Customer state determination method, device and computer program product

Abstract

The invention discloses a client state determination method and device and a computer program product. Relates to the field of financial science and technology, and the method comprises the steps: obtaining N static feature data and K dynamic feature data of a to-be-evaluated customer, each static feature data being data representing customer information of the to-be-evaluated customer, and each dynamic feature data being financial feature data of the to-be-evaluated customer changing with time; inputting the N static feature data into a first evaluation model, processing to obtain a first index, inputting the K dynamic feature data into a second evaluation model, and outputting a second index; and obtaining a preset weight, performing weighted calculation on the first index and the second index based on the preset weight to obtain probability data of the to-be-evaluated customer, and determining the state of the to-be-evaluated customer based on the probability data. Through the method and the device, the technical problem of low evaluation precision during evaluation of the state of the customer in related technologies is solved.

Description

Technical Field

[0001] This application relates to the field of financial technology, and more specifically, to a method, apparatus, and computer program product for determining customer status. Background Technology

[0002] As financial institutions deepen their digital transformation, they and their branches are increasingly focusing on identifying and serving long-tail customers, striving to enhance their value and contribution through online operational strategies. Long-tail customers typically refer to those with lower average daily assets. While individual customers may contribute less, their sheer number makes their potential value undeniable. Currently, financial institutions' operational strategies for long-tail customers are often rather crude, particularly in the promotion of asset retention programs, which generally employ a uniform end-of-month approach. This one-size-fits-all strategy ignores the differences in asset retention duration among different customer groups and fails to effectively identify potential trends in customer assets. Consequently, the timing of these promotions is often imprecise, resulting in many customers having already experienced asset downgrades before the promotion begins, making it exceptionally difficult to win back downgraded customers through promotions.

[0003] Specifically, when financial institutions promote their activities only at the end of the month, customers who have already downgraded at the beginning or middle of the month will not benefit from these activities, thus reducing the efficiency and effectiveness of the activities. Furthermore, because the timing of downgrades by long-tail customers cannot be predicted in advance, financial institutions struggle to intervene at the optimal time, leading to an increase in the number of customers downgrading their assets and consequently impacting the effectiveness of customer asset management. The timing of uniform promotion lacks flexibility: all customers receive information about the retention program at the same time, ignoring the differences in asset fluctuations among individual customers; promotional information is often pushed out after customers have already downgraded, resulting in wasted operational resources and reduced operational effectiveness; inaccurate timing of outreach may make customers feel disturbed rather than served, affecting customer satisfaction and loyalty; financial institutions fail to fully utilize historical data and customer characteristics to predict asset downgrade trends, lacking scientific data support for decision-making.

[0004] There is currently no effective solution to the technical problem of low accuracy in assessing customer status in related technologies. Summary of the Invention

[0005] The main objective of this application is to provide a method, apparatus, and computer program product for determining customer status, in order to solve the technical problem of low assessment accuracy in the related art when evaluating customer status.

[0006] To achieve the above objectives, according to one aspect of this application, a method for determining customer status is provided. The method includes: acquiring N static feature data and K dynamic feature data of a customer to be evaluated, wherein each static feature data refers to data characterizing customer information of the customer to be evaluated, and each dynamic feature data refers to financial feature data of the customer to be evaluated that changes over time, and N and K are positive integers; inputting the N static feature data into a first evaluation model to obtain a first indicator, inputting the K dynamic feature data into a second evaluation model to output a second indicator; obtaining preset weights, performing a weighted calculation on the first and second indicators based on the preset weights to obtain probability data of the customer to be evaluated, and determining the status of the customer to be evaluated based on the probability data.

[0007] Furthermore, obtaining N static feature data and K dynamic feature data of the customer to be evaluated includes: obtaining customer information of the customer to be evaluated, wherein the customer information is used to indicate the customer identity and account information of the customer to be evaluated; extracting M customer feature data of the customer to be evaluated within a preset time period from the database based on the customer information, wherein M is a positive integer; dividing the M customer feature data into data types to obtain N static feature data and K dynamic feature data, wherein the data types include at least dynamic type and static type, N+K=M.

[0008] Furthermore, the first and second evaluation models are trained as follows: Customer characteristic data of Y customers within a historical time period are obtained, resulting in Y sets of historical customer characteristic data, where Y is a positive integer; Amount storage data for each customer is extracted from the Y sets of historical customer characteristic data, resulting in Y amounts of stored data; Indicators for each customer are determined based on each amount of stored data, where each indicator type includes: a first type, a second type, and a third type; A model training set is constructed from the Y sets of historical customer characteristic data and Y labels; The first preset evaluation model is trained based on the model training set to obtain the first evaluation model; The second preset evaluation model is trained based on the model training set to obtain the second evaluation model.

[0009] Furthermore, the model training set, consisting of Y sets of historical customer feature data and Y labels, includes: dividing each set of historical customer feature data into data types to obtain set A of historical static feature data and set B of historical dynamic feature data, where A + B = Y, and A and B are positive integers; constructing a first model training set using set A of historical static feature data and Y labels, and constructing a second model training set using set B of historical dynamic feature data and Y labels; and combining the first model training set and the second model training set to obtain the final model training set.

[0010] Furthermore, before inputting N static feature data into the first evaluation model and processing them to obtain the first index, the method further includes: preprocessing the N static feature data to obtain N processed static feature data, wherein the preprocessing method includes at least one of the following: missing value imputation, outlier handling, and categorical feature encoding; inputting the N static feature data into the first evaluation model and processing them to obtain the first index includes: extracting features from the N processed static feature data by the feature extraction module in the first evaluation model to obtain T static features, and determining the contribution score of the static features by a first preset function in the first evaluation model, wherein T is a positive integer; extracting P static features from the T static features based on the T contribution scores by the feature extraction module, evaluating the P static features to obtain the first index, wherein P is a positive integer.

[0011] Further, inputting K dynamic feature data into the second evaluation model and outputting the second index includes: preprocessing the K dynamic feature data to obtain K processed dynamic feature data, and sorting each processed dynamic feature data according to the timestamp of each processed dynamic feature data to obtain a dynamic feature sequence; extracting features from the dynamic feature sequence by the second evaluation model to obtain R dynamic features, and converting the format of the R dynamic features to obtain R converted dynamic features, where R is a positive integer; predicting the R converted dynamic features by the second preset function in the second evaluation model to obtain feature prediction results, and converting the feature prediction results numerically to obtain the second index.

