Customer maintenance method, device, processor and electronic device
By acquiring the characteristic parameters of financial institution customers and using neural network models for evaluation and classification, the problem of customer churn in financial institutions has been solved, enabling accurate customer classification and personalized services, and enhancing customer loyalty to financial institutions.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- INDUSTRIAL AND COMMERCIAL BANK OF CHINA
- Filing Date
- 2023-02-15
- Publication Date
- 2026-08-04
AI Technical Summary
Financial institutions' applications categorize and maintain customers solely based on business logic, leading to a high rate of customer churn.
By acquiring characteristic parameters such as product purchase frequency, total purchase amount, account balance, customer risk level, and product risk level from customer information, a trained neural network model is used to predict evaluation values. Based on the evaluation values, customers are classified and products and services related to the target customer group are pushed to them.
It enables precise customer segmentation and personalized services, enhances customer loyalty, helps financial institutions make differentiated decisions, and reduces customer churn.
Smart Images
Figure CN116226740B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence, and more specifically, to a customer maintenance method, apparatus, processor, and electronic device. Background Technology
[0002] In related technologies, financial institutions' applications, as a product backing with high credibility, are increasingly used. Financial institutions are also broadening their customer segmentation, but this segmentation merely selects customer groups for product recommendations, without assessing customers based on their financial product purchase behavior. Furthermore, all current customer segmentation within financial institutions relies on business logic to directly filter eligible individuals, resulting in inaccurate customer group classifications and hindering differentiated daily customer maintenance.
[0003] There is currently no effective solution to the problem that customer churn is easily caused by classifying and maintaining customers solely based on the internal business logic of financial institutions' applications in related technologies. Summary of the Invention
[0004] The main purpose of this application is to provide a customer maintenance method, apparatus, processor, and electronic device to solve the problem in related technologies where customer maintenance is based solely on the internal business logic of a financial institution's application, leading to easy customer churn.
[0005] To achieve the above objectives, according to one aspect of this application, a customer maintenance method is provided. The method includes: acquiring customer information of a target customer stored in a target application, wherein the customer information includes at least one of the following: product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle; preprocessing the customer information to obtain feature parameters, and inputting the feature parameters into a target neural network model to obtain a target evaluation value for the target customer, wherein the target neural network model is trained from multiple sets of training samples, each set of training samples including historical feature parameters and historical evaluation values of a customer; classifying the target customer based on the target evaluation value to obtain the target customer group to which the target customer belongs; pushing target products associated with the target customer group to the target customer, and providing the target customer with target services associated with the target customer group.
[0006] Optionally, when the customer information includes product purchase frequency, the preprocessing of the customer information to obtain feature parameters includes: obtaining the number of product purchases and the total number of cash outflows of the target customer within a preset time period; calculating a first ratio of the number of product purchases to the total number of cash outflows, and determining the first ratio as the feature parameter corresponding to the product purchase frequency.
[0007] Optionally, when customer information includes total product purchase amount, account balance, and customer risk level, the preprocessing of customer information to obtain feature parameters includes: obtaining the total product purchase amount of the target customer within a preset time period, and normalizing the total product purchase amount to obtain the feature parameters corresponding to the total product purchase amount; normalizing the account balance to obtain the feature parameters corresponding to the account balance; obtaining the assessment value of the customer risk level within a preset time period, and normalizing the assessment value of the customer risk level to obtain the feature parameters corresponding to the customer risk level.
[0008] Optionally, if the customer information includes product risk levels, the preprocessing of the customer information to obtain feature parameters includes: obtaining the product risk value of each product purchased by the target customer within a preset time period and the total number of products; calculating the sum of the product risk values of all products, and calculating a second ratio of the sum of product risk values to the total number of products; and determining the second ratio as the feature parameter corresponding to the product risk level.
