Internet of Things user classification method, device and equipment and storage medium
By obtaining multi-dimensional information of IoT users and using support vector machine models, we can accurately judge whether there is a risk of churn by IoT users, solve the problem of inaccurate characterization of churn trends caused by single factors in the existing technology, and improve the accuracy of user classification and loss risk warning capability.
Patent Information
- Application Number
- CN202411087365.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-12-03
AI Technical Summary
The existing technology cannot accurately determine whether there is a risk of churn for Internet of Things users, and the analysis factors are single, and the user's churn trends cannot be objectively and accurately characterized, resulting in inaccurate classification identification.
By obtaining the basic information of IoT users, card usage information, operating status information and enterprise negative public opinion information, input it to the pre-trained support vector machine model, determine the relative position relationship between the feature set of churn indicators and the classification hyperplane, and judge the User churn category.
It realizes accurate classification and churn risk judgment of IoT users, improves the accuracy of classification, can objectively and accurately characterize user churn trends, and helps operators retain high-risk users in advance.
Smart Images

Figure CN119066501A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of data processing technology, and in particular, relates to a user churn warning method, device, equipment, storage medium and computer program product. Background Art
[0002] IoT cards are based on IoT dedicated networks, providing IoT terminals with basic communication functions such as voice, SMS, and traffic, and realizing communication connections between people, machines, and objects. They belong to the core network connection pipeline part of the IoT "end-pipe-cloud", enabling intelligent equipment to enter the cloud and build smart applications. They have gradually penetrated into various fields of infrastructure, people's livelihood, and industry production. In the process of using IoT cards, it is difficult to guarantee that users will always stay. User loss brings losses to operators. Therefore, it is necessary to classify and identify IoT users who are at risk of loss in order to retain users in advance.
[0003] The existing technology analyzes the card usage of IoT users, uses a weighted algorithm to calculate the churn factor of IoT users, and determines whether the user has a churn risk based on the churn factor. This method has a single analysis factor, and the churn factor obtained by weighted calculation cannot objectively and accurately describe the user's churn trend, cannot accurately classify IoT users, and cannot accurately determine whether IoT users have a churn risk. Summary of the invention
[0004] The embodiments of the present application provide a method, apparatus, device, storage medium and computer program product for classifying IoT users, which can accurately classify IoT users and accurately determine whether there is a risk of IoT user loss.
[0005] In a first aspect, an embodiment of the present application provides a method for classifying Internet of Things users, the method comprising:
[0006] Obtain the churn indicator feature set of IoT users; the churn indicator feature set includes basic information feature data, card usage information feature data, business status information feature data, and corporate negative public opinion information feature data;
[0007] Input the churn indicator feature set into a pre-trained target classification model to determine the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model; the target classification model is obtained by training based on a support vector machine model;
[0008] Determine the churn categories of IoT users based on their relative position relationships.
[0009] In a second aspect, an embodiment of the present application provides an IoT user classification device, the device comprising:
[0010] An acquisition module is used to acquire a churn indicator feature set of IoT users; the churn indicator feature set includes basic information feature data, card usage information feature data, business status information feature data, and corporate negative public opinion information feature data;
[0011] A determination module is used to input the churn indicator feature set into a pre-trained target classification model to determine the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model; the target classification model is obtained by training based on a support vector machine model;
[0012] The classification module is used to determine the churn category of IoT users based on relative position relationships.
[0013] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the method for classifying Internet of Things users as in the first aspect is implemented.
[0014] In a fourth aspect, an embodiment of the present application provides a computer storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, the method for classifying Internet of Things users as in the first aspect is implemented.
[0015] In a fifth aspect, an embodiment of the present application provides a computer program product. When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the Internet of Things user classification method as in the first aspect.
[0016] The user churn warning method, device, equipment, storage medium and computer program product of the embodiment of the present application obtains a churn indicator feature set, that is, obtains the basic information feature data, card usage information feature data, business status information feature data and corporate negative public opinion information feature data of the Internet of Things users, and conducts a comprehensive and multi-dimensional analysis of the user information data, objectively and accurately depicts the user churn trend, and can help the model improve the accuracy of the classification of Internet of Things users. The churn indicator feature set is input into the trained target classification model, and the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model is determined. Based on the relative position relationship, the churn category of the Internet of Things user is determined, which can improve the accuracy of the classification of Internet of Things users and accurately determine whether the Internet of Things users are at risk of churn. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.
[0018] Figure 1 It is a structural diagram of an Internet of Things user classification system provided by an embodiment of the present application;
[0019] Figure 2 This is a flow chart of a method for classifying Internet of Things users provided by an embodiment of the present application;
[0020] Figure 3 It is a flowchart of an implementation method of training a support vector machine model to obtain a target classification model provided by an embodiment of the present application;
[0021] Figure 4 It is a structural diagram of an Internet of Things user classification device provided by an embodiment of the present application;
[0022] Figure 5 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION
[0023] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.