[0012] Furthermore, after determining the status of the customer to be evaluated based on probability data, the method also includes: obtaining a list of operational strategies, wherein the list of operational strategies includes multiple probability data and the operational strategy corresponding to each probability data; determining a target operational strategy based on the probability data; and adjusting the strategy for the customer to be evaluated based on the target operational strategy.

[0013] To achieve the above objectives, according to another aspect of this application, a device for determining customer status is provided. The device includes: a first acquisition unit, configured to acquire N static feature data and K dynamic feature data of a customer to be evaluated, wherein each static feature data refers to data characterizing customer information of the customer to be evaluated, and each dynamic feature data refers to financial feature data of the customer to be evaluated that changes over time, and N and K are positive integers; an input unit, configured to input the N static feature data into a first evaluation model, process it to obtain a first indicator, input the K dynamic feature data into a second evaluation model, and output a second indicator; and a second acquisition unit, configured to acquire preset weights, perform weighted calculations on the first and second indicators based on the preset weights to obtain probability data of the customer to be evaluated, and determine the status of the customer to be evaluated based on the probability data.

[0014] According to another aspect of the present invention, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to execute any of the above-described methods for determining the client state.

[0015] According to another aspect of the present invention, an electronic device is also provided, including one or more processors and a memory, the memory storing an executable program, and the processor for running the program, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any of the above-described methods for determining the client state.

[0016] According to another aspect of the present invention, a computer program product is also provided, the computer program product including a computer program, wherein when the computer program is executed by a processor, it implements any of the above-described methods for determining the customer state.

[0017] In this embodiment, a method for determining customer status is adopted. This involves acquiring N static feature data and K dynamic feature data of the customer to be evaluated. Each static feature data represents customer information, and each dynamic feature data represents financial characteristic data of the customer changing over time. N and K are positive integers. The N static feature data are input into a first evaluation model to obtain a first indicator. The K dynamic feature data are input into a second evaluation model to output a second indicator. Preset weights are obtained, and the first and second indicators are weighted and calculated to obtain probability data of the customer to be evaluated. The customer status is determined based on the probability data. This solves the technical problem of low evaluation accuracy in related technologies when evaluating customer status. By acquiring the static and dynamic feature data of the customer to be evaluated, inputting them into the first and second evaluation models respectively, and obtaining the first and second indicators, the probability data is determined based on the first and second indicators, thereby determining the customer status. This improves the evaluation accuracy of customer status and enables the sending of operational strategies to customers based on their status. Attached Figure Description

[0018] 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:

[0019] Figure 1 This is a hardware structure block diagram of a computer terminal (or mobile device) used to implement a method for determining client status;

[0020] Figure 2This is a flowchart of a method for determining customer status according to an embodiment of this application;

[0021] Figure 3 This is a schematic diagram of the training methods for the first evaluation model and the second evaluation model provided according to the embodiments of this application;

[0022] Figure 4 This is a schematic diagram of the structure of the second evaluation model provided according to the embodiments of this application;

[0023] Figure 5 This is a schematic diagram of a customer status determination device according to an embodiment of this application;

[0024] Figure 6 This is a structural block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0025] 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.

[0026] 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 so that the embodiments of this application described herein can be implemented in orders other than those illustrated or 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.

[0027] It should be noted that all 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 application are information and data authorized by the user or fully authorized by all parties. For example, this system has interfaces with relevant users or organizations to provide users with corresponding operation data for them to choose to agree to or refuse automated decision results. Before obtaining relevant information, a request for obtaining the information needs to be sent to the aforementioned user or organization through the interface, and the relevant information is obtained after receiving consent from the aforementioned user or organization; if the user chooses to refuse, the expert decision-making process is initiated. Users can view the purpose of data use in real time through authorization decoding and have the right to withdraw authorization or delete data at any time. After the authorization is withdrawn, the system will terminate the relevant data processing within 24 hours.

[0028] It should be noted that the information collected in this application is 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 the relevant data all comply with the relevant laws, regulations and standards of the relevant regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation access points for users to choose to authorize use or refuse use.

[0029] Example 1

[0030] According to an embodiment of this application, a method embodiment for determining a customer's status is also provided. 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. Furthermore, 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.

[0031] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 This is a hardware structure block diagram of a computer terminal (or mobile device) used to implement a method for determining client status, such as... Figure 1 As shown, computer terminal 10 (or mobile device) may include one or more ( Figure 1(Illustrated as 102a, 102b, ..., 102n) Processor 102 (processor 102 may include, but is not limited to, a microprocessor MCU (Microcontroller Unit) or a programmable gate array (FPGA), etc.), memory 104 for storing data, and transmission device 106 for communication functions. In addition, it may include: a display, input / output interface (I / O interface), universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), network interface, keyboard, cursor control device, power supply and / or camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0032] It should be noted that the aforementioned one or more processors 102 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10 (or mobile device). As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0033] 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 status determination method in this embodiment. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the aforementioned customer status determination method. The memory 104 may include high-speed random access memory, and may also include 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 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0034] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network interface controller (NIC) and a network interface, which can be connected to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a radio frequency (RF) module, used for wireless communication with the Internet.

[0035] The display can be, for example, a touchscreen liquid crystal display (LCD), which allows the user to interact with the user interface of the computer terminal 10 (or mobile device).

[0036] Under the aforementioned operating environment, this application provides the following: Figure 2 The method for determining customer status is shown. Figure 2 This is a flowchart of a method for determining customer status according to an embodiment of this application, such as... Figure 2 As shown, the method includes the following steps:

[0037] Step S201: Obtain N static feature data and K dynamic feature data of the customer to be evaluated. Each static feature data refers to customer information representing the customer to be evaluated, and each dynamic feature data refers to financial feature data of the customer to be evaluated that changes over time. N and K are positive integers.

[0038] It should be noted that the customer to be evaluated refers to a customer whose assets are within a certain range (i.e., who has completed asset upgrade). Static characteristic data refers to the customer's basic information and long-term stable financial behavior characteristics, which do not change frequently over time. It provides a basic overview of the customer and long-term behavioral patterns, helping to understand the customer's background information and stable financial preferences. For example, static characteristic data can include the customer's basic information (such as age, gender, etc.), product holding status (such as whether an account is open, etc.), consumption preferences, and financial preferences (such as investment preferences, etc.). Dynamic characteristic data refers to customer financial behavior data that changes over time, revealing the changing trends of customer assets and consumption behavior over time. For example, dynamic characteristic data can include asset-related characteristics (such as total asset amount, etc.), consumption-related characteristics (such as total consumption amount, etc.), and expenditure scenario-related characteristics (the situation in various consumption scenarios, such as daily life, payment, etc.).