[0009] Optionally, if the customer information includes product cycles, the preprocessing of the customer information to obtain feature parameters includes: obtaining the product cycle of each product purchased by the target customer within a preset time period; calculating the sum of the product cycles of all products, and calculating a third ratio of the sum of product cycles to the total number of products; and determining the third ratio as the feature parameter corresponding to the product cycle.
[0010] Optionally, the target neural network model is obtained by: determining the input layer, hidden layer, and output layer of the convolutional neural network model; determining the number of nodes in the input layer based on the number of feature parameters, and determining the number of nodes in the output layer based on the number of types of target evaluation values; calculating the number of nodes in the hidden layer based on the number of nodes in the input layer and the number of nodes in the output layer; determining the activation function of the hidden layer and the transfer function of the output layer; constructing an initial neural network model based on the input layer and the number of nodes in the input layer, the hidden layer and the number of nodes in the hidden layer, the output layer and the number of nodes in the output layer, the activation function, and the transfer function; and training the initial neural network model with multiple sets of training samples to obtain the target neural network model.
[0011] Optionally, classifying target customers based on target evaluation values to obtain the target customer groups to which the target customers belong includes: identifying multiple different types of customer groups and determining the evaluation value range associated with each customer group; determining the target evaluation value range to which the target evaluation value belongs, and identifying the customer groups associated with the target evaluation value range as the target customer groups.
[0012] To achieve the above objectives, according to another aspect of this application, a customer maintenance apparatus is provided. The apparatus includes: an acquisition unit for acquiring customer information of a target customer stored in a target application, wherein the customer information includes at least one of the following: product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle; a processing unit for preprocessing the customer information to obtain feature parameters, and inputting the feature parameters into a target neural network model to obtain a target evaluation value for the target customer, wherein the target neural network model is trained from multiple sets of training samples, each set of training samples including historical feature parameters and historical evaluation values of a customer; a classification unit for classifying the target customer based on the target evaluation value to obtain the target customer group to which the target customer belongs; and a push unit for pushing target products associated with the target customer group to the target customer and providing the target customer with target services associated with the target customer group.
[0013] This application employs the following steps: obtaining customer information of target customers stored in the target application, wherein the customer information includes at least one of the following: product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle; preprocessing the customer information to obtain feature parameters, and inputting the feature parameters into a target neural network model to obtain the target customer's target evaluation value, wherein the target neural network model is trained from multiple sets of training samples, each set of training samples including a customer's historical feature parameters and historical evaluation value; classifying the target customers based on the target evaluation value to obtain the target customer group to which the target customer belongs; pushing target products associated with the target customer group to the target customer, and providing the target customer with target services associated with the target customer group, thereby solving the problem in related technologies where customer classification and maintenance are based solely on the internal business logic of the financial institution's application, leading to easy customer churn. By acquiring customer information related to product purchases, inputting this information into a trained target neural network model, and obtaining target evaluation values, target customers are categorized into target customer groups based on these evaluation values. Then, corresponding maintenance services are provided to target customers based on these target customer groups. This achieves the effect of effectively analyzing customer preferences, measuring the relative importance of customers to financial institutions, helping financial institutions make differentiated decisions, enhancing customer loyalty, and accurately pushing and providing personalized services to target customers. Attached Figure Description
[0014] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:
[0015] Figure 1 This is a flowchart of a customer maintenance method provided according to an embodiment of this application;
[0016] Figure 2 This is a schematic diagram of the network structure of the target neural network model provided in the embodiments of this application;
[0017] Figure 3 This is a schematic diagram of a customer maintenance device provided according to an embodiment of this application;
[0018] Figure 4 This is a schematic diagram of an electronic device provided according to an embodiment of this application. Detailed Implementation
[0019] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.
[0020] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.
[0021] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0022] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties.
[0023] The present invention will now be described in conjunction with preferred implementation steps. Figure 1 This is a flowchart of a customer maintenance method provided according to an embodiment of this application, such as... Figure 1 As shown, the method includes the following steps:
[0024] Step S101: Obtain customer information of the target customer stored in the target application, wherein the customer information includes at least one of the following: product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle.