[0024] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0025] The IoT card is based on the IoT dedicated network and provides basic communication functions such as voice, SMS, and traffic for IoT terminals. It is a dedicated phone card that realizes communication connections between people, machines, and objects. It belongs to the pipeline part of the core network connection in the IoT "end-pipe-cloud". It enables intelligent equipment to enter the cloud and then build smart applications. It has gradually penetrated into various fields of infrastructure, people's livelihood, and industry production. Vigorously developing the Internet of Things is the key to operators achieving innovative breakthroughs, and it is also the main battlefield for operators to compete in the market. Users cannot guarantee that they will always stay in the process of using IoT cards. User loss brings losses to operators. Therefore, it is of utmost importance to classify and identify IoT users who are at risk of loss, predict whether users are at risk of loss, and retain users in advance.
[0026] The existing solutions generally rely on staff, such as account managers, to maintain customer relationships, visit users offline or by phone, and understand the user's business status and whether there are business difficulties, etc., to determine whether a user has a risk of churn. This solution relies on staff to visit and analyze the user's business status to determine whether a user has a risk of churn. There are strong subjective factors, and it is impossible to objectively and accurately analyze users, and it is difficult to ensure the accuracy of the classification and identification of IoT users with a risk of churn, and it is impossible to accurately determine whether IoT users have a risk of churn.
[0027] Another solution is for operators to analyze the card usage of IoT users through their own business analysis systems, such as the number of cards opened, the number of activated cards, the activity level, the number of account cancellations, the use of voice, SMS or traffic, billing fees, etc., and calculate the churn factor of IoT users based on a weighted algorithm. Based on the churn factor, they screen users who need to be retained in advance and establish a churn user warning list. This solution only analyzes the card usage of IoT users, and the analysis factors are single, which cannot accurately portray the real churn behavior trends of IoT users. At the same time, the churn factor calculated by a simple weighted algorithm cannot objectively and accurately classify and identify IoT users, and cannot accurately determine whether IoT users are at risk of churn.
[0028] In order to solve the problems of the prior art, the embodiments of the present application provide a method, device, equipment, storage medium and computer program product for classifying Internet of Things users. By obtaining a churn indicator feature set, that is, obtaining basic information feature data, card usage information feature data, business status information feature data and corporate negative public opinion information feature data of Internet of Things users, a comprehensive and multi-dimensional analysis of user information data is performed, objectively and accurately depicting the trend of user churn, which can help the model improve the accuracy of Internet of Things user classification. The churn indicator feature set is input into the trained target classification model to determine the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model. Based on the relative position relationship, the churn category of the Internet of Things user is determined, which can improve the accuracy of Internet of Things user classification and accurately determine whether the Internet of Things user is at risk of churn.
[0029] The following first introduces the Internet of Things user classification system provided by the embodiment of the present application.
[0030] Figure 1 FIG. 1 is a schematic diagram showing the structure of an Internet of Things user classification system 100 provided by an embodiment of the present application. Figure 1 As shown, the system may include a user analysis subsystem 110 and a user early warning subsystem 120 .
[0031] The user analysis subsystem 110 can be used to train a target classification model based on a historical IoT user churn indicator feature set and its corresponding churn categories; it can also be used to analyze IoT users' basic information, card usage information, business information, and negative public opinion information, and classify IoT users based on the feature data obtained by analyzing the above information.
[0032] The user warning subsystem 120 can be used to receive the user churn category output by the user analysis subsystem, and when the user churn category indicates that the user has a churn risk, output an alarm message, and transmit the alarm message to the visual interaction module so that the staff can view the alarm message. The alarm message may include all information data of the IoT user and the churn category, and the churn category may include the existence of a churn risk and the absence of a churn risk.
[0033] The user analysis subsystem 110 may include a user basic information analysis module 111 , a user card usage analysis module 112 , a user business analysis module 113 , a user negative public opinion analysis module 114 , a training module 115 and a determination module 116 .
[0034] Among them, the basic information analysis module 111, the user card usage analysis module 112, the user business analysis module 113 and the user negative public opinion analysis module 114 can be used to pre-process the user basic information, card usage information, business status information, and corporate negative public opinion information respectively to obtain user basic information feature data, card usage information feature data, business status information feature data, and corporate negative public opinion information feature data.
[0035] The training module 115 may be used to train a target classification model based on a historical IoT user churn indicator feature set and its corresponding churn category.
[0036] A target classification model can be set in the determination module 116, which can be used to classify IoT users based on the user basic information feature data, card usage information feature data, business status information feature data and corporate negative public opinion information feature data output by the basic information analysis module 111, the user card usage analysis module 112, the user business analysis module 113 and the user negative public opinion analysis module 114, so as to determine whether there is a risk of user loss.
