A high net worth user screening method, device, electronic equipment and program product

Abstract

This application discloses a method, apparatus, electronic device, and program product for screening high-net-worth users, relating to the field of user profiling technology. The method includes: acquiring feature data of users to be screened, including user occupation information, terminal price information, luxury goods interaction information, car owner identity information, and average monthly consumption contribution; inputting the feature data into a user net worth screening model to obtain the net worth screening result for the users to be screened, the user net worth screening model being trained based on the XGBoost ensemble learning algorithm and through structural risk minimization; and determining whether the users to be screened are high-net-worth users based on the net worth screening result. This application, by using a user net worth screening model to determine whether a user belongs to high-net-worth users based on multi-dimensional feature data, effectively improves the accuracy of identifying high-net-worth users from communication users.

Description

Technical Field

[0001] This application relates to the field of user profiling technology, and in particular to a method, apparatus, electronic device, and program product for screening high-net-worth users. Background Technology

[0002] As the traditional telecommunications market approaches saturation, managing existing users and unlocking their value has become a core task for operators. Accurate identification of high-net-worth users is a crucial prerequisite for implementing refined operations and enhancing customer lifetime value. Existing technological solutions for user value assessment generally suffer from limitations in data dimensions and application scenarios. They are mostly based on single or limited-dimensional data such as consumption records, specific scenario behaviors, or vertical domain preferences for modeling. This results in models and rules being tightly coupled to pre-defined vertical domains, lacking cross-scenario transferability. When faced with comprehensive features from multiple sources, high dimensions, and varying sparsity, existing technological solutions struggle to deeply mine and integrate complex user characteristics, thus affecting the comprehensiveness and accuracy of high-net-worth user identification.

[0003] Therefore, there is an urgent need to provide a high-net-worth user screening method that can effectively integrate multi-dimensional heterogeneous features and accurately identify them. Summary of the Invention

[0004] This application provides a method, apparatus, electronic device, and program product for screening high-net-worth individuals.

[0005] Firstly, this application provides a method for screening high-net-worth individuals, including: Obtain the feature data of the users to be filtered; The feature data is input into the user net worth screening model to obtain the net worth screening results of the users to be screened; the user net worth screening model is based on the XGBoost ensemble learning algorithm and is trained by minimizing structural risk. Based on the net worth screening results, determine whether the users to be screened are high-net-worth users.

[0006] In some embodiments, the feature data includes user occupation information, terminal price information, luxury goods interaction information, car owner identity information, and user's average monthly consumption contribution value; The average monthly consumption contribution of users is calculated based on the transaction data of the users to be screened within a preset continuous time period.

[0007] In some embodiments, obtaining user occupation information includes: Based on users' online behavior data and location data, the occupational information of users to be screened is determined.

[0008] In some embodiments, obtaining terminal price information includes: Obtain the terminal device identifier of the user to be filtered; Based on the terminal device identifier, the terminal price information is obtained.

[0009] In some embodiments, obtaining luxury goods interaction information includes: Obtain preset network address information associated with luxury goods; By matching the internet access data of the users to be screened with the network address information associated with luxury goods, the interaction information of the users with luxury goods can be obtained.

[0010] In some embodiments, obtaining vehicle owner identity information includes: Calculate the correlation between the user's movement trajectory data and the movement trajectory data of at least one vehicle. Based on the degree of association, target vehicles that have a car owner relationship with the user are identified from the vehicles.

[0011] In some embodiments, calculating the correlation between the user to be screened and the vehicle includes: Extract feature indicators from the movement trajectory data of the users to be screened and the movement trajectory data of the vehicles, respectively. Calculate the similarity between users and vehicles to be screened on corresponding feature indicators; for numerical feature indicators, the correlation coefficient method is used to calculate similarity, and for character feature indicators, the edit distance method is used to calculate similarity. Based on the similarity of each feature indicator and the preset weight value, a weighted calculation is performed to obtain the correlation between the user to be screened and the vehicle.

[0012] Secondly, this application provides a high-net-worth user screening device, comprising: The data acquisition module is configured to acquire feature data of users to be filtered. The user screening module is configured to input feature data into the user net worth screening model to obtain the net worth screening results of the users to be screened; wherein, the user net worth screening model is based on the XGBoost ensemble learning algorithm and is trained by minimizing structural risk; The result determination module is configured to determine whether the user to be screened is a high-net-worth user based on the net worth screening results.

