a user portrait feature vector and a journey map feature vector; drawing a behavior graph based on the behavior data set of the target customer; and inputting the user portrait feature vector, the journey map feature vector and the behavior graph into a loss risk assessment model, and outputting a loss risk value of the target customer. According to the method and the device, the technical problem of relatively low accuracy of an assessment result caused by relatively single assessment data in a mode of performing loss risk assessment based on text mining or user portraits in related technologies is solved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, financial technology or other related technical fields, and in particular to a customer churn risk assessment method and device based on a neural network model. Background Art

[0002] With the rapid development of science and technology, the Internet era has fully penetrated into people's lives. Especially in the financial industry, from the initial offline single transaction scenario, it has developed to online transaction scenarios such as financial cards, Alipay, WeChat, microcredit, large loans, and even financial investment transaction scenarios including stock trading, insurance, and investment. In the financial industry, customers are the source of business growth, and customer churn is one of the main challenges facing financial institutions. At present, the customer churn rate in the financial industry is high, which has an adverse impact on the long-term development of financial institutions.

[0003] In the related technology, when predicting and evaluating the customer churn risk, the churn risk assessment can be based on text mining or user portraits, that is, analyzing the text data of customers on Internet platforms, online reviews and customer service conversations to understand the users' emotions, attitudes and opinions. By mining the users' language expressions, we can better understand the users' needs and preferences, and thus assess the users' churn risk. The user portrait-based method is to assess the churn risk based on the user's portrait retained in the financial institution. The above-mentioned method of assessing the churn risk based on text mining or user portraits has a low accuracy of the assessment result due to the relatively single assessment data.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiment of the present invention provides a customer churn risk assessment method and device based on a neural network model, so as to at least solve the technical problem in the related art that the assessment data is relatively single, resulting in low accuracy of the assessment result.

[0006] According to one aspect of an embodiment of the present invention, a customer churn risk assessment method based on a neural network model is provided, comprising: obtaining assessment data of a target customer, wherein the assessment data comprises at least: a user portrait, a journey map, and a behavior data set, wherein the journey map is used to describe the target customer's experience in the process of purchasing financial products and using financial services; performing feature extraction on the user portrait and the journey map of the target customer to obtain a user portrait feature vector and a journey map feature vector; drawing a behavior graph based on the behavior data set of the target customer; inputting the user portrait feature vector, the journey map feature vector, and the behavior graph into a churn risk assessment model, and outputting a churn risk value of the target customer, wherein the churn risk assessment model is a model pre-constructed based on a neural network model for assessing the churn risk of a customer.

[0007] Optionally, the user portrait feature vector, the journey map feature vector and the behavior graph are input into the churn risk assessment model, and the step of outputting the churn risk value of the target customer includes: inputting the journey map feature vector into the time series network layer of the churn risk assessment model, and outputting a purchase behavior feature vector, wherein the time series network layer is used to analyze the journey map feature vector according to the time series; inputting the behavior graph into the graph neural network layer of the churn risk assessment model, and outputting a behavior preference feature vector, wherein the graph neural network layer is used to learn the embedded representation of each node in the behavior graph; inputting the user portrait feature vector, the purchase behavior feature vector and the behavior preference feature vector into the feature fusion layer of the churn risk assessment model to obtain a fused feature vector; inputting the fused feature vector into the fully connected neural network of the churn risk assessment model, and outputting the churn risk value of the target customer.

[0008] Optionally, the step of drawing a behavior graph based on the behavior data of the target customer includes: traversing the behavior data in the behavior data set, determining the execution objects of the behavior data and the jump relationships between the execution objects, wherein the execution objects include at least: browsing pages and clicking links; and drawing a behavior graph with the execution objects as nodes and the jump relationships between the execution objects as boundaries.

[0009] Optionally, after inputting the user portrait feature vector, the journey map feature vector and the behavior graph into the churn risk assessment model, it also includes: outputting the recommendation probability value of the target customer through the churn risk assessment model; when the recommendation probability value of the target customer is greater than a pre-set recommendation probability threshold, issuing a recommendation task to the user terminal where the target customer is located, wherein the recommendation task is used to instruct the target customer to recommend a financial product.

[0010] Optionally, after outputting the churn risk value of the target customer, it also includes: comparing the churn risk value of the target customer with a pre-set churn risk threshold to obtain a comparison result; when the churn risk value is greater than the churn risk threshold, determining that the target customer has a churn risk, and generating a retention strategy for the target customer.

[0011] Optionally, before obtaining the evaluation data of the target customer, it also includes: collecting the target customer's financial product purchase information within the target time period, wherein the financial product purchase information includes at least: product type, purchase amount, number of purchases, purchase experience information, and the purchase experience information is used to characterize whether the purchased financial product meets the needs of the target customer; collecting the target customer's financial service usage information within the target time period, wherein the financial service usage information includes at least: service type, usage frequency, usage experience, and the usage experience is used to characterize whether the used financial services meet the needs of the target customer; based on the financial product purchase information and the financial service usage information, drawing the journey map from a time perspective.

[0012] Optionally, the step of constructing the churn risk assessment model includes: obtaining historical user portraits, historical journey maps, and historical behavior graphs drawn based on historical behavior data within a historical time period to obtain an initial sample data set; configuring a churn risk value and a recommended probability value for each initial sample data in the initial sample data set to obtain a sample set; splitting the sample set to obtain a training set and a test set; constructing an initial model, wherein the initial model includes at least: a time series network layer, a graph neural network layer, a feature fusion layer, and a fully connected neural network; iteratively training the initial model based on the training set to obtain an initial churn risk assessment model; testing the initial churn risk assessment model based on the test set to obtain a test result, and when the test result indicates that the initial churn risk assessment model passes the test, obtaining the final churn risk assessment model.