[0025] Specifically, the target application can be any mobile application developed by a financial institution to obtain customer information from the target application, such as product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle.
[0026] Step S102: Preprocess the customer information to obtain feature parameters, and input the feature parameters into the target neural network model to obtain the target evaluation value of the target customer. The target neural network model is trained by multiple sets of training samples, and each set of training samples includes the historical feature parameters and historical evaluation value of a customer.
[0027] Specifically, preprocessing refers to the preprocessing of the collected customer information, including data filtering and normalization. After preprocessing each type of customer information, the corresponding feature parameters can be obtained. The target neural network model can be a convolutional neural network model. Depending on the scenario requirements, one or more of the following information—product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle—can be selected as the feature parameters for the target neural network model as input values. By inputting the feature parameters into the target neural network model trained from multiple sets of training samples, the value assessment of the target customer corresponding to the customer information, i.e., the target assessment value, can be obtained.
[0028] Step S103: Classify target customers based on target evaluation values to obtain the target customer groups to which the target customers belong.
[0029] Specifically, based on the analysis of the product purchasing behavior of target customers, that is, the target customers' target evaluation values are classified into the corresponding evaluation value range, thereby classifying the target customers into the corresponding target customer groups.
[0030] Step S104: Push target products associated with the target customer group to the target customers, and provide target services associated with the target customer group to the target customers.
[0031] Specifically, customer groups selected based on customer information related to product purchase behavior take into account customer preferences, and different customer groups are associated with corresponding recommended products and routine maintenance services. For example, financial institutions may classify their customers into four groups: customers with no intention to purchase, potential customers, ordinary customers, and important customers. Customers with no intention to purchase may be recommended financial products with low investment and moderate returns, and provided with basic customer maintenance services. Potential customers may be recommended financial products with medium investment and higher returns, and provided with better customer maintenance services. Ordinary customers may be recommended financial products with medium investment and stable returns, and provided with frequent customer maintenance services. Important customers may be recommended financial products with higher investment and higher returns, and provided with dedicated maintenance services.
[0032] The customer maintenance method provided in this application embodiment obtains customer information of target customers stored in a target application. The customer information includes at least one of the following: product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle. The customer information is preprocessed to obtain feature parameters, which are then input into a target neural network model to obtain a target evaluation value for the target customer. The target neural network model is trained using multiple sets of training samples, each set including a customer's historical feature parameters and historical evaluation value. Based on the target evaluation value, the target customer is classified to obtain the target customer group to which the target customer belongs. Target products associated with the target customer group are pushed to the target customer, and target services associated with the target customer group are provided to the target customer. This method solves the problem in related technologies where customer classification and maintenance are based solely on the internal business logic of the financial institution's application, leading to customer churn. By acquiring customer information related to product purchases, inputting this information into a trained target neural network model, and obtaining target evaluation values, target customers are categorized into target customer groups based on these evaluation values. Then, corresponding maintenance services are provided to target customers based on these target customer groups. This achieves the effect of effectively analyzing customer preferences, measuring the relative importance of customers to financial institutions, helping financial institutions make differentiated decisions, enhancing customer loyalty, and accurately pushing and providing personalized services to target customers.
[0033] Optionally, in the customer maintenance method provided in this application embodiment, when the customer information includes the product purchase frequency, the preprocessing of the customer information to obtain the feature parameters includes: obtaining the number of product purchases and the total number of cash outflows of the target customer within a preset time period; calculating a first ratio of the number of product purchases to the total number of cash outflows, and determining the first ratio as the feature parameter corresponding to the product purchase frequency.