[0039] Specifically, because existing financial institutions allocate too uniformly the time for asset preservation activities for long-tail customers, failing to differentiate the time periods for different customers' specific preservation trends, customer churn occurs, thus affecting the preservation effect. Therefore, in order to reduce customer churn, effective intervention measures need to be taken before the critical moment of customer downgrade, that is, to implement different operational strategies for different customers, thereby improving the success rate of preservation and reducing the rate of customer asset downgrade. At this time, it can be determined whether a customer is in a downgrade period by assessing the customer's churn status.

[0040] Therefore, in order to determine the aforementioned customer status, it is first necessary to collect comprehensive feature data for each customer to be evaluated, that is, to obtain multiple static feature data and dynamic feature data. Static feature data provides the customer's static attributes and long-term preferences, while dynamic feature data captures the customer's immediate behavior and asset change trends. By combining the two types of feature data, the model can understand the customer from multiple dimensions, and then predict the asset retention period for different customer groups, that is, predict the customer status. This allows for the implementation of personalized outreach and operational strategies at the appropriate time before the customer is downgraded, effectively improving the conversion rate of operational activities and reducing the asset downgrade rate.

[0041] Step S202: Input N static feature data into the first evaluation model, process to obtain the first index, and input K dynamic feature data into the second evaluation model to output the second index.

[0042] Specifically, after obtaining the aforementioned dynamic and static characteristic data, different assessment models can be used to process different types of customer characteristic data, ultimately generating two indicators: the first indicator and the second indicator. These two indicators then work together in the subsequent model fusion stage to comprehensively predict the customer's status. The first indicator refers to the quantitative result obtained by the first assessment model after analyzing static characteristics, reflecting the likelihood of a customer's assets being downgraded under certain conditions (such as based solely on static characteristics), i.e., characterizing the degree to which the customer's basic attributes affect their ability to maintain asset status. The second indicator, output by the second assessment model, represents the risk level of customer asset downgrade assessed based on dynamic characteristics (the customer's short-term behavior and asset trends), characterizing the impact of recent dynamic factors on asset stability.

[0043] It should be noted that the first evaluation model can be constructed using an eXtremeGradient Boosting (XGBoost) model, which can handle structured static feature data, capture the complex interactions between static features, and continuously optimize the decision tree structure through multiple iterations to find the optimal feature split point, thereby assessing the future churn trend of customers. The second evaluation model can be constructed using a Long Short-Term Memory (LSTM) network model, which can capture dynamic patterns in time-series data, thereby understanding and inferring the impact of time-series-based dynamic features on customer asset retention.

[0044] Inputting static characteristic data into the first assessment model allows us to understand how customer attributes affect the stability of their assets, thus providing a baseline for subsequent dynamic analysis. This first indicator not only reveals the static aspect of a customer's ability to maintain their credit rating but also indirectly provides an important reference point for model fusion. Inputting dynamic characteristic data into the second assessment model captures real-time changes in a customer's financial status, revealing recent behavioral patterns and providing real-time data support for subsequent operational strategy design. This enables financial institutions to respond to immediate changes in a customer's asset status.

[0045] Step S203: Obtain preset weights, perform weighted calculations on the first and second indicators based on the preset weights to obtain probability data of the customer to be evaluated, and determine the status of the customer to be evaluated based on the probability data.

[0046] Specifically, after obtaining the first and second indicators, the first and second indicators can be weighted using preset weights to obtain probability data representing the possibility of asset downgrading for the customer to be evaluated. The preset weights are coefficients used to balance the influence of the prediction results of the first and second evaluation models. The probability data can be values ​​such as 0, 1, and 2. The larger the value, the higher the possibility of asset downgrading for the customer.

[0047] Then, based on probability data, the status of the customers to be evaluated is determined, that is, the customer classification based on probability data can be divided into three levels: "high risk," "medium risk," and "low risk." For example, if the probability of a customer to be evaluated is higher than a certain high-risk threshold, then the customer is classified as "high-risk," indicating that the customer has a higher risk of asset downgrade during the future maintenance period. Conversely, if the probability is lower than a certain low-risk threshold, it is considered "low-risk." Finally, different operational strategies are formulated based on the status to improve the effectiveness of activities and the accuracy of resource allocation, ultimately improving the effectiveness of long-tail customer management.

[0048] The customer status determination method provided in this application embodiment acquires N static feature data and K dynamic feature data of the customer to be evaluated. Each static feature data refers to customer information representing the customer to be evaluated, and each dynamic feature data refers to financial feature data of the customer to be evaluated that changes over time. N and K are positive integers. The N static feature data are input into a first evaluation model to obtain a first indicator. The K dynamic feature data are input into a second evaluation model to output a second indicator. Preset weights are obtained, and the first and second indicators are weighted and calculated to obtain probability data of the customer to be evaluated. The customer status is determined based on the probability data. This method solves the technical problem of low evaluation accuracy in related technologies when evaluating customer status. By acquiring the static and dynamic feature data of the customer to be evaluated, inputting the static and dynamic feature data into the first and second evaluation models respectively, processing to obtain the first and second indicators, and determining the probability data based on the first and second indicators, the customer status is determined, thereby improving the evaluation accuracy of customer status and enabling the sending of operational strategies to customers based on their status.

[0049] Optionally, in the method for determining customer status provided in this application embodiment, obtaining N static feature data and K dynamic feature data of the customer to be evaluated includes: obtaining customer information of the customer to be evaluated, wherein the customer information is used to indicate the customer identity and account information of the customer to be evaluated; extracting M customer feature data of the customer to be evaluated within a preset time period from the database based on the customer information, wherein M is a positive integer; dividing the M customer feature data into data types to obtain N static feature data and K dynamic feature data, wherein the data types include at least: dynamic type and static type, N+K=M.

[0050] Specifically, when acquiring static and dynamic feature data, the first step is to obtain complete information about the customer to be evaluated from the relevant databases of financial institutions, including but not limited to the customer's identity information (such as name and ID number) and account information (such as account type), to ensure that each customer can be accurately located, thereby extracting all historical data related to that customer.

[0051] Furthermore, after obtaining customer information, customer characteristic data of the customer to be evaluated within a preset time period can be extracted from the database based on the customer information. For example, customer characteristic data of the customer within 60 days can be obtained. It should be noted that the customer characteristic data at this time includes static characteristics and dynamic characteristics, such as static attributes such as age and gender, and dynamic characteristics such as recent changes in total assets.