[0037] The user warning subsystem 120 may include a warning module 121 and a visualization interaction module 122 .
[0038] The early warning module 121 may be used to receive the user churn category output by the user analysis subsystem, and when the user churn category indicates that the user is at risk of churn, output an alarm message and transmit the alarm message to the visualization interaction module.
[0039] The visualization interaction module 122 can be used to receive alarm information and perform visualization to facilitate staff to view information of IoT users who are at risk of loss.
[0040] Information can be transmitted between the user analysis subsystem and the user early warning subsystem via the system interaction interface 101 .
[0041] The following is an introduction to the IoT user classification method provided in the embodiment of the present application in conjunction with the accompanying drawings.
[0042] Figure 2 FIG. 1 is a flow chart of a method for classifying Internet of Things users provided by an embodiment of the present application. Figure 2 As shown, the method may include the following steps: S201 to S203.
[0043] S201: Obtain a feature set of churn indicators of IoT users.
[0044] In some embodiments, the churn indicator feature set is used to analyze IoT users from multiple dimensions, including basic information, card usage information, business conditions, and negative corporate public opinion, to characterize the churn behavior trends of IoT users. By inputting the churn indicator feature set into the target classification model of the embodiment of the present application, IoT users can be accurately classified and whether IoT users are at risk of churn can be accurately determined.
[0045] In some embodiments, the churn indicator feature set may include basic information feature data, card usage information feature data, business status information feature data, and corporate negative public opinion information feature data.
[0046] In some embodiments, the process of obtaining the churn indicator feature set of IoT users can be performed by Figure 1 The user analysis subsystem of the Internet of Things user classification system shown in the figure is used to complete the task.
[0047] S202: Input the churn indicator feature set into a pre-trained target classification model to determine the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model.
[0048] In some embodiments, the target classification model may be used to classify IoT users to determine the churn category of IoT users.
[0049] In some embodiments, the target classification model can be trained based on a support vector machine (SVM) model.
[0050] In some embodiments, the target classification model classifies IoT users through a classification hyperplane in the target classification model, and the classification hyperplane is used to determine the churn category corresponding to the churn indicator feature set through the relative position relationship between the churn indicator feature set and the classification hyperplane.
[0051] In some embodiments, the relative position relationship is used to represent the relative position of the churn indicator feature set and the classification hyperplane in the target classification model in the feature space.
[0052] In some embodiments, the relative position of the above-mentioned churn indicator feature set and the classification hyperplane in the target classification model in the feature space may include: the churn indicator feature set is located above the classification hyperplane in the feature space, and the churn indicator feature set is located below the classification hyperplane in the feature space.
[0053] The existing technology analyzes the card usage of IoT users and adopts a weighted algorithm to calculate the churn factor of IoT users, and judges whether the user is at risk of churn based on the churn factor. The analysis factors are single, and the churn factor obtained by weighted calculation cannot objectively and accurately describe the user's churn trend, cannot accurately classify IoT users, and cannot accurately judge whether IoT users are at risk of churn.
[0054] Therefore, this step trains the target classification model, inputs the churn indicator feature set into the trained target classification model, and determines the churn category corresponding to the churn indicator feature set through the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model. This can improve the accuracy of IoT user classification and accurately determine whether IoT users are at risk of churn.
[0055] S203: Determine the churn category of the IoT user according to the relative position relationship.
[0056] In some embodiments, the churn category is used to determine whether an IoT user has a churn risk, and the churn category of an IoT user may include a churn risk and a churn risk.
[0057] This step determines the churn category corresponding to the churn indicator feature set based on the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model, which can improve the accuracy of IoT user classification and accurately determine whether IoT users have a churn risk.
[0058] In some embodiments, determining the churn category of the IoT user according to the relative position relationship may include:
[0059] When the churn indicator feature set is located above the classification hyperplane in the feature space, the churn category of IoT users is that there is a churn risk;
[0060] When the churn indicator feature set is located below the classification hyperplane in the feature space, the churn category of IoT users is that there is no churn risk.
[0061] The embodiment of the present application obtains a churn indicator feature set, that is, obtains the basic information feature data, card usage information feature data, business status information feature data and corporate negative public opinion information feature data of IoT users, and conducts a comprehensive and multi-dimensional analysis of user information data, objectively and accurately depicts the user churn trend, and can help the model improve the accuracy of IoT user classification. The churn indicator feature set is input into the trained target classification model, and the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model is determined. Based on the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model, the churn category of IoT users is determined, which can improve the accuracy of IoT user classification and accurately determine whether IoT users are at risk of churn.