[0013] According to another aspect of an embodiment of the present invention, a customer churn risk assessment device based on a neural network model is also provided, including: an acquisition unit, used to acquire assessment data of a target customer, wherein the assessment data includes at least: a user portrait, a journey map, and a behavior data set, wherein the journey map is used to describe the experience of the target customer in the process of purchasing financial products and using financial services; an extraction unit, used to perform feature extraction on the user portrait and the journey map of the target customer to obtain a user portrait feature vector and a journey map feature vector; a drawing unit, used to draw a behavior graph based on the behavior data set of the target customer; and an output unit, used to input the user portrait feature vector, the journey map feature vector, and the behavior graph into a churn risk assessment model, and output the churn risk value of the target customer, wherein the churn risk assessment model is a model pre-constructed based on a neural network model for assessing the churn risk of a customer.

[0014] Optionally, the output unit includes: a first output module, used to input the journey map feature vector into the time series network layer of the churn risk assessment model, and output a purchase behavior feature vector, wherein the time series network layer is used to analyze the journey map feature vector according to the time series; a second output module, used to input the behavior graph into the graph neural network layer of the churn risk assessment model, and output a behavior preference feature vector, wherein the graph neural network layer is used to learn the embedded representation of each node in the behavior graph; a first acquisition module, used to input the user portrait feature vector, the purchase behavior feature vector and the behavior preference feature vector into the feature fusion layer of the churn risk assessment model to obtain a fused feature vector; a third output module, used to input the fused feature vector into the fully connected neural network of the churn risk assessment model, and output the churn risk value of the target customer.

[0015] Optionally, the drawing unit includes: a first determination module, used to traverse the behavior data in the behavior data set, determine the execution objects of the behavior data and the jump relationships between the execution objects, wherein the execution objects at least include: browsing pages, clicking links; a first drawing module, used to draw a behavior graph with the execution objects as nodes and the jump relationships between the execution objects as boundaries.

[0016] Optionally, the customer churn risk assessment based on the neural network model also includes: a fourth output module, used to output the recommendation probability value of the target customer through the churn risk assessment model; a first publishing module, used to publish a recommendation task to the user terminal where the target customer is located when the recommendation probability value of the target customer is greater than a preset recommendation probability threshold, wherein the recommendation task is used to instruct the target customer to recommend a financial product.

[0017] Optionally, the customer churn risk assessment based on the neural network model also includes: a first comparison module, used to compare the churn risk value of the target customer with a pre-set churn risk threshold to obtain a comparison result; a first generation module, used to determine that the target customer has a churn risk when the churn risk value is greater than the churn risk threshold, and generate a retention strategy for the target customer.

[0018] Optionally, the customer churn risk assessment based on the neural network model also includes: a first acquisition module, used to collect the target customer's financial product purchase information within a target time period, wherein the financial product purchase information includes at least: product type, purchase amount, number of purchases, purchase experience information, and the purchase experience information is used to characterize whether the purchased financial product meets the needs of the target customer; a second acquisition module, used to collect the target customer's financial service usage information within the target time period, wherein the financial service usage information includes at least: service type, frequency of use, and usage experience, and the usage experience is used to characterize whether the used financial services meet the needs of the target customer; a second drawing module, used to draw the journey map from a time perspective based on the financial product purchase information and the financial service usage information.

[0019] Optionally, the customer churn risk assessment based on the neural network model also includes: a second acquisition module, used to acquire historical user portraits, historical journey maps, and historical behavior graphs drawn based on historical behavior data within a historical time period to obtain an initial sample data set; a third acquisition module, used to configure a churn risk value and a recommendation probability value for each initial sample data in the initial sample data set to obtain a sample set; a first splitting module, used to split the sample set to obtain a training set and a test set; a first construction module, used to construct an initial model, wherein the initial model includes at least: a time series network layer, a graph neural network layer, a feature fusion layer, and a fully connected neural network; a first training module, used to iteratively train the initial model based on the training set to obtain an initial churn risk assessment model; a first testing module, used to test the initial churn risk assessment model based on the test set to obtain a test result, and when the test result indicates that the initial churn risk assessment model passes the test, the final churn risk assessment model is obtained.

[0020] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned customer churn risk assessment methods based on the neural network model.

[0021] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any one of the above-mentioned customer churn risk assessment methods based on a neural network model.

[0022] In the present application, the following steps are performed: obtaining evaluation data of target customers, wherein the evaluation data includes at least: user portrait, journey map, and behavior data set, wherein the journey map is used to describe the target customer's experience in the process of purchasing financial products and using financial services, then performing feature extraction on the target customer's user portrait and journey map to obtain a user portrait feature vector and a journey map feature vector, and drawing a behavior graph based on the target customer's behavior data set, and finally inputting the user portrait feature vector, the journey map feature vector, and the behavior graph into a churn risk assessment model to output the target customer's churn risk value, wherein the churn risk assessment model is a model pre-built based on a neural network model for assessing the customer's churn risk.

[0023] In the present application, based on multi-dimensional data such as user portraits, journey maps, and behavioral data as evaluation data, the churn risk value of the target customers is automatically output through a pre-built neural network model, so as to achieve the purpose of accurately quantifying the customer churn risk. The technical effect of evaluating the customer churn risk through multi-dimensional evaluation data and improving the accuracy of the evaluation results is achieved, thereby solving the technical problem in the related technology of churn risk assessment based on text mining or user portraits, where the evaluation data is relatively single, resulting in low accuracy of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0025] Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing a customer churn risk assessment method based on a neural network model is shown;

[0026] Figure 2 is a flow chart of an optional customer churn risk assessment method based on a neural network model according to an embodiment of the present invention;

[0027] Figure 3 is a schematic diagram of an optional customer churn risk assessment process based on a neural network model according to an embodiment of the present invention;

[0028] Figure 4is a schematic diagram of an optional customer churn risk assessment device based on a neural network model according to an embodiment of the present invention;

[0029] Figure 5 It is a hardware structure block diagram of an electronic device (or mobile device) that optionally executes a customer churn risk assessment method based on a neural network model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0031] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] It should be noted that the customer churn risk assessment method and device based on the neural network model in the present application can be used in the field of artificial intelligence. When the neural network model is used to assess the customer churn risk, it can also be used in any field except the field of artificial intelligence technology. When the neural network model is used to assess the customer churn risk, the present application does not limit the application field of the customer churn risk assessment method and device based on the neural network model.