[0052] The extracted customer feature data is then categorized into data types: static feature data and dynamic feature data. This categorization can be based on the nature of the features. Static feature data refers to customer attributes that do not change or change slowly over time, such as age and gender. Dynamic feature data, on the other hand, focuses on features that change rapidly over time, such as asset balance and consumption records.

[0053] This embodiment divides customer feature data to obtain static and dynamic feature data, providing high-quality input data for the evaluation model. This enables the fusion model to understand customer behavior from multiple perspectives, guide the optimization of operational strategies, and enhance the customer experience.

[0054] Optionally, in the customer status determination method provided in this application embodiment, the first evaluation model and the second evaluation model are trained in the following manner: acquiring customer feature data of Y customers within a historical time period to obtain Y sets of historical customer feature data, where Y is a positive integer; extracting the amount storage data of each customer from the Y sets of historical customer feature data to obtain Y amount storage data; determining the indicators of each customer based on each amount storage data, wherein the indicator type of each indicator includes: a first type, a second type, and a third type; constructing a model training set from the Y sets of historical customer feature data and Y labels; training the first preset evaluation model based on the model training set to obtain the first evaluation model; and training the second preset evaluation model based on the model training set to obtain the second evaluation model.

[0055] To improve the accuracy of both the first and second assessment models, and thus implement precise operational strategies before customer downgrades, reducing the asset downgrade rate, the following approach is taken: First, customer characteristic data from multiple customers over a historical period can be obtained, resulting in multiple sets of historical customer characteristic data. This data includes static characteristics (such as age) and dynamic characteristics (such as recent asset changes and consumption behavior). Then, the amount stored for each customer is extracted from each set of historical customer characteristic data. For example, financial storage data could indicate the customer's asset storage status, such as an average monthly asset value greater than or equal to 10,000. Based on this amount stored data, indicators are determined for each customer. For instance, indicators can categorize the duration of asset retention into three levels. For example, if a customer's average monthly asset value remains above 10,000 for no more than 7 days within a month, the customer is labeled as Type 1, representing short-term retention; if the retention period is between 8 and 15 days, it is labeled as Type 2, corresponding to medium-term retention; and if the customer's average monthly asset value remains above 10,000 throughout the entire observation period (e.g., one month), it is labeled as Type 3, representing long-term retention.

[0056] Furthermore, historical customer characteristic data and indicators are integrated to form a model training set for training the evaluation model. Finally, this model training set is used to train the first and second preset evaluation models respectively. That is, an algorithm suitable for processing static features (such as XGBoost) is selected, and historical static feature data and labels are used as input for training. The algorithm parameters are continuously optimized to improve the accuracy of the prediction indicators, resulting in the first evaluation model. A neural network suitable for processing time-series data (such as LSTM) is selected and trained using historical dynamic feature data and labels. Through deep learning of sequence data, the second evaluation model can capture the trend and pattern of asset changes and predict the likelihood of the customer maintaining their rating in the future.

[0057] Figure 3 This is a schematic diagram of the training methods for the first and second evaluation models provided in the embodiments of this application, as shown below. Figure 3 As shown, historical dynamic feature data is input into a first preset evaluation model constructed by an LSTM model, which outputs lstm_output. Historical static feature data is input into a second preset evaluation model constructed by an Xgboost model, which outputs Xgboost_output. Then, a preset weight is obtained through a model fusion layer. Using this weight, the following calculation is performed: Weight*Lstm_output + Weight*Xgboost_output, thus obtaining the model output, which is the initial label. Finally, the initial label and the label are compared, and the algorithm parameters are optimized based on the comparison results to improve the accuracy of the prediction index. Finally, the corresponding first evaluation model and second evaluation model are obtained.

[0058] This embodiment utilizes static feature data to train a model that can identify the impact of customer basic information and product holding status on asset retention duration. It also utilizes dynamic feature data to train a model that can capture and analyze the dynamic impact of asset changes and transaction behavior on retention duration. This effectively establishes a dual-assessment model system targeting both static and dynamic features, making the prediction of customer asset retention duration more comprehensive and accurate. This not only improves the management efficiency of financial institutions for long-tail customer groups but also enhances customer satisfaction, reduces operating costs, and provides strong support for the digital transformation of financial institutions.

[0059] Optionally, in the customer status determination method provided in this application embodiment, the model training set composed of Y groups of historical customer feature data and Y labels includes: dividing each group of historical customer feature data into data types to obtain group A of historical static feature data and group B of historical dynamic feature data, where A+B=Y, and A and B are positive integers; constructing a first model training set using group A of historical static feature data and Y labels, and constructing a second model training set using group B of historical dynamic feature data and Y labels; and combining the first model training set and the second model training set to obtain the model training set.

[0060] Specifically, when constructing the model training set, the collected historical customer characteristic data can first be meticulously categorized. Specifically, each set of historical customer characteristic data can be divided into two parts: historical static characteristic data and historical dynamic characteristic data. This categorization is based on whether customer characteristics change significantly over time. Static characteristic data includes attributes such as age and occupation that remain unchanged or change slowly over a long period, while dynamic characteristic data covers financial behavior characteristics that fluctuate significantly over time, such as total assets and consumption amounts.

[0061] Furthermore, to train a first assessment model capable of evaluating the impact of static characteristics on customer asset retention, a training set for the first model can be constructed using historical static characteristic data and labels. This associates each customer's static characteristics at a specific point in time with the labels of the representation results, forming feature-label pairs. This ensures that the first assessment model can learn during training to identify which combinations of static characteristics might lead to future asset instability, thereby enabling risk assessment for new customers. Simultaneously, a second model training set can be constructed using historical dynamic characteristic data and labels. This associates each customer's dynamic characteristics at a specific point in time with the labels of the representation results, forming feature-label pairs. This ensures that the second assessment model, by learning the correlation between these dynamic data and results, can capture behavioral patterns that might lead to asset downgrades in the short term, thereby predicting the immediate state of customer assets. Finally, the first and second model training sets together constitute the complete model training set.

[0062] This embodiment constructs different model training sets by dividing the data, avoiding the prediction bias caused by the mixed data types of traditional single models, and can effectively improve the prediction accuracy of customer asset preservation trends.

[0063] Optionally, in the customer status determination method provided in this application embodiment, before inputting N static feature data into the first evaluation model and processing to obtain the first indicator, the method further includes: preprocessing the N static feature data to obtain N processed static feature data, wherein the preprocessing method includes at least one of the following: missing value imputation, outlier handling, and categorical feature encoding; inputting the N static feature data into the first evaluation model and processing to obtain the first indicator includes: extracting features from the N processed static feature data by the feature extraction module in the first evaluation model to obtain T static features, and determining the contribution score of the static features by a first preset function in the first evaluation model, wherein T is a positive integer; extracting P static features from the T static features based on the T contribution scores by the feature extraction module, evaluating the P static features to obtain the first indicator, wherein P is a positive integer.