[0062] In some embodiments, in order to improve the accuracy of the target classification model in classifying IoT users, before step S202: inputting the churn indicator feature set into the pre-trained target classification model, the IoT user classification method may include:
[0063] Obtain a training sample set, the training sample set includes multiple training samples, each training sample includes a churn indicator feature set of historical IoT users and a corresponding churn category;
[0064] The support vector machine model is trained according to the training samples, and the target classification model is obtained when the preset training stop condition is met.
[0065] Among them, the support vector machine model is a supervised binary classification model, which distinguishes positive and negative samples through the optimal classification hyperplane to achieve target classification.
[0066] In one embodiment, the historical IoT users may include IoT users who have churned and IoT users who have not churned.
[0067] In one embodiment, training the target classification model may be accomplished through a machine automatic learning algorithm.
[0068] In one embodiment, the above-mentioned machine automatic learning algorithm may be a binary classification algorithm.
[0069] In one embodiment, obtaining a feature set of churn indicators of historical IoT users may be accomplished through feature engineering.
[0070] In one embodiment, the churn indicator feature set of historical IoT users can be represented by x, the churn category of historical IoT users can be represented by y, and the churn indicator feature set of the i-th IoT user can be represented by x i To represent, the churn category of the i-th IoT user can be represented by y i In this way, the above training sample set can be represented by X=((x 1 ,y 1 ),(x 2 ,y 2 )......,(x n ,y n )) to indicate.
[0071] In some embodiments, Figure 3 As shown, the above training of the support vector machine model according to the training samples, when the preset training stop condition is met, to obtain the target classification model, may include the following steps:
[0072] S301: Based on the loss category of the training sample, determine at least one classification hyperplane corresponding to the training sample through a preset correspondence relationship between the loss category and the classification hyperplane;
[0073] S302: Determine, from at least one classification hyperplane, a candidate classification hyperplane that satisfies a minimum geometric interval with the training sample;
[0074] S303: Determine a target classification hyperplane whose minimum geometric interval satisfies a preset condition from among candidate classification hyperplanes by using a sequence minimum optimization algorithm;
[0075] S304: Determine a target classification model according to the target classification hyperplane.
[0076] Among them, the sequential minimal optimization (SMO) algorithm is used to solve the quadratic optimization problem in the support vector machine model. Its basic idea is to optimize only two variables in one iteration and fix the remaining variables, decomposing the large optimization problem into multiple small optimization problems to obtain the optimal solution.
[0077] The minimum geometric interval is used to represent the minimum geometric distance between all training sample points and the classification plane in the feature space.
[0078] In one embodiment, the above classification hyperplane can be expressed by the following expression:
[0079] w * x+b * =0 (1)
[0080] Among them, w * Represents the normal vector of the classification hyperplane, which is used to determine the direction of the classification hyperplane; b * Represents the intercept of the classification hyperplane, which is used to represent the distance from the classification hyperplane to the origin in the feature space.
[0081] In one embodiment, the correspondence between the preset loss categories and the classification hyperplane can be obtained by the classification decision function sgn(w * x+b).
[0082] In one embodiment, the classification decision function sgn(w * x+b) can be expressed as follows:
[0083]
[0084] Among them, w * x i +b>0 means that the churn indicator feature set of the i-th IoT user is located above the classification hyperplane in the feature space, w * x i+b<0 means that the churn indicator feature set of the i-th IoT user is below the classification hyperplane in the feature space, y i =0 means that the churn category of the i-th IoT user is at risk of churn, y i =1 indicates that the churn category of the i-th IoT user is that there is no churn risk.
[0085] In one embodiment, the above-mentioned determining the target classification hyperplane whose minimum geometric interval satisfies the preset conditions from the candidate classification hyperplanes by using the sequential minimum optimization algorithm may include:
[0086] (1) Use the sequential minimum optimization algorithm to find the optimal solution α of the following expression * :
[0087]
[0088] Where α=(α 1 ,......,α n ), α i is the Lagrange multiplier.
[0089] (2) The optimal solution α * Substitute the following expression to get w * :
[0090]
[0091] (3) Find all the nodes that satisfy α i >0 corresponding to the training sample (x j ,y j ), calculate each training sample (x j ,y j ) corresponds to the following expression, we can get
[0092]
[0093] (4) Substitute w obtained in step (2) * and b obtained in step (3) * , substitute into the classification hyperplane formula w * x+b * =0, the target classification hyperplane can be determined and taken as the optimal classification hyperplane.
[0094] In some embodiments, in order to achieve a comprehensive characterization of the churn behavior trend of IoT users, so that the information data characterizing the churn behavior trend of IoT users can be applied to the target classification model classification, before step S201: obtaining the churn indicator feature set of IoT users, the IoT user classification method may further include:
[0095] Obtain information data of IoT users, including basic information data, card usage information data, business status information data and negative corporate public opinion information data;
[0096] The information data is preprocessed to obtain a loss indicator feature set; the preprocessing includes cleaning the information data.