[0033] It should be noted that the collected information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this application are information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data are in compliance with relevant laws, regulations and standards, necessary confidentiality measures are taken, and public order and good customs are not violated, and corresponding operation entrances are provided for users to choose to authorize or refuse. For example, an interface is set up between this system and relevant users or institutions to provide users with corresponding operation entrances for users to choose to agree or refuse the results of automated decision-making; if the user chooses to refuse, the expert decision-making process will be entered.

[0034] The following embodiments of the present invention can be applied to various churn risk assessment systems / applications / devices. The present invention comprehensively analyzes user portraits, user behavior preferences, and journey maps, automatically outputs the customer's churn risk value through a neural network model, quantifies the churn risk value, and performs comprehensive analysis through multi-dimensional data to improve the accuracy of the assessment results.

[0035] The present invention is described in detail below in conjunction with various embodiments.

[0036] Embodiment 1

[0037] According to an embodiment of the present invention, an embodiment of a customer churn risk assessment method based on a neural network model is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0038] The method embodiment provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 The hardware structure block diagram of a computer terminal (or mobile device) for implementing a customer churn risk assessment method based on a neural network model is shown. Figure 1 As shown, the computer terminal 10 (or mobile device) may include one or more (102a, 102b, ..., 102n are used to illustrate) processors 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0039] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10 (or mobile device). As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0040] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the customer churn risk assessment method based on the neural network model in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, the customer churn risk assessment method based on the neural network model is realized. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0041] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0042] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 (or mobile device).

[0043] Under the above operating environment, this application provides Figure 2 The customer churn risk assessment method based on the neural network model shown in the figure is implemented by a customer churn risk assessment system. Through multi-dimensional data collection, deep feature extraction, feature fusion and model training, a system that can accurately predict customer churn risk is constructed. This system can not only improve the accuracy of prediction, but also effectively reduce customer churn rate through real-time monitoring and timely intervention, improve the overall operational efficiency and customer satisfaction of financial services, and is suitable for credit card business of financial institutions and other scenarios.

[0044] Figure 2 is a flowchart of an optional method for assessing customer churn risk based on a neural network model according to an embodiment of the present invention. Figure 2 As shown, the method comprises the following steps:

[0045] Step S201, obtaining evaluation data of target customers.

[0046] It should be noted that in order to implement the assessment of customer churn risk, it is first necessary to obtain assessment data of target customers. The assessment data includes three parts: user portrait, journey map, and behavioral data set. User portrait contains basic information of customers, behavioral preferences, demand characteristics, interests and hobbies, consumption habits and other information. It is a virtual image description drawn by financial institutions for users based on data collected from multiple channels. Journey map is a visualization tool used to describe and display the experience and emotional changes of target customers in the process of purchasing financial products and using financial services. Journey map is user-centric and presents the interaction and emotional changes between users and products or services through timelines or step sequences. Based on journey maps, customers' goals and needs at each stage can be deeply understood.

[0047] Optionally, before obtaining the evaluation data of the target customers, it also includes: collecting the target customers' financial product purchase information within the target time period, wherein the financial product purchase information includes at least: product type, purchase amount, number of purchases, purchase experience information, and the purchase experience information is used to characterize whether the purchased financial products meet the needs of the target customers; collecting the target customers' financial service usage information within the target time period, wherein the financial service usage information includes at least: service type, usage frequency, and usage experience, and the usage experience is used to characterize whether the used financial services meet the needs of the target customers; and drawing a journey map from a time perspective based on the financial product purchase information and the financial service usage information.

[0048] Specifically, the journey map is pre-drawn based on the user's financial behavior. First, the target customer's financial product purchase information within the target time period is collected, including but not limited to product type, purchase amount, purchase frequency, and purchase experience information. Product type reflects the customer's preference for different financial products; purchase amount and purchase frequency help analyze the customer's consumption capacity and investment behavior; and purchase experience information is used to evaluate the customer's satisfaction with whether the purchased financial product meets their needs. The purpose of this step is to construct the customer's behavior pattern in the purchase of financial products and provide key time series data for subsequent journey map drawing.

[0049] Next, you need to collect information about the target customers’ use of financial services during the target time period, such as service type, frequency of use, and user experience. Service type covers the scope of financial institution services used by customers, frequency of use reflects the degree of interaction between customers and financial institutions, and user experience is used to measure customer satisfaction with financial services. With the above information, you can understand customers’ behavioral preferences and satisfaction when using financial services, providing a more detailed perspective for journey mapping.

[0050] Furthermore, time series data is constructed based on the above financial product purchase information and financial service usage information, and the time series data is analyzed to identify changes in customers' purchase behavior and usage experience at different time points. The journey map intuitively shows the process of customers' interaction with financial institutions, including the key stages that customers go through in the process of using credit cards and other financial products, as well as the satisfaction and behavioral characteristics at each stage. By drawing a journey map, we can better understand the feelings and experiences of customers throughout the process of using financial products or financial services.

[0051] Finally, the evaluation data also includes a collection of behavioral data. The behavioral data is obtained based on the customer's interactive behavior on the Internet, and may include page browsing behavior data and click behavior data, etc.

[0052] Step S202: extract features from the user portrait and journey map of the target customer to obtain a user portrait feature vector and a journey map feature vector.