[0064] Specifically, before starting model evaluation, static feature data needs to be preprocessed to ensure data quality and integrity. For example, missing value imputation, outlier handling, and categorical feature encoding can be performed. Categorical feature encoding can be done using one-hot encoding, label encoding, or embedding. Encoded features can be more effectively integrated into the model training process, enabling the model to understand and utilize categorical information for prediction.

[0065] After preprocessing the static feature data, the first evaluation model can be used to process the feature data. Firstly, the feature extraction module extracts features from the processed static feature data to identify those with greater predictive value. At this stage, feature selection techniques (such as mutual information, chi-square test, or recursive feature elimination) can be used to evaluate the correlation between each feature and the customer (i.e., asset downgrade), thereby selecting the set of features that contribute the most. Then, for the extracted static features, the feature extraction module further calculates their contribution scores using the first preset function in the first evaluation model, quantifying the impact of each feature on the prediction results. This helps the model determine which features are more critical in predicting the duration of customer asset retention, thus optimizing the model's learning focus.

[0066] Furthermore, after obtaining the contribution score of each static feature, a higher contribution score can be obtained based on the contribution score. Then, multiple static features with higher contribution scores are input into the first evaluation model for evaluation, thereby obtaining the first indicator. The evaluation process can utilize the algorithm and learning mechanism inside the model to combine static features to predict the likelihood of customer asset preservation, thereby generating a quantitative indicator that reflects customer risk based on static features.

[0067] This embodiment processes static feature data using a first evaluation model to output a first indicator, which helps financial institutions more accurately identify the risk of customer asset downgrades. This allows them to take effective operational measures at the appropriate time to reduce customer downgrades, improve the efficiency and effectiveness of customer asset management, and provide strong data support for financial institutions' customer relationship management and asset operation strategies.

[0068] Optionally, in the customer status determination method provided in this application embodiment, inputting K dynamic feature data into the second evaluation model and outputting the second indicator includes: preprocessing the K dynamic feature data to obtain K processed dynamic feature data, and sorting each processed dynamic feature data according to the timestamp of each processed dynamic feature data to obtain a dynamic feature sequence; extracting features from the dynamic feature sequence by the second evaluation model to obtain R dynamic features, and converting the format of the R dynamic features to obtain R converted dynamic features, where R is a positive integer; predicting the R converted dynamic features by the second preset function in the second evaluation model to obtain feature prediction results, and converting the feature prediction results numerically to obtain the second indicator.

[0069] Specifically, when processing dynamic feature data using the second evaluation model, the collected dynamic feature data can first be preprocessed, such as cleaning and removing missing or outlier values ​​to ensure the accuracy of model training. Furthermore, since the dynamic feature data includes changes in customer assets over time and data on consumption behavior, each dynamic feature data point can be associated with a different timestamp. At this point, the data can be sorted according to the timestamp carried by each data point to construct a dynamic feature sequence.

[0070] Furthermore, a second evaluation model is used to extract several key dynamic features from the sorted dynamic feature sequence. These extracted dynamic features are then format-converted to ensure they can be correctly read and processed by the model, resulting in converted dynamic features. This format conversion may include converting the features to a tensor format required by the model to adapt to the input requirements of the LSTM. Further, the converted dynamic features are predicted using a second preset function in the second evaluation model. This can be achieved through steps such as weight updates and activation function applications, ultimately generating feature prediction results. These results represent the model's preliminary estimate of the likelihood of future asset downgrades based on the input dynamic feature sequence, characterizing the model's judgment on the potential liquidity of customer assets. To transform the feature prediction results into more actionable indicators, numerical conversion can be performed, converting the model output into a specific second indicator.

[0071] It should be noted that, Figure 4This is a schematic diagram of the structure of the second evaluation model provided according to the embodiments of this application, such as... Figure 4 As shown, the LSTM (Long Short-Term Memory) network includes an input layer and an output layer. After receiving the input data, the input layer can convert it into input features representing a point in time in the time series. After calculation and processing by multiple layers of LSTM units, the network generates the final output sequence, which is output by the output layer.

[0072] Each unit in an LSTM can include a forget gate, an input gate, a cell state, and an output gate. The forget gate processes the input sequence and the previous hidden state using a sigmoid activation function, outputting a vector between 0 and 1. This vector is used to multiply the cell state from the previous time step, thus controlling whether information is retained or forgotten, and deciding which information to discard from the cell state. The input gate controls which new information is added to the cell state. It uses another sigmoid function to determine whether new information should be added to the cell state, and a tanh function generates a candidate value vector. This vector is multiplied element-wise by the output vector of the input gate, ultimately updating the cell state. The cell state can be considered the memory carrier of the LSTM, propagating linearly along the timeline, but its content can be modified through the control of the forget gate and the input gate. The output gate determines which parts of the cell state output are used as the hidden state at the current time step. The output gate also uses a sigmoid function to process the input and the previous hidden state, and then uses a tanh function to transform the cell state to a certain range. The two are multiplied together to obtain the output hidden state at the current time step.

[0073] Figure 4 The diagram illustrates the multi-layered structure of the LSTM network, with each layer containing multiple LSTM units. This structure allows the network to capture more complex temporal dependencies, increasing the model's expressive power and learning accuracy. The output of each LSTM unit serves as the input to the next layer, forming a deep recurrent network that helps handle long-term dependency issues. Furthermore, the LSTM network employs a GRU with an attention mechanism when processing temporal features, enabling the model to focus on the most critical parts for prediction while ignoring irrelevant information, thus improving its predictive ability on complex sequences. Each layer of the LSTM network can process sequence features independently or collaboratively. Through its multi-layered structure and attention mechanism, the model can more accurately predict the duration of customer asset retention.

[0074] This embodiment utilizes a second assessment model to evaluate dynamic characteristic data, providing financial institutions with a key basis for predicting the duration of customer asset protection. This helps financial institutions take effective measures before a customer is about to be downgraded, and facilitates decision-makers' understanding and action.

[0075] Optionally, in the method for determining customer status provided in the embodiments of this application, after determining the status of the customer to be evaluated based on probability data, the method further includes: obtaining an operational strategy list, wherein the operational strategy list includes multiple probability data and an operational strategy corresponding to each probability data; determining a target operational strategy based on the probability data; and adjusting the strategy for the customer to be evaluated based on the target operational strategy.