[0013] Thirdly, this application provides an electronic device, including a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement any one of the high-net-worth user screening methods.

[0014] Fourthly, this application provides a computer program product, including a computer-readable storage medium on which a computer program is stored, wherein the computer program, when executed by a processor, implements any one of the high-net-worth user screening methods.

[0015] The at least one technical solution adopted in this application can achieve the following beneficial effects: by acquiring feature data including user occupation information, terminal price information, luxury goods interaction information, car owner identity information, and user's average monthly consumption contribution value, and by using a user net worth screening model trained with parameter optimization based on the XGBoost ensemble learning algorithm and minimizing structural risk to process the feature data, it is possible to determine whether a user belongs to a high-net-worth user based on multi-dimensional feature data. This effectively overcomes the shortcomings of existing technologies such as single data dimensions and insufficient model generalization ability, and can more comprehensively and accurately identify high-net-worth users from communication users, providing a reliable basis for subsequent precision marketing and differentiated services.

[0016] It should be understood that both the foregoing general description and the following detailed description are exemplary and intended to provide further illustration of the claimed technology. Attached Figure Description

[0017] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0018] Figure 1 This illustration schematically shows an overall process diagram of a high-net-worth user screening method according to an embodiment of this application; Figure 2 This illustration schematically shows a process for obtaining user occupation information in a high-net-worth user screening method according to an embodiment of this application; Figure 3 This illustration shows a schematic diagram of the terminal price information acquisition process of a high-net-worth user screening method according to an embodiment of this application; Figure 4 This illustration shows a schematic diagram of the luxury goods interactive information acquisition process of a high-net-worth user screening method according to an embodiment of this application; Figure 5 This illustration schematically shows a process for obtaining vehicle owner identity information according to an embodiment of the high-net-worth user screening method of this application; Figure 6 This illustration schematically shows a flowchart of the user-vehicle correlation calculation process for a high-net-worth user screening method according to an embodiment of this application; Figure 7 This illustration schematically shows an overall structural diagram of a high-net-worth user screening device according to an embodiment of this application; Figure 8 This illustration schematically shows a data acquisition module structure of a high-net-worth user screening device according to an embodiment of this application; Figure 9 An exemplary block diagram of a computer program product of a high-net-worth user screening method according to an embodiment of this application is shown schematically.

[0019] In the diagram: 701, Data Acquisition Module; 702, User Filtering Module; 703, Result Confirmation Module; 704, Occupational Information Acquisition Submodule; 705, Terminal Price Acquisition Submodule; 706, Luxury Goods Interaction Acquisition Submodule; 707, Car Owner Identity Acquisition Submodule; 708, Average Monthly Income Acquisition Submodule; 901, Computer Program. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application more apparent, exemplary embodiments according to this application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0021] Figure 1 The illustration shows an overall flowchart of a high-net-worth user screening method according to an embodiment of this application.

[0022] like Figure 1 As shown, the steps include: S101, Obtain the feature data of the users to be screened. Feature data is a collection of multi-dimensional information used to assess a user's net worth level. Specifically, features are acquired and constructed, including but not limited to the following dimensions: user occupation information obtained from analysis of user online behavior and location data; price information of the user's terminal device obtained by matching the user's terminal device identifier with pre-collected market prices; user's luxury goods interaction information obtained by matching user online access records with preset luxury goods network addresses; user's vehicle owner identity and associated vehicle information calculated based on the similarity between the user and vehicle movement trajectories; and the user's average monthly consumption contribution value statistically obtained from the user's communication consumption records. After acquisition, the feature data undergoes cleaning, transformation, and standardization to form a normalized feature vector for subsequent user net worth screening model input.

[0023] S102, the feature data is input into the user net worth screening model. The user net worth screening model is a machine learning classification model built based on the gradient boosting ensemble learning framework. Specifically, in a preferred embodiment, the user net worth screening model is built and trained based on the XGBoost model. During the offline training phase, the user net worth screening model is trained using a historical user dataset containing labeled multidimensional features that includes both positive and negative samples, and the parameters are tuned according to the principle of minimizing structural risk, so as to effectively control the model complexity while ensuring the model's prediction accuracy, thereby improving its generalization ability.