[0053] It should be noted that for the evaluation data, preprocessing is first required to convert data of different formats into feature vectors that can be recognized by the neural network model. First, the basic information of the target customers is encoded, which includes converting discrete or continuous features such as age, gender, and occupation into vector forms that can be understood by the model. Specifically, for continuous variables such as age and income level, standardization can be performed to ensure numerical stability during model training; for categorical variables such as occupation, they are converted through unique hot encoding or embedding so that the model can capture feature information of different categories. In addition, the text description information in the user portrait (such as hobbies, personal preferences, etc.) can be converted into a fixed-dimensional vector using the word embedding model in natural language processing technology to reflect the semantic characteristics of the text. Finally, these encoded feature vectors will be spliced ​​or fused to form a user portrait feature vector.

[0054] Furthermore, features can be extracted from the journey map by using time series analysis to extract features such as the customer's purchase amount of financial products, purchase category, and purchase frequency from the time series data, thereby obtaining a journey map feature vector.

[0055] It should be noted that the user portrait feature vector reflects the basic attributes and static information of the customer, while the journey map feature vector captures the dynamic behavior pattern of the customer's interaction with financial services. The integration of the two can not only fully reflect the customer characteristics, but also reveal the changing trend of customer behavior over time, which is of great significance for predicting the potential risk of customer churn.

[0056] Step S203: drawing a behavior graph based on the behavior data set of the target customer.

[0057] It should be noted that for the customer's behavior data set, it can be analyzed based on graph neural networks. First, it is necessary to build a behavior graph. The behavior graph uses the browsed pages, clicked links, etc. as nodes, and the jump relationships between nodes as edges. The correlation between the content browsed and clicked by customers is presented in the form of an image.

[0058] Optionally, the step of drawing a behavior graph based on the behavior data of the target customers includes: traversing the behavior data in the behavior data set, determining the execution objects of the behavior data and the jump relationships between the execution objects, wherein the execution objects include at least: browsing pages and clicking links; and drawing the behavior graph with the execution objects as nodes and the jump relationships between the execution objects as boundaries.

[0059] Furthermore, by constructing a behavior graph, we can gain a deeper understanding of customers' interaction patterns and behavior preferences on the Internet, thereby providing richer behavioral features for the measurement of customer churn risk. First, we need to traverse the behavior data set, which contains all the behavior records of target customers within a certain time range. The purpose of the traversal is to identify each specific execution behavior, including but not limited to operations such as users browsing pages and clicking links. The system needs to record in detail the specific object of each execution behavior, as well as the context in which the execution behavior occurs, such as time, location, and page information before and after the operation. During the traversal process, the system will extract the execution objects of the behavior data, such as the specific pages visited by users, the links clicked, etc., and identify the jump relationships between these objects. The jump relationship reflects the user's browsing path and click order on the Internet, and is an important manifestation of the user's behavior preferences and habits. By determining the execution object and jump relationship, the user's interactive behavior pattern on the Internet can be captured, providing core data for subsequent behavior graph drawing. Finally, the determined execution objects are used as the nodes of the graph, and the jump relationships between the execution objects become the edges of the graph, constructing a behavior graph. Subsequently, the graph neural network can be used to learn the embedded representations of the nodes in the graph. These representations not only contain the characteristics of the nodes themselves, but also integrate the characteristics of the nodes connected to them, forming a comprehensive vector that can reflect user behavior patterns and preference relationships.

[0060] Step S204: input the user portrait feature vector, the journey map feature vector and the behavior graph into a churn risk assessment model, and output the churn risk value of the target customer.

[0061] It should be noted that the embodiment of the present invention identifies the churn risk of target customers through a pre-built churn risk assessment model, takes the user portrait feature vector, the journey map feature vector and the behavior graph as input data, and analyzes the multi-dimensional assessment data through multiple modules in the churn risk assessment model, thereby predicting the churn risk value of the target customers, quantifying the potential churn risk into a specific prediction value, so that the output result is intuitive and reliable.

[0062] Optionally, the step of constructing a churn risk assessment model includes: obtaining historical user portraits and historical journey maps within a historical time period, and historical behavior graphs drawn based on historical behavior data to obtain an initial sample data set; configuring a churn risk value and a recommended probability value for each initial sample data in the initial sample data set to obtain a sample set; splitting the sample set to obtain a training set and a test set; constructing an initial model, wherein the initial model includes at least: a time series network layer, a graph neural network layer, a feature fusion layer, and a fully connected neural network; iteratively training the initial model based on the training set to obtain an initial churn risk assessment model; testing the initial churn risk assessment model based on the test set to obtain test results, and when the test results indicate that the initial churn risk assessment model passes the test, obtaining a final churn risk assessment model.

[0063] It should be noted that the above-mentioned churn risk assessment model is a model pre-built based on a neural network model for assessing the churn risk of customers. Specifically, when constructing a churn risk assessment model, it is first necessary to prepare sample data, including historical user portraits, historical journey maps, and historical behavior graphs of each customer in a historical time period, and then configure labels for each sample data, including churn risk values ​​and recommendation probability values. The recommendation probability value indicates the intensity of the customer's willingness to recommend financial products or financial services. A sample set is constructed based on the historical user portraits, historical journey maps, and historical behavior graphs and their corresponding churn risk values ​​and recommendation probability values. The sample set is then split to obtain a training set and a test set. At the same time, an initial model is constructed through a neural network framework, including: a time series network layer for extracting time series features, a graph neural network layer for extracting features from images, a feature fusion layer for fusing multiple feature vectors, and a fully connected neural network for predicting the churn risk value of the target customer. Then, the initial model is continuously iterated and trained according to the training set, and the iteration is stopped when the maximum number of iterations is reached. The trained model is tested through the test set, and a trained churn risk assessment model is obtained if the test passes.