[0034] Specifically, the preset time period can be set to six months, the number of product purchases is the number of times the target customer purchases financial products on the financial institution's application, and the total number of transaction expenditures can be the number of times the target customer makes consumption on the financial institution's application within six months. By calculating the first ratio of the number of product purchases to the total number of transaction expenditures, the proportion of the target customer's product purchase expenditures in the total expenditures can be obtained, thereby enabling the analysis of the target customer's willingness to purchase various financial products. Therefore, the first ratio is used as the characteristic parameter corresponding to the product purchase frequency.
[0035] For example, if Customer A made 100 transactions on a financial institution's application within six months, 30 of which were for purchasing financial products, it means that the target customer is willing to purchase financial products accounting for 30% of their total spending. The higher the percentage, or the larger the first ratio, the higher the customer's willingness to purchase financial products, and the higher the customer's evaluation value.
[0036] Optionally, in the customer maintenance method provided in this application embodiment, when the customer information includes the total product purchase amount, account balance, and customer risk level, the preprocessing of the customer information to obtain feature parameters includes: obtaining the total product purchase amount of the target customer within a preset time period, and normalizing the total product purchase amount to obtain the feature parameters corresponding to the total product purchase amount; normalizing the account balance to obtain the feature parameters corresponding to the account balance; obtaining the assessment value of the customer risk level within the preset time period, and normalizing the assessment value of the customer risk level to obtain the feature parameters corresponding to the customer risk level.
[0037] Specifically, the preset time period can be set to six months, and the customer risk level can be the financial institution's assessment of the customer's risk-bearing capacity within six months. Normalization processing unifies the values corresponding to various customer information into a nearby range. For example, if the total product purchase amount is 150,000 yuan, the characteristic parameter corresponding to the total product purchase amount after normalization is 150,000 yuan. By normalizing customer information such as the total product purchase amount and account balance, which differ significantly from the product purchase frequency, corresponding characteristic parameters are obtained for comprehensive evaluation among multiple characteristic parameters.
[0038] Optionally, in the customer maintenance method provided in this application embodiment, when the customer information includes product risk levels, the preprocessing of the customer information to obtain feature parameters includes: obtaining the product risk value of each product purchased by the target customer within a preset time period and the total number of products; calculating the sum of the product risk values of all products, and calculating a second ratio of the sum of product risk values to the total number of products; and determining the second ratio as the feature parameter corresponding to the product risk level.
[0039] Specifically, the preset time period can be six months, and the product risk value can be the probability that a customer purchasing the financial product may lose money. The average risk value of the financial products purchased by the target customer within six months, i.e., the second ratio, is calculated as a characteristic parameter for evaluating the target customer. The higher the second ratio, the higher the target customer's willingness to purchase high-risk financial products, and thus the higher the evaluation value of the target customer.
[0040] Optionally, in the customer maintenance method provided in this application embodiment, when the customer information includes product cycles, the preprocessing of the customer information to obtain feature parameters includes: obtaining the product cycle of each product purchased by the target customer within a preset time period; calculating the sum of the product cycles of all products, and calculating a third ratio of the sum of product cycles to the total number of products; and determining the third ratio as the feature parameter corresponding to the product cycle.
[0041] Specifically, the preset time period can be six months. The product cycle refers to the investable period of a financial product, that is, the length of time a customer can hold the financial product after each purchase. For example, if the product cycle of product B is three months, the target customer can hold product B for three months and then need to repurchase product B. By calculating the average product cycle of all financial products purchased by the target customer within six months, also known as the third ratio, the third ratio is used as a characteristic parameter corresponding to the product cycle. The higher the third ratio, the higher the target customer's willingness to purchase financial products with longer product cycles, and thus the higher the target customer's evaluation value.