[0097] Among them, cleaning is used to review and verify the data, with the aim of deleting duplicate information, correcting existing errors, and providing data consistency.
[0098] In the embodiments of the present application, there is no limitation on the source and method of obtaining the information data of IoT users.
[0099] In one embodiment, the basic information data, card usage information data and business status information data of the IoT user can be obtained from the operator's IoT business support system.
[0100] In one embodiment, the negative public opinion information data of the enterprise of the Internet of Things users is obtained from an Internet public opinion monitoring system.
[0101] In one embodiment, the basic information data of IoT users may include the industry to which they belong, user level, payment type, scale level and user card activation information. The user card activation information may include card activation time, card activation province, activation functions, tariff discounts, etc.
[0102] In one embodiment, the card usage information data of IoT users may include new users, net new users, off-network users, revenue contribution, activity, etc.
[0103] In one embodiment, the business status information data of IoT users may include tax credit, registration status (revoked, cancelled, continued) and risk information. The risk information may include administrative penalties, judicial risks, business anomalies, etc.
[0104] In one embodiment, the negative public opinion information data of the enterprise of the Internet of Things users may include enterprise personnel turnover, salary cuts and layoffs, business fluctuations, etc.
[0105] In some embodiments, the basic information feature data in step S201 may include application industry features, user level features, payment type features, scale level features, and user type features, as shown in Table 1 below.
[0106]
[0107] Among them, the application industry feature can be used to indicate the industry to which the user's IoT application belongs, such as the smart meter industry; the user level feature can be used to indicate the user's classification when joining the network, such as ordinary, silver, and gold; the payment type feature can be used to indicate the payment method selected by the user, such as prepaid and postpaid; the scale level feature can be used to indicate the scale of the user's IoT card business, such as millions or hundreds of thousands; the user type feature can be used to indicate the user type, such as state-owned enterprise employees, public institution employees, and private enterprise employees.
[0108] In some embodiments, the card usage information feature data in step S201 may include card activation features and card usage features;
[0109] User card opening characteristics may include card opening time distribution characteristics, most recent card opening time characteristics, subscription rate characteristics, rate discount characteristics, and activation function characteristics;
[0110] Card usage characteristics may include new user number characteristics, net increase user number characteristics, off-network user number characteristics, call bill usage characteristics, revenue contribution characteristics, active user number characteristics and card usage risk characteristics;
[0111] Card usage risk characteristics may include usage ratio characteristics in key areas, usage ratio characteristics in sensitive areas, usage ratio characteristics in mobile terminals, card-machine separation ratio characteristics, one card for multiple cards ratio characteristics, one card for multiple locations ratio characteristics, and visitor network ratio characteristics. The specific characteristics are shown in Table 2.
[0112]
[0113]
[0114] Among them, the card activation time distribution feature can be used to indicate the time distribution of users' newly opened Internet of Things cards; the most recent card activation time feature can be used to indicate the time when the user last opened a new Internet of Things card; the subscription rate feature can be used to indicate the Internet of Things card rate package subscription information; the rate discount feature can be used to indicate the Internet of Things card rate discount information; and the activation function feature can be used to indicate the activation status of the Internet of Things card function.
[0115] Among them, the new user number feature can be used to indicate the number of new IoT cards added by users; the net increase in users feature can be used to indicate the net increase in the number of IoT cards added by users; the off-network user number feature can be used to indicate the number of IoT cards that users have left the network / canceled; the call bill usage feature can be used to indicate the user's voice, SMS, and traffic call bill usage; the revenue contribution feature can be used to indicate the user's monthly bill contribution; the active user number feature can be used to indicate the number of active IoT cards under the user; the card usage risk feature can be used to indicate the risk feature of the user's IoT card, which may include the usage proportion feature in key areas, the usage proportion feature in sensitive areas, the usage proportion feature in mobile terminals, the machine-card separation proportion feature, the one-card-multiple-card proportion feature, the one-card-multiple-location proportion feature, and the visiting human network proportion feature.
[0116] Among them, the key area usage ratio feature can be used to indicate the number of users using the user's Internet of Things card in the key areas defined by the Ministry of Industry and Information Technology / the total number of users of the user's Internet of Things card; the sensitive area usage ratio feature can be used to indicate the number of users using the user's Internet of Things card in the sensitive areas defined by the Ministry of Industry and Information Technology / the total number of users of the user's Internet of Things card; the mobile terminal usage ratio feature can be used to indicate the number of users using the user's Internet of Things card on mobile terminals / the total number of users of the user's Internet of Things card; the machine-card separation ratio feature can be used to indicate the number of users whose Internet of Things cards have machine-card separation / the total number of users of the user's Internet of Things card; the one-certificate-multiple-card ratio feature can be used to indicate the number of users with one-certificate-multiple-cards for the user's Internet of Things card / the total number of users of the user's Internet of Things card; the one-certificate-multiple-location usage ratio feature can be used to indicate the number of users whose Internet of Things card is used in multiple locations with one certificate / the total number of users of the user's Internet of Things card; the human network access ratio feature can be used to indicate the number of users whose Internet of Things card has human network access / the total number of users of the user's Internet of Things card.