[0064] Optionally, the step of inputting the user portrait feature vector, the journey map feature vector and the behavior graph into the churn risk assessment model, and outputting the churn risk value of the target customer includes: inputting the journey map feature vector into the time series network layer of the churn risk assessment model, and outputting the purchase behavior feature vector, wherein the time series network layer is used to analyze the journey map feature vector according to the time series; inputting the behavior graph into the graph neural network layer of the churn risk assessment model, and outputting the behavior preference feature vector, wherein the graph neural network layer is used to learn the embedded representation of each node in the behavior graph; inputting the user portrait feature vector, the purchase behavior feature vector and the behavior preference feature vector into the feature fusion layer of the churn risk assessment model to obtain a fused feature vector; inputting the fused feature vector into the fully connected neural network of the churn risk assessment model, and outputting the churn risk value of the target customer.

[0065] It should be noted that when predicting the churn risk of target customers through the churn risk assessment model, the input journey map feature vector and behavior graph are first processed synchronously through the time series network layer and the graph neural network layer. The journey map feature vector is processed through the time series network layer to capture the time dependency in the feature vector to obtain the purchase behavior feature vector. The graph neural network layer learns the embedded representation of each node in the behavior graph to obtain the behavior preference feature vector. Then, the input user portrait feature vector, the purchase behavior feature vector output by the time series network layer, and the behavior preference feature vector output by the graph neural network layer are input into the feature fusion layer for feature fusion to obtain the fused feature vector. Finally, the fused feature vector is input into the fully connected neural network, and after multiple hidden layers for nonlinear transformation and further feature refinement, the final output layer uses an activation function to output the churn risk value.

[0066] Optionally, after inputting the user portrait feature vector, the journey map feature vector and the behavior graph into the churn risk assessment model, it also includes: outputting the recommendation probability value of the target customer through the churn risk assessment model; when the recommendation probability value of the target customer is greater than a pre-set recommendation probability threshold, issuing a recommendation task to the user terminal where the target customer is located, wherein the recommendation task is used to instruct the target customer to recommend a financial product.

[0067] It should be noted that in addition to outputting the churn risk value, the churn risk assessment model can also predict the target user's recommendation intention and output a recommendation probability value. If the target customer's recommendation probability value is high, it indicates that the client's recommendation intention is high, and a recommendation task containing rewards can be issued to the customer, thereby completing the promotion of financial products or financial services through customer channels.

[0068] Optionally, after outputting the target customer's churn risk value, the method further includes: comparing the target customer's churn risk value with a preset churn risk threshold to obtain a comparison result; when the churn risk value is greater than the churn risk threshold, determining that the target customer has a churn risk and generating a retention strategy for the target customer.

[0069] It should be noted that after outputting the target customer's churn risk value, the nature of the target customer is determined by threshold comparison, that is, whether the target customer is a potential churn customer. For the target customer determined to be at risk of churn, the system will generate a personalized retention strategy based on their specific characteristics and behavior patterns. This may include sending customized preferential information, providing more personalized customer service, or optimizing customer-related credit card product features. The purpose of generating a retention strategy is to improve customer satisfaction, reduce customer churn rate, and enhance customer loyalty by providing services and products that are closer to customer needs.

[0070] Through the above steps, the evaluation data of the target customer is obtained, wherein the evaluation data at least includes: user portrait, journey map, and behavior data set. The journey map is used to describe the experience of the target customer in the process of purchasing financial products and using financial services. Then, the user portrait and journey map of the target customer are feature extracted to obtain the user portrait feature vector and the journey map feature vector, and a behavior graph is drawn based on the behavior data set of the target customer. Finally, the user portrait feature vector, the journey map feature vector and the behavior graph are input into the churn risk assessment model, and the churn risk value of the target customer is output. The churn risk assessment model is a model pre-built based on the neural network model for assessing the churn risk of customers.

[0071] In this embodiment, based on multi-dimensional data such as user portraits, journey maps, and behavioral data as evaluation data, the churn risk value of the target customer is automatically output through a pre-built neural network model, so as to achieve the purpose of accurately quantifying the customer churn risk, and achieve the technical effect of evaluating the customer churn risk through multi-dimensional evaluation data and improving the accuracy of the evaluation results, thereby solving the technical problem in the related technology of churn risk assessment based on text mining or user portraits, where the evaluation data is relatively single, resulting in low accuracy of the evaluation results.

[0072] Another optional specific implementation is described in detail below.

[0073] Figure 3 FIG. 1 is a schematic diagram of an optional customer churn risk assessment process based on a neural network model according to an embodiment of the present invention. Figure 3 As shown in the figure, when evaluating the customer churn risk based on the neural network model, it specifically includes:

[0074] Step 1: Obtain evaluation data, including user portraits, journey maps, and behavior maps;

[0075] Step 2: extract features from the user portrait through a portrait feature extraction module to obtain a user portrait feature vector;

[0076] Step 3: Input the journey map and behavior graph into the churn risk assessment model, extract features from the journey map through the time series network layer, output the purchase behavior feature vector, and extract features from the behavior graph through the graph neural network layer to obtain the behavior preference feature vector;

[0077] Step 4: The feature fusion layer combines the input feature vectors of the user portrait feature vector, the purchase behavior feature vector, and the behavior preference feature vector to form a fused feature vector. The feature fusion layer concatenates the input feature vectors, or uses an attention mechanism to dynamically assign weights to the features according to their importance for the prediction of the churn risk, thereby forming a fused feature vector.

[0078] Step 5: Input the fused feature vector into the fully connected neural network, and after multiple hidden layers for nonlinear transformation and further feature extraction, the final output layer uses a specific activation function (because the predicted churn risk value and recommendation probability value are both probability values, ranging from 0 to 1), and outputs two results: one is the churn risk value (a value between 0 and 1, the closer to 1, the greater the churn risk), and the other is the user recommendation willingness score (also a value between 0 and 1, the closer to 1, the stronger the recommendation willingness).

[0079] In the embodiment of the present invention, user portraits, user behavior preferences and journey maps are comprehensively analyzed, and the customer's churn risk value is automatically output through a neural network model, the churn risk value is quantified, and a comprehensive analysis is performed through multi-dimensional data to improve the accuracy of the evaluation results.

[0080] The following is a detailed description in conjunction with another embodiment.