[0042] To obtain the target evaluation value of the target customer, a target neural network model needs to be trained based on training samples. Optionally, in the customer maintenance method provided in this application embodiment, the target neural network model is obtained in the following way: determining the input layer, hidden layer, and output layer of the convolutional neural network model; determining the number of nodes in the input layer based on the number of feature parameters, and determining the number of nodes in the output layer based on the number of types of target evaluation values; calculating the number of nodes in the hidden layer based on the number of nodes in the input layer and the number of nodes in the output layer; determining the activation function of the hidden layer and the transfer function of the output layer; constructing an initial neural network model based on the input layer and the number of nodes in the input layer, the hidden layer and the number of nodes in the hidden layer, the output layer and the number of nodes in the output layer, the activation function, and the transfer function; and training the initial neural network model with multiple sets of training samples to obtain the target neural network model.
[0043] Specifically, a convolutional neural network consists of three layers: an input layer, hidden layers, and an output layer. The input layer contains the input feature parameters, and the output layer contains the output parameters, i.e., the target evaluation value, which is output based on a transfer function. The hidden layer is the core layer of the convolutional neural network, handling a large amount of computation. The number of hidden layers and nodes has a decisive impact on the performance of the neural network. Considering overall performance, this application uses a neural network model containing one hidden layer. There is no fixed standard for the number of nodes in the hidden layer, but it does affect the learning effect. Too few nodes will not achieve the desired learning effect, while too many nodes exponentially increase network complexity and reduce learning efficiency. The number of nodes in the hidden layer can be calculated using any of the following formulas:
[0044] s = log₂n
[0045] s = 2n + 1
[0046]
[0047] Where s is the number of nodes in the hidden layer, n is the number of nodes in the input layer, i.e. the number of feature parameters, and m is the number of nodes in the output layer, i.e. the number of types of target evaluation values. Figure 2 This is a schematic diagram of the network structure of the target neural network model provided in the embodiments of this application, such as... Figure 2 As shown, if customer information includes product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle, then the number of nodes in the input layer is 6, and the number of nodes in the output layer is 1. Therefore, the number of nodes in the hidden layer can be selected between 2 and 13, with the specific value determined based on the training situation. The activation function of the hidden layer can be the hyperbolic tangent function, and the transfer function of the output layer can be a purelin-type linear function. After determining the convolutional neural network model, 80% of the collected customer information data is used as the training set, and the remaining 20% is used as the validation set. After training with the training samples, the mean squared error of the training results is set to be less than 0.01. When the final error is less than 0.01, training stops to generate the target neural network model. By training the target neural network model, the target evaluation value for the target customer is determined.
[0048] After determining the target evaluation value, the target customers are classified into target customer groups based on the target evaluation value. Optionally, in the customer maintenance method provided in this application embodiment, classifying the target customers based on the target evaluation value to obtain the target customer group to which the target customer belongs includes: determining multiple different types of customer groups and determining the evaluation value range associated with each customer group; determining the target evaluation value range to which the target evaluation value belongs, and determining the customer group associated with the target evaluation value range as the target customer group.
[0049] For example, four customer groups can be defined: customers with no purchase intention, potential customers, regular customers, and key customers. The total evaluation value can range from 10. The evaluation value range for customers with no purchase intention is 0 to 2.5, for potential customers it's 2.5 to 5, for regular customers it's 5 to 7.5, and for key customers it's 7.5 to 10. If the target evaluation value is 4, falling between 2.5 and 5, then the target evaluation value range is 2.5 to 5, and the target customer group is the potential customer group. By identifying the target customer group to which the target customer belongs, targeted customer maintenance services and recommendations of financial products that match the customer's preferences can be provided.
[0050] It should be noted that the steps shown in the flowchart in 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 may be executed in a different order than that shown here.
[0051] This application also provides a customer maintenance device. It should be noted that the customer maintenance device of this application can be used to execute the customer maintenance method provided in this application. The customer maintenance device provided in this application is described below.
[0052] Figure 3 This is a schematic diagram of a customer maintenance device provided according to an embodiment of this application. For example... Figure 3 As shown, the device includes:
[0053] The acquisition unit 10 is used to acquire customer information of target customers stored in the target application, wherein the customer information includes at least one of the following: product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle.