[0117] In some embodiments, the business status information feature data in step S201 may include unit credit features, registration status features, administrative penalty features, business abnormality features, industrial and commercial change features, and judicial risk features, as shown in Table 3 below.
[0118]
[0119] Among them, the unit credit feature can be used to determine whether the unit credit is normal based on the tax credit level and whether it is included in the industry user blacklist database; the registration status feature can be used to indicate whether the registration status is normal, and the registration status includes revocation, cancellation, and continuation; the administrative penalty feature can be used to indicate whether it has been subject to administrative penalties for fraud, illegal operation, environmental violations, labor violations, illegal finance and other behaviors; the abnormal operation feature can be used to determine whether the operation is abnormal based on the average number of equity freezes, equity pledges, abnormal operations, land pledges, tax arrears announcements, movable mortgages, intellectual property pledges, and external guarantees within a preset period; the industrial and commercial change feature can be used to determine whether the change is normal based on whether the registered capital remains unchanged or changes to increase or decrease within a preset period, the average number of legal person changes, the average number of shareholder changes, and the average number of equity changes; the judicial risk feature can be used to determine whether there is a judicial risk based on the average number of administrative penalties, serious violations of law and trust, executions, and executions of breaches of trust within a preset period.
[0120] In some embodiments, the characteristic data of negative corporate public opinion information in step S201 may include corporate personnel turnover characteristics, salary reduction and layoff characteristics, and business fluctuation characteristics, as shown in Table 4 below.
[0121]
[0122] Among them, the enterprise personnel flow characteristics can be used to indicate whether the enterprise personnel flow is abnormal; the salary reduction and layoff characteristics can be used to indicate whether there is a large-scale salary reduction or layoff public opinion in the enterprise; the business fluctuation characteristics can be used to indicate the fluctuations in the enterprise's market share, sales orders, and financial revenue.
[0123] By analyzing IoT users from multiple dimensions including basic information, card usage information, business conditions, and negative corporate public opinion, we comprehensively consider various factors that may cause user churn. Comprehensive evaluation of multiple factors of IoT users can objectively and accurately characterize the churn behavior trends of IoT users. Based on the churn behavior trends of IoT users, we can improve the accuracy of the target classification model in classifying IoT users and accurately determine whether IoT users are at risk of churn.
[0124] In some embodiments, after step S203: determining the churn category of the IoT user, the IoT user classification method may further include:
[0125] When the IoT user churn category is at risk of churn, an alarm message is output.
[0126] By timely outputting alarm information when the IoT user churn category is at risk of churn, it is possible to monitor IoT users in real time and promptly discover IoT users at risk of churn, making it easier to formulate targeted measures in advance to retain users.
[0127] In one embodiment, the above-mentioned warning information may include information data and loss category of the Internet of Things user.
[0128] In one embodiment, the process of outputting the alarm information can be performed by Figure 1 This is completed through the early warning module in the IoT user classification system.
[0129] In some embodiments, after step S203: determining the churn category of the IoT user, the IoT user classification method may further include:
[0130] When the IoT user loss category is at risk of loss, store the IoT user's information data;
[0131] Displays the stored information data of IoT users.
[0132] By storing and displaying the information data of IoT users who are at risk of churn, an IoT user churn warning list can be established for staff to review. Staff can contact users through the list and take user retention measures in advance to improve user retention rate.
[0133] In one embodiment, the process of storing the information data of the Internet of Things user and displaying the stored information data of the Internet of Things user can be performed by Figure 1 This is done through the visual interaction module in the IoT user classification system.
[0134] Figure 4 4 is a schematic diagram of the structure of an IoT user classification device 400 provided in an embodiment of the present application. Figure 4 As shown, the apparatus may include an acquisition module 410 , a determination module 420 , and a classification module 430 .