[0081] Embodiment 2

[0082] A customer churn risk assessment system based on a neural network model provided in this embodiment includes multiple implementation units, each of which corresponds to each implementation step in the above-mentioned embodiment 1. Its specific implementation methods and beneficial effects can refer to the aforementioned method embodiments and will not be repeated here.

[0083] Figure 4 is a schematic diagram of an optional customer churn risk assessment device based on a neural network model according to an embodiment of the present invention. Figure 4 As shown, the customer churn risk assessment device based on the neural network model may include: an acquisition unit 41, an extraction unit 42, a drawing unit 43, and an output unit 44, wherein:

[0084] An acquisition unit 41 is used to acquire evaluation data of a target customer, wherein the evaluation data includes at least: a user portrait, a journey map, and a behavior data set, wherein the journey map is used to describe the experience of the target customer in the process of purchasing financial products and using financial services;

[0085] An extraction unit 42 is used to extract features from the user portrait and journey map of the target customer to obtain a user portrait feature vector and a journey map feature vector;

[0086] A drawing unit 43, used for drawing a behavior graph based on the behavior data set of the target customer;

[0087] The output unit 44 is used to input the user portrait feature vector, the journey map feature vector and the behavior graph into the churn risk assessment model, and output the churn risk value of the target customer, wherein the churn risk assessment model is a model pre-built based on the neural network model for assessing the churn risk of customers.

[0088] The above-mentioned customer churn risk assessment device based on the neural network model obtains the assessment data of the target customer through the acquisition unit 41, wherein the assessment data at least includes: user portrait, journey map, and behavior data set, and the journey map is used to describe the experience of the target customer in the process of purchasing financial products and using financial services; the user portrait and journey map of the target customer are subjected to feature extraction through the extraction unit 42 to obtain the user portrait feature vector and the journey map feature vector; the behavior graph is drawn based on the behavior data set of the target customer through the drawing unit 43; the user portrait feature vector, the journey map feature vector and the behavior graph are input into the churn risk assessment model through the output unit 44, and the churn risk value of the target customer is output, wherein the churn risk assessment model is a model pre-constructed based on the neural network model for assessing the churn risk of customers.

[0089] In this embodiment, based on multi-dimensional data such as user portraits, journey maps, and behavioral data as evaluation data, the churn risk value of the target customer is automatically output through a pre-built neural network model, so as to achieve the purpose of accurately quantifying the customer churn risk, and achieve the technical effect of evaluating the customer churn risk through multi-dimensional evaluation data and improving the accuracy of the evaluation results, thereby solving the technical problem in the related technology of churn risk assessment based on text mining or user portraits, where the evaluation data is relatively single, resulting in low accuracy of the evaluation results.

[0090] Optionally, the output unit 44 includes: a first output module, used to input the journey map feature vector into the time series network layer of the churn risk assessment model, and output the purchase behavior feature vector, wherein the time series network layer is used to analyze the journey map feature vector according to the time series; a second output module, used to input the behavior graph into the graph neural network layer of the churn risk assessment model, and output the behavior preference feature vector, wherein the graph neural network layer is used to learn the embedded representation of each node in the behavior graph; a first acquisition module, used to input the user portrait feature vector, the purchase behavior feature vector and the behavior preference feature vector into the feature fusion layer of the churn risk assessment model to obtain a fused feature vector; a third output module, used to input the fused feature vector into the fully connected neural network of the churn risk assessment model, and output the churn risk value of the target customer.

[0091] Optionally, the drawing unit 43 includes: a first determination module, used to traverse the behavior data in the behavior data set, determine the execution objects of the behavior data and the jump relationships between the execution objects, wherein the execution objects at least include: browsing pages, clicking links; a first drawing module, used to draw a behavior graph with the execution objects as nodes and the jump relationships between the execution objects as boundaries.

[0092] Optionally, the customer churn risk assessment based on the neural network model also includes: a fourth output module, used to output the recommendation probability value of the target customer through the churn risk assessment model; a first publishing module, used to publish a recommendation task to the user terminal where the target customer is located when the recommendation probability value of the target customer is greater than a preset recommendation probability threshold, wherein the recommendation task is used to instruct the target customer to recommend a financial product.

[0093] Optionally, the customer churn risk assessment based on the neural network model also includes: a first comparison module, used to compare the churn risk value of the target customer with a pre-set churn risk threshold to obtain a comparison result; a first generation module, used to determine that the target customer has a churn risk when the churn risk value is greater than the churn risk threshold, and generate a retention strategy for the target customer.

[0094] Optionally, the customer churn risk assessment based on the neural network model also includes: a first acquisition module, used to collect financial product purchase information of target customers within a target time period, wherein the financial product purchase information includes at least: product type, purchase amount, number of purchases, purchase experience information, and the purchase experience information is used to characterize whether the purchased financial products meet the needs of target customers; a second acquisition module, used to collect financial service usage information of target customers within a target time period, wherein the financial service usage information includes at least: service type, frequency of use, and usage experience, and the usage experience is used to characterize whether the used financial services meet the needs of target customers; a second drawing module, used to draw a journey map from a time perspective based on the financial product purchase information and the financial service usage information.

[0095] Optionally, the customer churn risk assessment based on the neural network model also includes: a second acquisition module, which is used to acquire historical user portraits, historical journey maps, and historical behavior graphs drawn based on historical behavior data within a historical time period to obtain an initial sample data set; a third acquisition module, which is used to configure a churn risk value and a recommendation probability value for each initial sample data in the initial sample data set to obtain a sample set; a first splitting module, which is used to split the sample set to obtain a training set and a test set; a first construction module, which is used to construct an initial model, wherein the initial model includes at least: a time series network layer, a graph neural network layer, a feature fusion layer, and a fully connected neural network; a first training module, which is used to iteratively train the initial model based on the training set to obtain an initial churn risk assessment model; a first testing module, which is used to test the initial churn risk assessment model based on the test set to obtain a test result, and when the test result indicates that the initial churn risk assessment model passes the test, a final churn risk assessment model is obtained.