[0024] (Formula 1) Formula 1 is a mathematical formula for minimizing structural risk, used to define the structural risk of the user net worth screening model. Specifically, structural risk Risk from experience With regularization term We get the weighted average, where To balance experience risk With regularization term Hyperparameters.

[0025] (Formula 2) Formula 2 represents empirical risk. The calculation formula, where The number of training samples, The loss function is used to measure the true label of a single sample. Predicted values ​​from user net worth screening model The error is calculated by averaging the loss over all samples, reflecting the degree of fit of the user net worth screening model to the training data.

[0026] (Formula 3) Formula 3 is the regularization term for quantifying model complexity. Adapted to decision tree models, among which It is the number of leaf nodes in the decision tree. It is the first The weight of each leaf node, and It is a regularization hyperparameter, where, Constrain the number of leaf nodes Penalize node weights to control model complexity and prevent the model from overfitting noise in the training data.

[0027] Among them, positive samples correspond to high-net-worth users, and negative samples correspond to general users.

[0028] S103, the user net worth screening model processes and calculates the feature data to obtain the net worth screening results for the users to be screened. Specifically, the user net worth screening model processes the input feature vector based on its optimal parameter set trained according to the principle of minimizing structural risk. The user net worth screening model performs hierarchical judgment and combination learning on the input features through multiple base decision trees generated sequentially within it. Each newly added decision tree is trained to fit the residual between the prediction results of all previous tree combinations and the true labeled values ​​of the samples, and in this process, the importance of features is automatically evaluated to select split points. Finally, the user net worth screening model obtains the net worth screening results representing the net worth level of the users to be screened by weighted integration of the outputs of all base decision trees.

[0029] (Formula 4) Formula 4 is the XGBoost model's first... The objective function for each iteration is derived based on a second-order Taylor expansion. For loss function, and Loss function pairs The first and second partial derivatives, For the first The prediction output of each tree for this sample. For the first The regularization term for each tree.

[0030] (Formula 5) Formula 5 is the loss function for the previous... -1 round of forecast value The first-order partial derivative.

[0031] (Formula 6) Formula 6 is the loss function for the previous... -1 round of forecast value The second-order partial derivatives of .

[0032] (Formula 7) Formula 7 is the first The regularization term of the decision tree, where This represents the number of leaf nodes in the tree. For the first The weight of each leaf node, and For hyperparameters, Constrain the number of leaf nodes. The weights of leaf nodes are penalized to control the complexity of the tree and prevent the model from overfitting noise in the training data.

[0033] S104. Based on the net worth screening results, determine whether the user to be screened is a high-net-worth user. Specifically, compare the net worth screening results output by the user net worth screening model with a preset judgment threshold. If the net worth screening results meet the judgment threshold, the user to be screened is determined to be a high-net-worth user; otherwise, they are determined to be a general user. Based on the determination result of whether the user to be screened is a high-net-worth user, decisions are made regarding subsequent processes such as targeted marketing, differentiated services, or resource allocation.

[0034] Figure 2 The illustration shows a schematic diagram of the user occupation information acquisition process of a high-net-worth user screening method according to an embodiment of this application.

[0035] like Figure 2 As shown, the steps include: S201, Obtain internet access behavior data and location data of the users to be filtered. Internet access behavior data includes at least the identifiers of the applications used by the user within a preset time period, the single and cumulative usage duration of each application, the data traffic generated, and the access time sequence. Location data includes at least the identifiers of the base stations the user connected to within the preset time period, the identifiers of the cells to which the user belongs, and the corresponding timestamps and dwell time information.

[0036] S202, the internet browsing behavior data is associated with a pre-defined set of occupation-related applications to generate user occupational application metrics. The occupational application set defines the mapping relationship between different occupational categories and one or more typical applications. During processing, the application identifiers in the acquired internet browsing behavior data are matched with the occupational application set. For each matched occupational category, statistics such as total usage time, total data traffic, access frequency, and active time period distribution for that user are calculated to form user occupational application metrics. These metrics quantify the occupational tendencies exhibited by the user through application usage behavior.