[0054] The processing unit 20 is used to preprocess customer information to obtain feature parameters, and input the feature parameters into the target neural network model to obtain the target evaluation value of the target customer. The target neural network model is trained by multiple sets of training samples, and each set of training samples includes the historical feature parameters and historical evaluation value of a customer.
[0055] Classification unit 30 is used to classify target customers based on target evaluation values to obtain the target customer group to which the target customer belongs;
[0056] Push unit 40 is used to push target products associated with the target customer group to the target customer and to provide target services associated with the target customer group to the target customer.
[0057] The customer maintenance device provided in this application embodiment acquires customer information of target customers stored in a target application through an acquisition unit 10. The customer information includes at least one of the following: product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle. A processing unit 20 preprocesses the customer information to obtain feature parameters and inputs these parameters into a target neural network model to obtain a target evaluation value for the target customer. The target neural network model is trained from multiple sets of training samples, each set including historical feature parameters and historical evaluation values for one customer. A classification unit 30 classifies the target customer based on the target evaluation value to obtain the target customer group to which the target customer belongs. A push unit 40... This technology pushes target products associated with target customer groups to target customers and provides them with target services associated with those groups. It addresses the problem of customer churn caused by relying solely on the internal business logic of financial institutions' applications for customer classification and maintenance. By acquiring customer information related to product purchases and inputting this information into a trained target neural network model, a target evaluation value is obtained. Based on this evaluation value, target customers are classified into target customer groups, and corresponding maintenance services are provided to these groups. This effectively analyzes customer preferences, measures the relative importance of customers to financial institutions, helps financial institutions make differentiated decisions, enhances customer loyalty, and accurately pushes and provides personalized services to target customers.
[0058] Optionally, in the customer maintenance device provided in this application embodiment, the processing unit 20 includes: a first acquisition module, used to acquire the number of product purchases and the total number of cash outflows of a target customer within a preset time period; and a first calculation module, used to calculate a first ratio of the number of product purchases to the total number of cash outflows, and to determine the first ratio as a characteristic parameter corresponding to the product purchase frequency.
[0059] Optionally, in the customer maintenance device provided in this application embodiment, the processing unit 20 includes: a second acquisition module, used to acquire the total product purchase amount of the target customer within a preset time period, and normalize the total product purchase amount to obtain the characteristic parameters corresponding to the total product purchase amount; a processing module, used to normalize the account balance to obtain the characteristic parameters corresponding to the account balance; and a third acquisition module, used to acquire the assessment value of the customer risk level within a preset time period, and normalize the assessment value of the customer risk level to obtain the characteristic parameters corresponding to the customer risk level.
[0060] Optionally, in the customer maintenance device provided in this application embodiment, the processing unit 20 includes: a fourth acquisition module, used to acquire the product risk value of each product purchased by the target customer within a preset time period and the total number of products; a second calculation module, used to calculate the sum of the product risk values of all products and calculate a second ratio of the sum of product risk values to the total number of products; and a first determination module, used to determine the second ratio as a characteristic parameter corresponding to the product risk level.
[0061] Optionally, in the customer maintenance device provided in this application embodiment, the processing unit 20 includes: a fifth acquisition module, used to acquire the product cycle of each product purchased by the target customer within a preset time period; a third calculation module, used to calculate the sum of the product cycles of all products, and calculate a third ratio of the sum of product cycles to the total number of products; and a second determination module, used to determine the third ratio as a characteristic parameter corresponding to the product cycle.
[0062] Optionally, in the customer maintenance device provided in this application embodiment, the target neural network model is obtained in the following manner: determining the input layer, hidden layer, and output layer of the convolutional neural network model; determining the number of nodes in the input layer based on the number of feature parameters, and determining the number of nodes in the output layer based on the number of types of target evaluation values; calculating the number of nodes in the hidden layer based on the number of nodes in the input layer and the number of nodes in the output layer; determining the activation function of the hidden layer and the transfer function of the output layer; constructing an initial neural network model based on the input layer and the number of nodes in the input layer, the hidden layer and the number of nodes in the hidden layer, the output layer and the number of nodes in the output layer, the activation function, and the transfer function; and training the initial neural network model with multiple sets of training samples to obtain the target neural network model.