[0135] The acquisition module 410 is used to obtain a churn indicator feature set of IoT users; the churn indicator feature set includes basic information feature data, card usage information feature data, business status information feature data, and corporate negative public opinion information feature data;
[0136] Basic information feature data includes application industry features, user level features, payment type features, scale level features, and user type features;
[0137] The card usage information feature data includes card opening features and card usage features;
[0138] User card opening characteristics include card opening time distribution characteristics, latest card opening time characteristics, subscription rate characteristics, rate discount characteristics and activation function characteristics;
[0139] Card usage characteristics include new user number characteristics, net increase user number characteristics, off-network user number characteristics, call bill usage characteristics, revenue contribution characteristics, active user number characteristics and card usage risk characteristics;
[0140] The card usage risk characteristics include the usage ratio characteristics in key areas, the usage ratio characteristics in sensitive areas, the usage ratio characteristics in mobile terminals, the usage ratio characteristics of the card and the machine, the usage ratio characteristics of multiple cards with one ID, the usage ratio characteristics of multiple locations with one ID, and the usage ratio characteristics of the visitor network;
[0141] The characteristic data of business status information include unit credit characteristics, registration status characteristics, administrative penalty characteristics, business abnormality characteristics, industrial and commercial change characteristics and judicial risk characteristics;
[0142] The characteristic data of negative public opinion information of enterprises include the characteristics of enterprise personnel turnover, salary reduction and layoffs, and business fluctuations;
[0143] Determination module 420, used for inputting the churn indicator feature set into a pre-trained target classification model to determine the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model; the target classification model is obtained by training based on a support vector machine model;
[0144] The classification module 430 is used to determine the churn category of the Internet of Things users according to the relative position relationship.
[0145] This embodiment obtains the churn indicator feature set, that is, obtains the basic information feature data, card usage information feature data, business status information feature data and corporate negative public opinion information feature data of IoT users, and conducts a comprehensive and multi-dimensional analysis of user information data, objectively and accurately depicts the user churn trend, which can help the model improve the accuracy of IoT user classification. The churn indicator feature set is input into the trained target classification model, and the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model is determined. Based on the relative position relationship, the churn category of IoT users is determined, which can improve the accuracy of IoT user classification and accurately determine whether IoT users are at risk of churn.
[0146] In some embodiments, the IoT user classification device 400 further includes a training module;
[0147] The acquisition module 410 is further used to acquire a training sample set before inputting the churn indicator feature set into the pre-trained target classification model, where the training sample set includes a plurality of training samples, each training sample includes a churn indicator feature set of historical IoT users and a corresponding churn category;
[0148] The training module is used to train the support vector machine model according to the training samples, and obtain the target classification model when the preset training stop condition is met.
[0149] In some embodiments, the training module is further used to determine at least one classification hyperplane corresponding to the training sample based on the loss category of the training sample and through the correspondence between the preset loss category and the classification hyperplane;
[0150] Determine a candidate classification hyperplane from at least one classification hyperplane that satisfies a minimum geometric interval with the training sample;
[0151] Through the sequential minimum optimization algorithm, the target classification hyperplane with the minimum geometric interval that meets the preset conditions is determined from the candidate classification hyperplanes;
[0152] According to the target classification hyperplane, the target classification model is determined.
[0153] In some embodiments, the IoT user classification device 400 further includes a processing module;
[0154] The acquisition module 410 is further used to acquire information data of IoT users before acquiring the churn indicator feature set of IoT users, the information data including basic information data, card usage information data, business status information data and negative corporate public opinion information data;
[0155] The processing module is used to pre-process the information data to obtain a loss indicator feature set; the pre-processing includes cleaning the information data.
[0156] In some embodiments, the IoT user classification device 400 further includes an alarm module;
[0157] The alarm module is used to output alarm information after determining the churn category of the IoT user and in the case that the churn category of the IoT user is a churn risk.
[0158] In some embodiments, the IoT user classification device 400 further includes a display module;
[0159] The display module is used to store the information data of the Internet of Things user after determining the loss category of the Internet of Things user and in the case that the loss category of the Internet of Things user is a loss risk; and display the stored information data of the Internet of Things user.
[0160] Figure 4 The various modules in the device shown can be implemented Figure 2 Each step in the above steps can achieve the corresponding technical effect, which will not be repeated here for the sake of brevity.
[0161] Figure 5 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application is shown.
[0162] The electronic device may include a processor 501 and a memory 502 storing computer program instructions.
[0163] Specifically, the processor 501 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.
[0164] The memory 502 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 502 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 502 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 502 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 502 is a non-volatile solid-state memory.
[0165] In certain embodiments, the memory may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Thus, typically, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to one aspect of the present application.
[0166] The processor 501 implements any one of the IoT user classification methods in the above embodiments by reading and executing computer program instructions stored in the memory 502 .
[0167] In one example, the electronic device may further include a communication interface 503 and a bus 510. Figure 5 As shown, the processor 501, the memory 502, and the communication interface 503 are connected via a bus 510 and communicate with each other.
[0168] The communication interface 503 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.
[0169] Bus 510 includes hardware, software or both, and the parts of electronic equipment are coupled to each other. For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 510 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the application considers any suitable bus or interconnection.
[0170] In addition, in combination with the IoT user classification method in the above embodiment, the present application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, any IoT user classification method in the above embodiment is implemented.
[0171] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed, implements any one of the Internet of Things user classification methods in the above embodiments.