[0076] Specifically, after determining the status of the customer to be evaluated, since the probability data for determining the status can be values ​​of 0, 1, or 2, an operational strategy list can be obtained. This list contains a series of operational activities or service strategies for different probability levels. Each strategy is matched with specific probability data, that is, different operational responses set for the potential risk of customer asset downgrade. For example, for customers with a high probability (probability data of 2), they can be reached through manual outbound calls, while also receiving a large amount of benefits; for customers with a medium probability (probability data of 1), they can be given a medium amount of benefits and intelligent outbound calls can be used; and for customers with a low probability (probability data of 0), they can be reached only through SMS.

[0077] After obtaining the aforementioned list of operational strategies, the target operational strategy for each customer to be evaluated can be determined based on probability data. This allows for precise matching of corresponding operational activities or service plans from the operational strategy list, ensuring that the strategy aligns with the customer's risk level. Once the target operational strategy is determined, specific adjustments can be made to the operational strategy for each customer to be evaluated. For example, for high-probability customers, methods such as outbound human calls can be prioritized; while for low-probability customers, service information can be sent via SMS to maintain good customer relationships.

[0078] This embodiment utilizes probability data to determine target operational strategies, avoiding excessive operations on downgraded or low-risk customers, reducing operational costs, and increasing the probability of high-risk customers responding to retention activities, thereby improving the effectiveness of personal customer asset management and overall operational efficiency of financial institutions.

[0079] 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.

[0080] Example 2

[0081] This application also provides a customer status determination device. It should be noted that the customer status determination device of this application can be used to execute the customer status determination method provided in this application. The customer status determination device provided in this application will be described below.

[0082] According to an embodiment of this application, an apparatus for implementing the above-described method for determining customer status is also provided. Figure 5 This is a schematic diagram of a customer status determination device according to an embodiment of this application, such as... Figure 5 As shown, the device includes: a first acquisition unit 50, an input unit 51, and a second acquisition unit 52.

[0083] The first acquisition unit 50 is used to acquire N static feature data and K dynamic feature data of the customer to be evaluated. Each static feature data refers to the customer information of the customer to be evaluated, and each dynamic feature data refers to the financial feature data of the customer to be evaluated that changes over time. N and K are positive integers.

[0084] Input unit 51 is used to input N static feature data into the first evaluation model, process it to obtain the first index, and input K dynamic feature data into the second evaluation model to output the second index.

[0085] The second acquisition unit 52 is used to acquire preset weights, perform weighted calculations on the first indicator and the second indicator based on the preset weights to obtain probability data of the customer to be evaluated, and determine the status of the customer to be evaluated based on the probability data.

[0086] The customer status determination device provided in this application embodiment acquires N static feature data and K dynamic feature data of the customer to be evaluated through a first acquisition unit 50. Each static feature data refers to data representing customer information of the customer to be evaluated, and each dynamic feature data refers to financial feature data of the customer to be evaluated that changes over time. N and K are positive integers. The input unit 51 inputs the N static feature data into a first evaluation model, processes it to obtain a first indicator, and inputs the K dynamic feature data into a second evaluation model, outputting a second indicator. The second acquisition unit 52 acquires preset weights, performs weighted calculations on the first and second indicators based on the preset weights, obtains probability data of the customer to be evaluated, and determines the status of the customer to be evaluated based on the probability data. This solves the technical problem of low evaluation accuracy in evaluating customer status in related technologies. By acquiring the static and dynamic feature data of the customer to be evaluated, inputting the static and dynamic feature data into the first and second evaluation models respectively, processing them to obtain the first and second indicators, and determining the probability data based on the first and second indicators, the status of the customer to be evaluated is determined. This achieves the technical effect of improving the evaluation accuracy of customer status and sending operational strategies to customers based on their status.

[0087] Optionally, in the customer status determination device provided in this application embodiment, the first acquisition unit 50 includes: a first acquisition module, used to acquire customer information of the customer to be evaluated, wherein the customer information is used to indicate the customer identity and account information of the customer to be evaluated; a first extraction module, used to extract M customer feature data of the customer to be evaluated from the database within a preset time period based on the customer information, wherein M is a positive integer; and a first division module, used to divide the M customer feature data into data types to obtain N static feature data and K dynamic feature data, wherein the data types include at least: dynamic type and static type, and N+K=M.

[0088] Optionally, in the customer status determination device provided in this application embodiment, the input unit 51 includes: a second acquisition module, used to acquire customer feature data of Y customers within a historical time period to obtain Y sets of historical customer feature data, where Y is a positive integer; a second extraction module, used to extract the amount storage data of each customer from the Y sets of historical customer feature data to obtain Y amount storage data, and determine the indicator of each customer based on each amount storage data, wherein the indicator type of each indicator includes: a first type, a second type, and a third type; a composition module, used to construct a model training set from the Y sets of historical customer feature data and Y labels; and a training module, used to train a first preset evaluation model based on the model training set to obtain a first evaluation model, and train a second preset evaluation model based on the model training set to obtain a second evaluation model.

[0089] Optionally, in the customer status determination device provided in this application embodiment, the input unit 51 includes: a second division module, used to divide each group of historical customer feature data into data types to obtain group A of historical static feature data and group B of historical dynamic feature data, wherein A+B=Y, and A and B are positive integers; a construction module, used to construct a first model training set using group A of historical static feature data and Y labels, and construct a second model training set using group B of historical dynamic feature data and Y labels; and a combination module, used to combine the first model training set and the second model training set to obtain a model training set.

[0090] Optionally, in the customer status determination device provided in the embodiments of this application, the device further includes: a processing unit, used to preprocess the N static feature data before inputting N static feature data into the first evaluation model and processing them to obtain the first index, to obtain N processed static feature data, wherein the preprocessing method includes at least one of the following: missing value imputation, outlier handling, and category feature encoding; the input unit 51 includes: a third extraction module, used to extract features from the N processed static feature data by the feature extraction module in the first evaluation model to obtain T static features, and to determine the contribution score of the static features by a first preset function in the first evaluation model, wherein T is a positive integer; a fourth extraction module, used to extract P static features from the T static features based on the T contribution scores by the feature extraction module, to evaluate the P static features, and to obtain the first index, wherein P is a positive integer.

[0091] Optionally, in the customer status determination device provided in this application embodiment, the input unit 51 includes: a processing module, used to preprocess K dynamic feature data to obtain K processed dynamic feature data, and sort each processed dynamic feature data according to the timestamp of each processed dynamic feature data to obtain a dynamic feature sequence; a fifth extraction module, used to extract features from the dynamic feature sequence by a second evaluation model to obtain R dynamic features, and convert the format of the R dynamic features to obtain R converted dynamic features, where R is a positive integer; and a prediction module, used to predict the R converted dynamic features by a second preset function in the second evaluation model to obtain feature prediction results, and convert the feature prediction results numerically to obtain a second index.