[0037] S203, the location data is matched with preset workplace base station information to generate a user workplace index. The workplace base station information identifies the set of base stations or cells covered by typical physical workplaces for specific professions, such as courts, hospitals, schools, and industrial parks. During processing, the base station or cell identifiers in the acquired location data are matched with the workplace base station information. For each matched workplace category, statistics such as the user's total dwell time, number of visits, and regularity within the relevant base station / cell range are calculated to form the user workplace index. This index quantifies the workplace relevance exhibited by the user's physical location-based dwelling behavior.

[0038] S204, Based on user occupation application metrics and user occupation location metrics, determine the occupational information of the users to be screened. Analyze and determine the generated user occupation application metrics and user occupation location metrics. In one embodiment, by constructing an occupational classification model, the joint feature vector composed of user occupation application metrics and user occupation location metrics is input into the occupational classification model. The occupational classification model learns the combined features of different occupations in application usage patterns and location dwell patterns, and outputs the probability or confidence level of the occupational category to which the user to be screened belongs. The occupational category with the highest probability or confidence level meeting the threshold is determined as the occupational information of the user to be screened. The occupational classification model can be based on classification models including but not limited to logistic regression, decision trees, random forests, or support vector machines.

[0039] In another embodiment, rule matching is used, such as directly determining that an occupation is a user's occupation if it matches both application metrics and location metrics.

[0040] Figure 3 The illustration shows a schematic diagram of the terminal price information acquisition process of a high-net-worth user screening method according to an embodiment of this application.

[0041] like Figure 3 As shown, the steps include: S301, Obtain the terminal device identifier of the user to be screened. The terminal device identifier is a code used to uniquely identify the hardware model of the user's terminal in the mobile network.

[0042] S302, based on the terminal device identifier, query the pre-set mapping relationship between the terminal device identifier and the market price. This mapping relationship is achieved by using network information collection technology to obtain, in batches, the official suggested retail price, current market transaction price, or price range for each terminal model from data sources such as the terminal manufacturer's official website, mainstream e-commerce platforms, and publicly available market quotations. The collected price data is then cleaned, deduplicated, and aggregated to obtain a mapping relationship with the terminal device identifier as the key and one or more corresponding price indicators as the value.

[0043] S303: Based on the mapping relationship between terminal device identifiers and market prices, the terminal price information corresponding to the terminal device identifier is obtained. When a record matching the terminal device identifier is successfully retrieved in the mapping relationship between terminal device identifiers and market prices, the price data stored in that record is extracted as the terminal price information. The terminal price information can be a specific price value, such as the average market price, or a price range, such as the lowest price and the highest price.

[0044] In one embodiment, if the retrieved terminal price information is a price range, its average value is taken as the representative value. If no exact match is found, fuzzy matching or default value processing can be performed. The obtained terminal price information is used as a quantitative feature reflecting the user's spending power and preferences and is incorporated into the feature data of the users to be screened.

[0045] Figure 4 The illustration shows a schematic diagram of the luxury goods interactive information acquisition process of a high-net-worth user screening method according to an embodiment of this application.

[0046] like Figure 4 As shown, the steps include: S401, Obtain preset luxury brand-related network address information. This information includes, but is not limited to: Uniform Resource Locators (URLs) of well-known luxury brand official websites, bag names or identifiers of their official mobile applications, paths to official mini-programs, and official social media accounts or public accounts. This information is collected from brand websites, app stores, and publicly available information using web crawling technology, and then verified and cleaned.

[0047] S402, Obtain Internet access data of the users to be screened. The Internet access data records the network activities of the users to be screened within a preset observation period, including at least: the domain names or complete URLs of the websites visited by the user, the network service requests initiated, the mobile application identifiers used, and the corresponding access timestamps, access frequency and / or session duration, etc., obtained from data sources such as core network signaling of the mobile network, deep packet inspection system, or terminal application logs authorized by the user.