[0172] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.
[0173] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.
[0174] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.
[0175] The above reference is to the flowchart and / or block diagram of the method, device, equipment, storage medium and computer program product according to the embodiment of the present application. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.
[0176] The above is only a specific implementation of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.
Claims
1. A method for classifying Internet of Things users, characterized in that: include: Obtain a churn indicator feature set of IoT users; the churn indicator feature set includes basic information feature data, card usage information feature data, business status information feature data, and corporate negative public opinion information feature data; Inputting the churn indicator feature set into a pre-trained target classification model to determine the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model; The target classification model is obtained based on support vector machine model training; According to the relative position relationship, a churn category of the Internet of Things user is determined.
2. The method according to claim 1, characterized in that Before inputting the churn indicator feature set into the pre-trained target classification model, it includes: Acquire a training sample set, wherein the training sample set includes a plurality of training samples, each of the training samples includes a churn indicator feature set of historical IoT users and a corresponding churn category; The support vector machine model is trained according to the training samples, and the target classification model is obtained when a preset training stop condition is met.
3. The method according to claim 2, characterized in that The step of training the support vector machine model according to the training samples to obtain the target classification model when a preset training stop condition is met includes: Based on the loss category of the training sample, determining at least one classification hyperplane corresponding to the training sample through a preset correspondence between the loss category and the classification hyperplane; Determine, from the at least one classification hyperplane, a candidate classification hyperplane that satisfies a minimum geometric interval with the training sample; Determine, by a sequential minimum optimization algorithm, a target classification hyperplane whose minimum geometric interval satisfies a preset condition from among the candidate classification hyperplanes; The target classification model is determined according to the target classification hyperplane.
4. The method according to any one of claim 1, characterized in that: Before obtaining the churn indicator feature set of the IoT user, the method further includes: Obtaining information data of IoT users, including basic information data, card usage information data, business status information data, and negative corporate public opinion information data; The information data is preprocessed to obtain the loss indicator feature set; the preprocessing includes cleaning the information data.
5. The method according to any one of claims 1 to 4, characterized in that: The basic information feature data includes application industry characteristics, user level characteristics, payment type characteristics, scale level characteristics and user type characteristics.
6. The method according to any one of claims 1 to 4, characterized in that: The card usage information feature data includes card opening features and card usage features; The user card activation characteristics include card activation time distribution characteristics, the most recent card activation time characteristics, subscription fee characteristics, fee discount characteristics and activation function characteristics; The card usage characteristics include the number of new users, the number of net new users, the number of off-network users, the number of call bill usage, the revenue contribution, the number of active users and the card usage risk characteristics; The card usage risk characteristics include the usage proportion characteristics in key areas, the usage proportion characteristics in sensitive areas, the usage proportion characteristics in mobile terminals, the usage proportion characteristics of machine and card, the usage proportion characteristics of one card for multiple cards, the usage proportion characteristics of one card for multiple locations and the usage proportion characteristics of visitors connected to the Internet.
7. The method according to any one of claims 1 to 4, characterized in that: The business status information characteristic data includes unit credit characteristics, registration status characteristics, administrative penalty characteristics, business abnormality characteristics, industrial and commercial change characteristics and judicial risk characteristics.
8. The method according to any one of claims 1 to 4, characterized in that: The characteristic data of negative public opinion information of an enterprise include characteristics of enterprise personnel turnover, salary reduction and layoffs, and business fluctuations.
9. The method according to any one of claims 1 to 4, characterized in that: The churn category of the Internet of Things user includes a churn risk and a churn risk. After determining the churn category of the Internet of Things user, the method further includes: When the churn category of the IoT user is a churn risk, an alarm message is output.
10. The method according to any one of claims 1 to 4, characterized in that: After determining the churn category of the IoT user, the method further includes: When the loss category of the IoT user is a loss risk, storing the information data of the IoT user; Displays the stored information data of IoT users.
11. An Internet of Things user classification device, characterized in that: The device comprises: An acquisition module is used to acquire a churn indicator feature set of IoT users; the churn indicator feature set includes basic information feature data, card usage information feature data, business status information feature data, and corporate negative public opinion information feature data; A determination module, used to input the churn indicator feature set into a pre-trained target classification model to determine the relative position relationship between the churn indicator feature set and the classification hyperplane in the target classification model; the target classification model is obtained by training based on a support vector machine model; A classification module is used to determine the churn category of the Internet of Things user according to the relative position relationship.
12. An electronic device, characterized in that: The device comprises: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the method for classifying Internet of Things users as described in any one of claims 1 to 10 is implemented.
13. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by the processor, the method for classifying Internet of Things users according to any one of claims 1 to 10 is implemented.
14. A computer program product, characterized in that When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the method for classifying Internet of Things users as described in any one of claims 1 to 10.