[0096] It should be noted that the acquisition unit 41, extraction unit 42, drawing unit 43, and output unit 44 correspond to steps S201 to S204 in the first embodiment, and the examples and application scenarios implemented by the above units and corresponding steps are the same, but are not limited to the contents disclosed in the first embodiment. It should be noted that the above modules or units can be hardware components or software components stored in a memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n), and the above modules or units can also be part of the device and can be run in the computer terminal 10 provided in the first embodiment.

[0097] The present invention is described below in conjunction with another optional embodiment.

[0098] Embodiment 3

[0099] An embodiment of the present invention may also provide an electronic device, Figure 5is a hardware structure block diagram of an electronic device (or mobile device) for optionally executing a customer churn risk assessment method based on a neural network model according to an embodiment of the present invention, such as Figure 5 As shown, the electronic device may include: one or more ( Figure 5 (only one is shown) processor 502, memory 504, storage controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0100] Among them, the memory can be used to store software programs and modules, such as program instructions / modules corresponding to the methods and devices in the embodiments of the present application, and the processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0101] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain evaluation data of the target customer, wherein the evaluation data at least includes: user portrait, journey map, and behavior data set, and the journey map is used to describe the target customer's experience in the process of purchasing financial products and using financial services; extract features from the user portrait and journey map of the target customer to obtain a user portrait feature vector and a journey map feature vector; draw a behavior graph based on the behavior data set of the target customer; input the user portrait feature vector, the journey map feature vector, and the behavior graph into a churn risk assessment model, and output the churn risk value of the target customer, wherein the churn risk assessment model is a model pre-built based on a neural network model for assessing the churn risk of customers.

[0102] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: input the journey map feature vector into the time series network layer of the churn risk assessment model, and output the purchase behavior feature vector, wherein the time series network layer is used to analyze the journey map feature vector according to the time series; input the behavior graph into the graph neural network layer of the churn risk assessment model, and output the behavior preference feature vector, wherein the graph neural network layer is used to learn the embedded representation of each node in the behavior graph; input the user portrait feature vector, the purchase behavior feature vector and the behavior preference feature vector into the feature fusion layer of the churn risk assessment model to obtain a fused feature vector; input the fused feature vector into the fully connected neural network of the churn risk assessment model, and output the churn risk value of the target customer.

[0103] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: traverse the behavior data in the behavior data set, determine the execution objects of the behavior data and the jump relationships between the execution objects, wherein the execution objects at least include: browsing pages, clicking links; draw a behavior graph with the execution objects as nodes and the jump relationships between the execution objects as boundaries.

[0104] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: output the recommendation probability value of the target customer through the churn risk assessment model; when the recommendation probability value of the target customer is greater than a preset recommendation probability threshold, issue a recommendation task to the user terminal where the target customer is located, wherein the recommendation task is used to instruct the target customer to recommend a financial product.

[0105] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: compare the target customer's churn risk value with a preset churn risk threshold to obtain a comparison result; when the churn risk value is greater than the churn risk threshold, determine that the target customer has a churn risk and generate a retention strategy for the target customer.

[0106] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: collect the target customer's financial product purchase information within the target time period, wherein the financial product purchase information at least includes: product type, purchase amount, number of purchases, purchase experience information, and the purchase experience information is used to characterize whether the purchased financial product meets the needs of the target customer; collect the target customer's financial service usage information within the target time period, wherein the financial service usage information at least includes: service type, usage frequency, and usage experience, and the usage experience is used to characterize whether the used financial services meet the needs of the target customer; draw a journey map from a time perspective based on the financial product purchase information and the financial service usage information.

[0107] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain historical user portraits, historical journey maps, and historical behavior graphs drawn based on historical behavior data within a historical time period to obtain an initial sample data set; configure a churn risk value and a recommendation probability value for each initial sample data in the initial sample data set to obtain a sample set; split the sample set to obtain a training set and a test set; construct an initial model, wherein the initial model includes at least: a time series network layer, a graph neural network layer, a feature fusion layer, and a fully connected neural network; iteratively train the initial model based on the training set to obtain an initial churn risk assessment model; test the initial churn risk assessment model based on the test set to obtain a test result, and when the test result indicates that the initial churn risk assessment model passes the test, obtain a final churn risk assessment model.

[0108] By adopting the embodiment of the present invention, a technical solution for evaluating customer churn risk based on a neural network is provided. Based on multi-dimensional data such as user portraits, journey maps, and behavioral data as evaluation data, the churn risk value of the target customer is automatically output through a pre-built neural network model, so as to achieve the purpose of accurately quantifying the customer churn risk, and obtain the technical effect of evaluating the customer churn risk through multi-dimensional evaluation data and improving the accuracy of the evaluation results, thereby solving the technical problem in the related technology that the evaluation data for churn risk evaluation based on text mining or user portraits is relatively single, resulting in low accuracy of the evaluation results.

[0109] It can be understood by those skilled in the art that Figure 5 The structure shown is for illustration only, and the electronic device may also be a smart phone, a tablet computer, a PDA, a mobile Internet device (MID), a PAD or other terminal device. Figure 5 The structure of the electronic device is not limited. Figure 5 More or fewer components (such as network interfaces, display devices, etc.) shown in, or having Figure 5 Different configurations are shown.

[0110] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0111] The present invention is described below in conjunction with another optional embodiment.

[0112] Embodiment 4

[0113] The embodiment of the present invention further provides a computer-readable storage medium. Optionally, in the embodiment of the present invention, the computer-readable storage medium can be used to store the program code executed by the customer churn risk assessment method based on the neural network model provided in the first embodiment.