[0048] S403, Match internet access data with luxury brand-related network address information, and determine luxury brand interaction information based on the matched access patterns. Perform correlation analysis on the acquired internet access data and luxury brand-related network address information. Specifically, compare and match each network resource access recorded in the internet access data, such as the accessed URL and used app, with the luxury brand-related network address information. Based on successfully matched records, determine the user's interaction behavior with a specific luxury brand. Based on statistical analysis of these interactions, generate luxury brand interaction information.

[0049] S404 outputs luxury goods interaction information from the users to be screened. This outputs the identified and quantified luxury goods interaction information, reflecting user consumption preferences, brand awareness, and potential spending power.

[0050] Figure 5 The illustration shows a schematic diagram of the process for obtaining vehicle owner identity information according to an embodiment of the high-net-worth user screening method of this application.

[0051] like Figure 5 As shown, the steps include: S501, acquire the movement trajectory data of the user to be screened and the movement trajectory data of at least one vehicle. The movement trajectory data is a sequence of location changes of the target within a preset time period. Specifically, the movement trajectory data of the user to be screened is generated based on the signaling interaction between the terminal device and the cellular network, and may include information such as the base stations or geographical locations passed, dwell time, movement speed, and timestamp sequence. The movement trajectory data of the vehicle can be acquired through in-vehicle intelligent terminals, vehicle networking systems, or mobile devices bound to the vehicle, and it also includes information such as the base stations or geographical locations passed, dwell time, movement speed, and timestamp sequence to facilitate subsequent comparison and analysis.

[0052] S502, based on the movement trajectory data of the users to be screened and the movement trajectory data of vehicles, calculate the correlation between the two. Extract comparable feature indicators from the movement trajectory data of the users and vehicles to be screened, including but not limited to: the number of times they pass through the same or adjacent geographical locations within the same time period, the overlap of stop points, the similarity of movement paths, and the consistency of travel time patterns. One or more similarity calculation methods are used to calculate the similarity of the feature indicators. For example, for the number of overlapping geographical locations, an algorithm based on overlapping point counting is used; for the spatial morphological similarity of movement paths, an algorithm based on trajectory distance is used; for the consistency of travel time patterns, a dynamic time warping algorithm based on time-series matching is used. The various similarities are weighted and summed according to preset weights to obtain a numerical value representing the degree of correlation between the user and vehicle trajectories, i.e., the correlation degree.

[0053] S503, based on the calculated correlation degree, determine the target vehicles that have a car owner relationship with the user to be screened. Specifically, sort the correlation degree between the user to be screened and each vehicle or compare it with a preset correlation degree threshold. In one embodiment, the vehicle with the highest correlation degree with the user to be screened is determined as the target vehicle with a car owner relationship.

[0054] In another embodiment, if the correlation between a vehicle and a user continues to exceed a preset correlation threshold, then the vehicle is determined to be the target vehicle.

[0055] S504 outputs the vehicle owner identity information of the users to be filtered. Specifically, the vehicle owner identity information includes at least: whether the user to be filtered is identified as a vehicle owner and the associated vehicle information. The associated vehicle information includes vehicle identification and price information.

[0056] Figure 6 The illustration shows a schematic diagram of the user-vehicle correlation calculation process of a high-net-worth user screening method according to an embodiment of this application.

[0057] like Figure 6 As shown, the steps include: S601, extract feature indicators from the movement trajectory data of users and vehicles to be screened. Feature indicators are parameters that reflect the spatiotemporal behavior patterns of users and vehicles, calculated or directly extracted from their trajectory data. These include both numerical and character indicators. Numerical indicators are continuous or discrete values ​​that can be arithmetically calculated, such as the number of the same base stations or geographical locations passed within a preset time period, the difference in dwell time at the same station, and the average movement speed. Character indicators are strings or classification labels that represent specific patterns or states, such as the name of the station with the longest dwell time in a day or its latitude and longitude coordinates, high-frequency travel time labels, and codes for the type of frequently visited area.

[0058] S602. For numerical indicators among the feature indicators, the correlation coefficient method is used to calculate their similarity. Specifically, the Pearson correlation coefficient method is used to calculate a correlation coefficient between -1 and 1 by analyzing the covariance and standard deviation between the user indicator value series and the vehicle indicator value series. The closer the absolute value of this coefficient is to 1, the higher the linear correlation between the two series, that is, the more similar the behavioral patterns exhibited on this indicator. Finally, this correlation coefficient is used as the similarity score for this numerical indicator.