[0092] Optionally, in the customer status determination device provided in the embodiments of this application, the device further includes: a third acquisition unit, configured to acquire an operational strategy list after determining the status of the customer to be evaluated based on probability data, wherein the operational strategy list includes multiple probability data and an operational strategy corresponding to each probability data; and a determination unit, configured to determine a target operational strategy based on the probability data, and adjust the strategy for the customer to be evaluated based on the target operational strategy.

[0093] It should be noted that the first acquisition unit 50, input unit 51, and second acquisition unit 52 mentioned above correspond to steps S201 to S203 in Embodiment 1. The instances and application scenarios implemented by the above units and corresponding steps are the same, but are not limited to the content disclosed in Embodiment 1. It should be noted that the above modules or units can be hardware or software components stored in memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n). The above units can also be part of a device and run in the computer terminal 10 provided in Embodiment 1.

[0094] Example 3

[0095] Embodiments of this application may provide a computer terminal, which may be any computer terminal device in a group of computer terminals. Optionally, in this embodiment, the aforementioned computer terminal may also be replaced with a mobile terminal or an electronic device, etc.

[0096] Optionally, in this embodiment, the computer terminal may be located in at least one of a plurality of network devices in a computer network.

[0097] In this embodiment, the computer terminal described above can execute the program code for the following steps in the method for determining customer status: acquiring N static feature data and K dynamic feature data of the customer to be evaluated, wherein each static feature data refers to data representing customer information of the customer to be evaluated, and each dynamic feature data refers to financial feature data of the customer to be evaluated that changes over time, and N and K are positive integers; inputting the N static feature data into a first evaluation model and processing it to obtain a first indicator; inputting the K dynamic feature data into a second evaluation model and outputting a second indicator; acquiring preset weights, performing weighted calculations on the first and second indicators based on the preset weights to obtain probability data of the customer to be evaluated, and determining the status of the customer to be evaluated based on the probability data.

[0098] Optionally, the computer terminal described above can execute the program code for the following steps in the method for determining customer status: obtaining customer information of the customer to be evaluated, wherein the customer information is used to indicate the customer identity and account information of the customer to be evaluated; extracting M customer feature data of the customer to be evaluated from the database within a preset time period based on the customer information, wherein M is a positive integer; dividing the M customer feature data into data types to obtain N static feature data and K dynamic feature data, wherein the data types include at least: dynamic type and static type, N+K=M.

[0099] Optionally, the aforementioned computer terminal may execute the following steps in the method for determining customer status: acquiring customer characteristic data of Y customers within a historical time period to obtain Y sets of historical customer characteristic data, where Y is a positive integer; extracting the amount storage data of each customer from the Y sets of historical customer characteristic data to obtain Y amount storage data; determining the indicators for each customer based on each amount storage data, wherein the indicator type of each indicator includes: a first type, a second type, and a third type; constructing a model training set from the Y sets of historical customer characteristic data and Y labels; training a first preset evaluation model based on the model training set to obtain a first evaluation model; and training a second preset evaluation model based on the model training set to obtain a second evaluation model.

[0100] Optionally, the aforementioned computer terminal can execute the program code for the following steps in the method for determining customer status: dividing each group of historical customer feature data into data types to obtain group A of historical static feature data and group B of historical dynamic feature data, where A+B=Y, and A and B are positive integers; constructing a first model training set using group A of historical static feature data and Y labels, and constructing a second model training set using group B of historical dynamic feature data and Y labels; and combining the first model training set and the second model training set to obtain a model training set.

[0101] Optionally, the aforementioned computer terminal may execute the program code for the following steps in the method for determining the client status: preprocessing N static feature data to obtain N processed static feature data, wherein the preprocessing method includes at least one of the following: missing value imputation, outlier handling, and categorical feature encoding; inputting the N static feature data into a first evaluation model and processing to obtain a first indicator, including: extracting features from the N processed static feature data by the feature extraction module in the first evaluation model to obtain T static features, and determining the contribution score of the static features by a first preset function in the first evaluation model, wherein T is a positive integer; extracting P static features from the T static features based on the T contribution scores by the feature extraction module, evaluating the P static features to obtain the first indicator, wherein P is a positive integer.

[0102] Optionally, the aforementioned computer terminal may execute the following steps in the method for determining the client status: preprocessing K dynamic feature data to obtain K processed dynamic feature data, and sorting each processed dynamic feature data according to the timestamp of each processed dynamic feature data to obtain a dynamic feature sequence; extracting features from the dynamic feature sequence by the second evaluation model to obtain R dynamic features, and converting the format of the R dynamic features to obtain R converted dynamic features, where R is a positive integer; predicting the R converted dynamic features by the second preset function in the second evaluation model to obtain feature prediction results, and converting the feature prediction results numerically to obtain a second index.

[0103] Optionally, the aforementioned computer terminal may execute program code for the following steps in the method for determining customer status: obtaining an operational strategy list, wherein the operational strategy list includes multiple probability data and an operational strategy corresponding to each probability data; determining a target operational strategy based on the probability data; and adjusting the strategy for the customer to be evaluated based on the target operational strategy.

[0104] Optionally, Figure 6 This is a structural block diagram of an electronic device according to an embodiment of this application. Figure 6 As shown, the electronic device may include: one or more ( Figure 6 (Only one is shown) processor 602, memory 604, memory controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0105] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the customer state determination method and apparatus in this application embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned customer state determination method. The memory may include high-speed random access memory, and may also include 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 memory remotely located relative to the processor, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0106] The processor can invoke the information and application program stored in the memory through the transmission device to execute the steps described above in the method for determining the client state.

[0107] This application provides a scheme for determining customer status. It acquires N static feature data and K dynamic feature data of the customer to be evaluated. Each static feature data represents customer information, and each dynamic feature data represents financial characteristic data of the customer changing over time. N and K are positive integers. The N static feature data are input into a first evaluation model to obtain a first indicator. The K dynamic feature data are input into a second evaluation model to output a second indicator. Preset weights are obtained, and the first and second indicators are weighted and calculated to obtain probability data of the customer to be evaluated. The customer status is determined based on the probability data. This solves the technical problem of low evaluation accuracy in related technologies when evaluating customer status. By acquiring the static and dynamic feature data of the customer to be evaluated, inputting them into the first and second evaluation models respectively, processing them to obtain the first and second indicators, and determining the probability data based on the first and second indicators, the customer status is determined, thereby improving the evaluation accuracy of customer status and enabling the delivery of operational strategies to customers based on their status.