[0063] Optionally, in the customer maintenance device provided in this application embodiment, the classification unit 30 includes: a third determining module, used to determine multiple different types of customer groups and determine the evaluation value range associated with each customer group; and a fourth determining module, used to determine the target evaluation value range to which the target evaluation value belongs and determine the customer group associated with the target evaluation value range as the target customer group.
[0064] The customer maintenance device includes a processor and a memory. The aforementioned acquisition unit 10, processing unit 20, classification unit 30, and push unit 40 are all stored in the memory as program units. The processor executes the aforementioned program units stored in the memory to realize the corresponding functions.
[0065] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and by adjusting kernel parameters, personalized services can be precisely targeted and delivered to specific clients.
[0066] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.
[0067] This invention provides a computer-readable storage medium having a program stored thereon that, when executed by a processor, implements a customer maintenance method.
[0068] This invention provides a processor for running a program, wherein the program executes a customer maintenance method during runtime.
[0069] Figure 4 This is a schematic diagram of an electronic device provided according to an embodiment of this application. For example... Figure 4 As shown, device 401 includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs the following steps: a customer maintenance method. The device in this document can be a server, PC, PAD, mobile phone, etc.
[0070] This application also provides a computer program product that, when executed on a data processing device, is suitable for executing a program with the following initialization method steps: a customer maintenance method.
[0071] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0072] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0073] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0074] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0075] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0076] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0077] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0078] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.
[0079] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0080] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.
Claims
1. A customer maintenance method, characterized in that, include: Obtain customer information of target customers stored in a target application, wherein the target application is an application developed by a financial institution and runs on a terminal, and the customer information includes at least: product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle; The customer information is preprocessed to obtain feature parameters, and the feature parameters are input into the target neural network model to obtain the target evaluation value of the target customer. The target neural network model is trained by multiple sets of training samples, and each set of training samples includes the historical feature parameters and historical evaluation value of a customer. The preprocessing of the customer information to obtain feature parameters includes: obtaining the product risk value and the total number of products purchased by the target customer within a preset time period; calculating the sum of the product risk values of all products, and calculating a second ratio of the sum of the product risk values to the total number of products; determining the second ratio as the feature parameter corresponding to the product risk level, wherein the product risk value is the probability that the customer who purchases the financial product will lose money; Based on the target evaluation value, the target customers are classified to obtain the target customer group to which the target customers belong; Push the target products associated with the target customer group to the target customers, and provide the target services associated with the target customer group to the target customers; The target neural network model is obtained in the following way: Determine the input layer, hidden layer, and output layer of the convolutional neural network model; The number of nodes in the input layer is determined based on the number of feature parameters, and the number of nodes in the output layer is determined based on the number of types of the target evaluation values. The number of nodes in the hidden layer is calculated based on the number of nodes in the input layer and the number of nodes in the output layer; Determine the activation function of the hidden layer and the transfer function of the output layer, wherein the activation function is a hyperbolic tangent function and the transfer function is a purelin-type linear function; An initial neural network model is constructed based on the input layer and the number of nodes in the input layer, the hidden layer and the number of nodes in the hidden layer, the output layer and the number of nodes in the output layer, the activation function, and the transfer function. The initial neural network model is trained using the multiple sets of training samples to obtain the target neural network model, wherein the mean square error of the training result is less than 0.
01.
2. The method according to claim 1, characterized in that, When the customer information includes product purchase frequency, the preprocessing of the customer information to obtain feature parameters includes: Obtain the number of product purchases and the total number of cash outflows for the target customer within a preset time period; Calculate a first ratio of the number of times the product is purchased to the total number of times the cash flow is spent, and determine the first ratio as a characteristic parameter corresponding to the product purchase frequency.