[0059] (Formula 8) Formula 8 is the formula for calculating the Pearson correlation coefficient, used to quantify the degree of linear correlation between the numerical characteristic index sequences of users and vehicles. Among them, The number of samples for the feature index. For the user's characteristic indicator sequence, For the vehicle's characteristic index sequence; The sum of the products of corresponding elements of the two sequences. , These are the sums of the user and vehicle sequence elements, respectively. , These are the sums of the squares of the user and vehicle sequence elements, respectively. The calculated coefficients are... The value range of is [-1, 1]. The closer the absolute value is to 1, the higher the linear correlation between the user and the vehicle on this numerical feature, meaning the stronger the similarity in their behavioral patterns.

[0060] S603, For character-type indicators among the feature indicators, the similarity is calculated using the edit distance method. In one embodiment, the Levenstein distance algorithm is used to measure the difference between two strings by calculating the minimum number of single-character editing operations (insertion, deletion, and replacement) required to convert the string of the user indicator into the string of the vehicle indicator. The smaller the edit distance, the more similar the two strings are. The edit distance is normalized or converted into a similarity score, which is used as the similarity of the character-type indicator.

[0061] (Formula 9) The minimum edit distance between the user and vehicle character feature strings is calculated based on Formula 9. Let... For user strings, For vehicle strings, For the two strings before , The distance of a substring of 1 character. When the user string When empty, When vehicle string When empty, When user string and vehicle strings If none of the values ​​are empty, then the minimum cost of the deletion, insertion, and replacement operations is taken. The final result is... The smaller the value, the better the user string. and vehicle strings The higher the feature similarity, the better. , respectively The lengths of the user string and the vehicle string.

[0062] S604. Based on the similarity of each feature indicator and its preset weight value, a weighted calculation is performed to obtain the correlation between the user to be screened and the vehicle. Specifically, a preset weight value is assigned to each feature indicator, which reflects the importance of that feature indicator in determining the relationship between the user and the vehicle. The similarity of each feature indicator and its corresponding weight value are weighted and summed to obtain the correlation between the user to be screened and the vehicle. The higher the correlation value, the more synchronized and consistent the user's and the vehicle's movement trajectories are across multiple dimensions, and the greater the possibility of a relationship between the user and the vehicle owner.

[0063] Figure 7 The illustration shows an overall structural diagram of a high-net-worth user screening device according to an embodiment of this application.

[0064] like Figure 7 As shown, the device includes a data acquisition module 701, a user screening module 702, and a result determination module 703.

[0065] The data acquisition module 701 is configured to acquire the feature data of the users to be screened; the feature data is a collection of multi-dimensional information used to evaluate the net worth level of users.

[0066] The user screening module 702 is configured to input feature data into a pre-trained user net worth screening model to obtain the net worth screening results for users to be screened. The user net worth screening model is a machine learning classification model trained based on the XGBoost ensemble learning algorithm and optimized for parameters using the principle of minimizing structural risk.

[0067] The result determination module 703 is configured to determine whether the user to be screened is a high-net-worth user based on the net value screening results.

[0068] Figure 8 The illustration shows a schematic diagram of the data acquisition module structure of a high-net-worth user screening device according to an embodiment of this application.

[0069] like Figure 8 As shown, the data acquisition module 701 includes a professional information acquisition submodule 704, a terminal price acquisition submodule 705, a luxury goods interaction acquisition submodule 706, a car owner identity acquisition submodule 707, and a monthly average income acquisition submodule 708.

[0070] Among them, the occupational information acquisition submodule 704 is configured to determine the occupational information of the users to be screened based on their online behavior data and location data.

[0071] The terminal price acquisition submodule 705 is configured to match the price information of the terminal device used by the user to be screened based on the terminal device identifier of the user to be screened.

[0072] The luxury goods interaction acquisition submodule 706 is configured to match the internet access data of the users to be screened with the preset luxury goods associated network address information to obtain the luxury goods interaction information of the users to be screened.

[0073] The vehicle owner identity acquisition submodule 707 is configured to determine the vehicle owner identity and associated vehicle information of the user to be screened based on the movement trajectory data of the user to be screened and the movement trajectory data of at least one vehicle.