[0108] Those skilled in the art will understand that Figure 6 The structure shown is for illustrative purposes only. Electronic devices can also be smartphones, tablets, handheld computers, mobile internet devices (MIDs), PADs, and other terminal devices. Figure 6 This does not limit the structure of the aforementioned electronic device. For example, electronic devices may also include components that are more... Figure 6 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 6 The different configurations shown.

[0109] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0110] Example 4

[0111] Embodiments of this application also provide a storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the customer status determination method provided in Embodiment 1.

[0112] Optionally, in this embodiment, the storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0113] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: acquiring N static feature data and K dynamic feature data of the customer to be evaluated, wherein each static feature data refers to data representing customer information of the customer to be evaluated, and each dynamic feature data refers to financial feature data of the customer to be evaluated that changes over time, and N and K are positive integers; inputting the N static feature data into a first evaluation model and processing it to obtain a first indicator; inputting the K dynamic feature data into a second evaluation model and outputting a second indicator; acquiring preset weights, performing weighted calculations on the first indicator and the second indicator based on the preset weights to obtain probability data of the customer to be evaluated, and determining the status of the customer to be evaluated based on the probability data.

[0114] This application also provides a computer program product, which, when executed on a data processing device, is adapted to perform method steps for determining a client state.

[0115] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0116] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0117] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of units or modules may be electrical or other forms.

[0118] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0119] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0120] If the integrated unit is implemented as 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 this application, 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. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0121] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for determining customer status, characterized in that, include: Obtain N static feature data and K dynamic feature data of the customer to be evaluated. Each static feature data refers to the customer information of the customer to be evaluated, and each dynamic feature data refers to the financial feature data of the customer to be evaluated that changes over time. N and K are positive integers. The N static feature data are input into the first evaluation model and processed to obtain the first index; the K dynamic feature data are input into the second evaluation model and output the second index. Obtain preset weights, perform weighted calculations on the first indicator and the second indicator based on the preset weights to obtain probability data of the customer to be evaluated, and determine the status of the customer to be evaluated based on the probability data.

2. The method according to claim 1, characterized in that, Obtaining N static feature data and K dynamic feature data of the customer to be evaluated includes: Obtain the customer information of the customer to be evaluated, wherein the customer information is used to indicate the customer identity and account information of the customer to be evaluated; Based on the customer information, extract M customer characteristic data of the customer to be evaluated within a preset time period from the database, where M is a positive integer; The M customer feature data are divided into data types to obtain N static feature data and K dynamic feature data, wherein the data types include at least: dynamic type and static type. N+K=M.

3. The method according to claim 1, characterized in that, The first evaluation model and the second evaluation model are trained in the following manner: Obtain customer characteristic data for Y customers within a historical time period to obtain Y sets of historical customer characteristic data, where Y is a positive integer; Extract the amount storage data of each customer from the Y sets of historical customer feature data to obtain Y amount storage data. Determine the indicators of each customer based on each amount storage data. The indicator type of each indicator includes: a first type, a second type and a third type. The model training set consists of the Y sets of historical customer feature data and the Y labels; The first preset evaluation model is trained based on the model training set to obtain the first evaluation model, and the second preset evaluation model is trained based on the model training set to obtain the second evaluation model.

4. The method according to claim 3, characterized in that, The model training set, consisting of the Y sets of historical customer feature data and Y labels, includes: Each group of historical customer feature data is divided into data types to obtain group A of historical static feature data and group B of historical dynamic feature data, where A+B=Y, and A and B are positive integers; A first model training set is constructed using the historical static feature data of group A and Y labels, and a second model training set is constructed using the historical dynamic feature data of group B and Y labels. The model training set is obtained by combining the first model training set and the second model training set.

5. The method according to claim 1, characterized in that, Before inputting the N static feature data into the first evaluation model and processing them to obtain the first index, the method further includes: The N static feature data are preprocessed to obtain N processed static feature data, wherein the preprocessing method includes at least one of the following: missing value imputation, outlier handling, and category feature encoding; The N static feature data are input into the first evaluation model, and the resulting first index includes: Features are extracted from the N processed static feature data by the feature extraction module in the first evaluation model to obtain T static features, and the contribution score of the static features is determined by the first preset function in the first evaluation model, where T is a positive integer; The feature extraction module extracts P static features from the T static features based on the T contribution scores, evaluates the P static features, and obtains the first index, where P is a positive integer.

6. The method according to claim 1, characterized in that, The K dynamic feature data are input into the second evaluation model, and the output second index includes: The K dynamic feature data are preprocessed to obtain K processed dynamic feature data, and each processed dynamic feature data is sorted according to the timestamp of each processed dynamic feature data to obtain a dynamic feature sequence. The second evaluation model extracts features from the dynamic feature sequence to obtain R dynamic features, and the format of the R dynamic features is converted to obtain R converted dynamic features, where R is a positive integer; The second preset function in the second evaluation model is used to predict the R transformed dynamic features to obtain feature prediction results, and the feature prediction results are numerically converted to obtain the second index.

7. The method according to claim 1, characterized in that, After determining the status of the customer to be evaluated based on the probability data, the method further includes: Obtain an operational strategy list, wherein the operational strategy list includes multiple probability data and the operational strategy corresponding to each probability data; Based on the probability data, a target operational strategy is determined, and the strategy is adjusted for the customer to be evaluated based on the target operational strategy.

8. A device for determining customer status, characterized in that, include: The first acquisition unit is used to acquire N static feature data and K dynamic feature data of the customer to be evaluated, wherein each static feature data refers to the customer information of the customer to be evaluated, and each dynamic feature data refers to the financial feature data of the customer to be evaluated that changes over time, and N and K are positive integers. The input unit is used to input the N static feature data into the first evaluation model, process it to obtain the first index, and input the K dynamic feature data into the second evaluation model to output the second index. The second acquisition unit is used to acquire a preset weight, perform a weighted calculation on the first indicator and the second indicator based on the preset weight to obtain the probability data of the customer to be evaluated, and determine the status of the customer to be evaluated based on the probability data.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device on which the computer-readable storage medium is located to perform the client status determination method according to any one of claims 1 to 7.

10. An electronic device, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, performs the method for determining the client state as described in any one of claims 1 to 7.

11. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the method for determining the customer status as described in any one of claims 1 to 7.

Images