[0114] Optionally, in an embodiment of the present invention, the above-mentioned storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0115] An embodiment of the present invention also provides a computer program product, which, when executed on a data processing device, is suitable for executing the steps of a customer churn risk assessment method based on a neural network model: obtaining assessment data of target customers, wherein the assessment data at least includes: user portraits, journey maps, and a behavior data set, and the journey map is used to describe the experience of target customers in the process of purchasing financial products and using financial services; extracting features from the user portraits and journey maps of the target customers to obtain user portrait feature vectors and journey map feature vectors; drawing a behavior graph based on the behavior data set of the target customers; inputting the user portrait feature vectors, journey map feature vectors, and behavior graphs into a churn risk assessment model, and outputting the churn risk value of the target customer, wherein the churn risk assessment model is a model pre-constructed based on the neural network model for assessing the churn risk of customers.

[0116] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0117] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0118] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0119] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0120] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0121] If the integrated unit is implemented in the form of a software functional unit 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 invention, in essence, or the part that contributes to the prior art, or all or 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 a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0122] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A customer churn risk assessment method based on a neural network model, characterized in that: include: Obtaining evaluation data of target customers, wherein the evaluation data includes at least: user portraits, journey maps, and behavioral data sets, wherein the journey maps are used to describe the target customers' experience in purchasing financial products and using financial services; Performing feature extraction on the user portrait of the target customer and the journey map to obtain a user portrait feature vector and a journey map feature vector; Drawing a behavior graph based on the behavior data set of the target customer; The user portrait feature vector, the journey map feature vector and the behavior graph are input into a churn risk assessment model, and the churn risk value of the target customer is output, wherein the churn risk assessment model is a model pre-built based on a neural network model for assessing the churn risk of customers.

2. The method according to claim 1, characterized in that The step of inputting the user portrait feature vector, the journey map feature vector and the behavior graph into a churn risk assessment model and outputting the churn risk value of the target customer comprises: Inputting the journey map feature vector into the time series network layer of the churn risk assessment model, and outputting a purchase behavior feature vector, wherein the time series network layer is used to analyze the journey map feature vector according to a time series; Inputting the behavior graph into the graph neural network layer of the churn risk assessment model, and outputting a behavior preference feature vector, wherein the graph neural network layer is used to learn the embedded representation of each node in the behavior graph; Inputting the user portrait feature vector, the purchase behavior feature vector, and the behavior preference feature vector into the feature fusion layer of the churn risk assessment model to obtain a fused feature vector; The fused feature vector is input into the fully connected neural network of the churn risk assessment model, and the churn risk value of the target customer is output.

3. The method according to claim 1, characterized in that The step of drawing a behavior map based on the behavior data of the target customer includes: Traversing the behavior data in the behavior data set, determining the execution object of the behavior data and the jump relationship between the execution objects, wherein the execution object at least includes: browsing a page and clicking a link; A behavior graph is drawn with the execution objects as nodes and the jump relationships between the execution objects as boundaries.

4. The method according to claim 1, characterized in that: After inputting the user portrait feature vector, the journey map feature vector and the behavior graph into the churn risk assessment model, the method further includes: Outputting the recommendation probability value of the target customer through the churn risk assessment model; When the recommendation probability value of the target customer is greater than a preset recommendation probability threshold, a recommendation task is issued to the user terminal where the target customer is located, wherein the recommendation task is used to instruct the target customer to recommend a financial product.

5. The method according to claim 1, characterized in that After outputting the target customer's loss risk value, the method further includes: Comparing the target customer's churn risk value with a preset churn risk threshold to obtain a comparison result; When the churn risk value is greater than the churn risk threshold, it is determined that the target customer has a churn risk, and a retention strategy is generated for the target customer.

6. The method according to claim 1, characterized in that Before obtaining the target customer's evaluation data, it also includes: Collecting the target customer's financial product purchase information within the target time period, wherein the financial product purchase information at least includes: product type, purchase amount, purchase frequency, and purchase experience information, wherein the purchase experience information is used to indicate whether the purchased financial product meets the needs of the target customer; Collecting the target customer's financial service usage information within the target time period, wherein the financial service usage information includes at least: service type, usage frequency, and usage experience, wherein the usage experience is used to indicate whether the used financial services meet the needs of the target customer; The journey map is drawn from a time perspective based on the financial product purchase information and the financial service usage information.

7. The method according to claim 1, characterized in that The steps of constructing the churn risk assessment model include: Obtain historical user portraits and historical journey maps within the historical time period, as well as historical behavior graphs drawn based on historical behavior data, to obtain an initial sample data set; Assigning a loss risk value and a recommendation probability value to each piece of initial sample data in the initial sample data set to obtain a sample set; Splitting the sample set to obtain a training set and a test set; Constructing an initial model, wherein the initial model at least includes: a time series network layer, a graph neural network layer, a feature fusion layer, and a fully connected neural network; Iteratively training the initial model based on the training set to obtain an initial churn risk assessment model; The initial churn risk assessment model is tested based on the test set to obtain a test result, and when the test result indicates that the initial churn risk assessment model passes the test, the final churn risk assessment model is obtained.

8. A customer churn risk assessment device based on a neural network model, characterized in that: include: An acquisition unit, configured to acquire evaluation data of a target customer, wherein the evaluation data includes at least: a user portrait, a journey map, and a behavior data set, wherein the journey map is used to describe the experience of the target customer in the process of purchasing financial products and using financial services; An extraction unit, configured to extract features from the user portrait of the target customer and the journey map to obtain a user portrait feature vector and a journey map feature vector; A drawing unit, used for drawing a behavior graph based on the behavior data set of the target customer; An output unit is used to input the user portrait feature vector, the journey map feature vector and the behavior graph into a churn risk assessment model, and output the churn risk value of the target customer, wherein the churn risk assessment model is a model pre-built based on a neural network model for assessing the churn risk of customers.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the customer churn risk assessment method based on a neural network model as described in any one of claims 1 to 7.

10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the customer churn risk assessment method based on a neural network model as described in any one of claims 1 to 7.

Images