[0074] The monthly average income acquisition submodule 708 is configured to calculate the monthly average consumption contribution value of the users to be screened based on their communication consumption records within a preset continuous time period.

[0075] Figure 9 An exemplary block diagram of a computer program product of a high-net-worth user screening method according to an embodiment of this application is shown schematically.

[0076] like Figure 9 As shown, the computer program product stores a computer program 901, which, when executed by a processor, implements the method provided in any embodiment of this application.

[0077] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0078] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0079] Additionally, as used herein, the "or" used in a list of items beginning with "at least one" indicates a separate list, such that a list of, for example, "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not imply that the described example is preferred or better than other examples.

[0080] It should also be noted that in the system and method of this application, the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalent solutions of this application.

[0081] Various changes, substitutions, and modifications can be made to the technology described herein without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of this application is not limited to the specific aspects of the processes, machines, manufacturing, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufacturing, events, means, methods, or actions that perform substantially the same function or achieve substantially the same result as the corresponding aspects described above can be utilized. Therefore, the appended claims include such processes, machines, manufacturing, events, means, methods, or actions within their scope.

[0082] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0083] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for screening high-net-worth individuals, characterized in that, include: Obtain the feature data of the users to be filtered; The feature data is input into the user net worth screening model to obtain the net worth screening results of the users to be screened; wherein, the user net worth screening model is based on the XGBoost ensemble learning algorithm and is trained by minimizing structural risk. Based on the net worth screening results, determine whether the users to be screened are high-net-worth users.

2. The high-net-worth user screening method according to claim 1, characterized in that, The feature data includes user occupation information, terminal price information, luxury goods interaction information, car owner identity information, and user's average monthly consumption contribution value; The average monthly consumption contribution of the user is calculated based on the transaction data of the user to be screened within a preset continuous time period.

3. The high-net-worth user screening method according to claim 2, characterized in that, Obtaining the user's occupational information includes: Based on the user's online behavior data and location data, the user's occupation information is determined for the users to be screened.

4. The high-net-worth user screening method according to claim 2, characterized in that, Obtaining the terminal price information includes: Obtain the terminal device identifier of the user to be screened; The terminal price information is obtained by matching the terminal device identifier.

5. The high-net-worth user screening method according to claim 2, characterized in that, Obtaining the interaction information of the luxury goods includes: Obtain preset network address information associated with luxury goods; The online access data of the users to be screened is matched with the network address information associated with the luxury goods to obtain the interaction information of the luxury goods of the users to be screened.

6. The high-net-worth user screening method according to claim 2, characterized in that, Obtaining the vehicle owner's identity information includes: Calculate the correlation between the user's movement trajectory data and the movement trajectory data of at least one vehicle. Based on the correlation, target vehicles that have a vehicle owner relationship with the user are identified from the vehicles.

7. The high-net-worth user screening method according to claim 6, characterized in that, Calculating the correlation between the user to be screened and the vehicle includes: Extract feature indicators from the movement trajectory data of the users to be screened and the movement trajectory data of the vehicles, respectively; Calculate the similarity between the user to be screened and the vehicle on the corresponding feature indicators; wherein, the correlation coefficient method is used to calculate the similarity for numerical feature indicators, and the edit distance method is used to calculate the similarity for character feature indicators; Based on the similarity of each feature indicator and the preset weight value, a weighted calculation is performed to obtain the correlation between the user to be screened and the vehicle.

8. A high-net-worth user screening device, characterized in that, include: The data acquisition module is configured to acquire feature data of users to be filtered. The user screening module is configured to input the feature data into the user net worth screening model to obtain the net worth screening results of the users to be screened; wherein, the user net worth screening model is based on the XGBoost ensemble learning algorithm and is trained by minimizing structural risk. The result determination module is configured to determine whether the user to be screened is a high-net-worth user based on the net worth screening results.

9. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the high-net-worth user screening method according to any one of claims 1 to 7.

10. A computer program product comprising a computer-readable storage medium on which a computer program is stored, characterized in that, When the computer program is executed by a processor, it implements the high-net-worth user screening method according to any one of claims 1 to 7.

Images