3. The method according to claim 1, characterized in that, When the customer information includes total product purchase amount, account balance, and customer risk level, the preprocessing of the customer information to obtain feature parameters includes: Obtain the total product purchase amount of the target customer within a preset time period, and normalize the total product purchase amount to obtain the characteristic parameters corresponding to the total product purchase amount. The account balance is normalized to obtain the characteristic parameters corresponding to the account balance; The assessment value of the customer risk level within the preset time period is obtained, and the assessment value of the customer risk level is normalized to obtain the feature parameters corresponding to the customer risk level.
4. The method according to claim 1, characterized in that, When the customer information includes product cycles, the preprocessing of the customer information to obtain feature parameters includes: Obtain the product cycle of each product purchased by the target customer within the preset time period; Calculate the sum of the product cycles of all products, and calculate the third ratio of the sum of the product cycles to the total number of products; The third ratio is determined as the characteristic parameter corresponding to the product cycle.
5. The method according to claim 1, characterized in that, Based on the target evaluation value, the target customers are classified to obtain the target customer groups to which the target customers belong, including: Identify multiple different customer groups and determine the range of evaluation values associated with each customer group; Determine the target evaluation value range to which the target evaluation value belongs, and identify the customer group associated with the target evaluation value range as the target customer group.
6. A customer maintenance device, characterized in that, include: The acquisition unit is used to acquire customer information of target customers stored in a target application, wherein the target application is an application developed by a financial institution and running on a terminal, and the customer information includes at least: product purchase frequency, total product purchase amount, account balance, customer risk level, product risk level, and product cycle. The processing unit is used to preprocess the customer information to obtain feature parameters, and input the feature parameters into the target neural network model to obtain the target evaluation value of the target customer. The target neural network model is trained by multiple sets of training samples, and each set of training samples includes the historical feature parameters and historical evaluation value of a customer. The processing unit includes: a fourth acquisition module, used to acquire the product risk value of each product purchased by the target customer within a preset time period and the total number of products; a second calculation module, used to calculate the sum of the product risk values of all products and calculate a second ratio of the sum of the product risk values to the total number of products; and a first determination module, used to determine the second ratio as a characteristic parameter corresponding to the product risk level, wherein the product risk value is the probability that the customer who purchases the financial product will lose money. A classification unit is used to classify the target customers based on the target evaluation value to obtain the target customer group to which the target customers belong; The push unit is used to push target products associated with the target customer group to the target customer and to provide target services associated with the target customer group to the target customer. The target neural network model is obtained through the following steps: determining the input layer, hidden layers, and output layer of the convolutional neural network model; determining the number of nodes in the input layer based on the number of feature parameters, and determining the number of nodes in the output layer based on the number of types of target evaluation values; calculating the number of nodes in the hidden layer based on the number of nodes in the input layer and the number of nodes in the output layer; determining the activation function of the hidden layer and the transfer function of the output layer, wherein the activation function is a hyperbolic tangent function and the transfer function is a purelin-type linear function; constructing an initial neural network model based on the input layer and the number of nodes in the input layer, the hidden layer and the number of nodes in the hidden layer, the output layer and the number of nodes in the output layer, the activation function, and the transfer function; and training the initial neural network model using multiple sets of training samples to obtain the target neural network model, wherein the mean square error of the training result is less than 0.
01.
7. A non-volatile storage medium, characterized in that, The non-volatile storage medium includes a stored program, wherein the program, when running, controls the device where the non-volatile storage medium is located to execute the customer maintenance method according to any one of claims 1 to 5.
8. An electronic device, characterized in that, It includes one or more processors and a memory, the memory being 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 cause the one or more processors to implement the customer maintenance method according to any one of claims 1 to 5.