User type prediction method

Abstract

The embodiment of the invention provides a user type prediction method. The method comprises the following steps: according to data generated by a network support domain system, obtaining service behavior perception characteristics and call behavior characteristics of non-off-network users; and according to the service behavior perception characteristics and the call behavior characteristics, determining service behavior portraits and voice service portraits corresponding to the non-off-network users through a user type prediction model, and predicting potential carry-out type users from the non-off-network users according to the service behavior portraits and the voice service portraits. According to the embodiment of the invention, the problem that the prediction of the carrying-out user is inaccurate according to the basic information of the user and the data of whether the user surfs the internet and the like in the related technology can be solved, so that the effect of improving the prediction accuracy of the potential carrying-out user is achieved.

Description

Technical Field

[0001] This invention relates to the field of mobile wireless communication, and more specifically, to a user type prediction method. Background Technology

[0002] With the development of communication services, users' demands for communication products are constantly evolving. To improve the quality of communication services, user needs are typically analyzed using basic user information and data such as whether users are online. Based on this analysis, predictions of users switching out can be made. However, the analysis results obtained through this method are often inaccurate, leading to inaccurate predictions of users switching out, which is detrimental to improving the quality of communication services. Furthermore, most related technologies focus on predicting users switching out, lacking technical solutions for predicting users switching to other services. Summary of the Invention

[0003] This invention provides a user type prediction method to at least solve the problem of inaccurate prediction of porting users based on user basic information and data such as whether the user is online in related technologies.

[0004] According to an embodiment of the present invention, a user type prediction method is provided, comprising: acquiring service behavior perception characteristics and call behavior characteristics of users who have not left the network based on data generated by a network support domain system; determining a service behavior profile and a voice service profile corresponding to each user who has not left the network based on the service behavior perception characteristics and call behavior characteristics through a user type prediction model; and predicting potential porting-out users from the users who have not left the network based on the service behavior profile and the voice service profile.

[0005] According to yet another embodiment of the present invention, a computer-readable storage medium is also provided, wherein a computer program is stored therein, wherein the computer program is configured to perform the steps in the above method embodiments when executed.

[0006] According to yet another embodiment of the present invention, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in the above method embodiments.

[0007] According to yet another embodiment of the present invention, a computer program product is also provided, comprising a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0008] Through the above embodiments of the present invention, based on the data generated by the network support domain system, business behavior perception features and call behavior features that better characterize user behavior are obtained. Thus, based on the business behavior perception features and call behavior features, potential porting-out users can be predicted more accurately. Therefore, the problem of inaccurate prediction of porting-out users based on user basic information and whether the user is online in related technologies can be solved, thereby improving the accuracy of prediction of potential porting-out users. Attached Figure Description

[0009] Figure 1 This is a hardware structure block diagram of a computer terminal that runs a user type prediction method according to an embodiment of the present invention;

[0010] Figure 2 This is a network architecture diagram of a potential user prediction system according to an embodiment of the present invention;

[0011] Figure 3 This is a flowchart of a user type prediction method according to an embodiment of the present invention;

[0012] Figure 4 This is a schematic diagram illustrating the training of a user type prediction model according to an embodiment of the present invention;

[0013] Figure 5 This is a flowchart of an AI model training method according to an embodiment of the present invention;

[0014] Figure 6 This is a schematic diagram of the AI ​​model training process according to an embodiment of the present invention. Detailed Implementation

[0015] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings and examples.

[0016] It should be noted that the terms "first," "second," etc., in the specification, claims, and drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0017] The methods and embodiments provided in this invention can be executed on a mobile terminal, a computer terminal, or a similar computing device. Taking execution on a computer terminal as an example... Figure 1 A hardware structure block diagram of a computer terminal running a user type prediction method according to an embodiment of the present invention. (See diagram below.) Figure 1 As shown, a computer terminal may include one or more ( Figure 1Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The computer terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the computer terminal described above. For example, the computer terminal may also include components that are more complex than those described above. Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0018] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the user type prediction method in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the computer programs stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to a computer terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0019] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module used for wireless communication with the Internet.

[0020] Figure 2 This is a network architecture diagram of a potential user prediction system according to an embodiment of the present invention, such as... Figure 2 As shown, the network architecture includes: a data source layer, a VMAX (Versatile Modular Architecture Storage) data preprocessing layer, a VMAX storage layer, an AI big data model functional layer, and a VMAX application layer.

[0021] Data source layer: includes O domain (Operation Support Domain) data, i.e. data generated by the network support domain system, and B domain (Business Support Domain) data, i.e. data generated by the business support domain system;

[0022] In this embodiment, the data used for predicting potential outgoing users includes not only the O-domain data and B-domain data contained in the data source layer, but also survey report data.

[0023] VMAX data preprocessing layer: used to obtain B-domain data from the data source through the interface and collect O-domain data from the data source layer through the probe, and to perform data cleaning on the B-domain data and O-domain data;

[0024] VMAX storage layer: includes the distributed file system (Hadoop Distributed File System, HDFS) and the database management system GBASE, used to store data preprocessed by the VMAX data preprocessing layer;

[0025] AI Big Data Model Functional Layer: Used to obtain relevant user data from the VMAX storage layer, such as mobile internet access log data and voice service data, and based on the relevant user data, use the AI ​​Big Data Model to predict user types using NPS (Net Promoter Score) to predict potential porting types of users.

[0026] VMAX Application Layer: Displays the prediction results fed back from the AI ​​big data model functional layer through the VMAX application interface.

[0027] This embodiment provides a user type prediction method that runs on the aforementioned computer terminal or potential user prediction system. This method obtains more refined user data from the O-domain system and the B-domain system. Therefore, the terminal / device running this method needs to be connected to the O-domain system and the B-domain system via a network. At the same time, this embodiment has a large demand for O-domain data and a high requirement for network transmission rate. Therefore, the server should support a high-performance central processing unit (CPU), large memory, and massive storage space.

[0028] Figure 3 This is a flowchart of a user type prediction method according to an embodiment of the present invention, such as... Figure 3 As shown, the process includes the following steps:

[0029] Step S302: Based on the data generated by the network support domain system, obtain the service behavior perception characteristics and call behavior characteristics of users who have not left the network;

[0030] In step S302 of this embodiment, the service behavior perception characteristics of the non-disconnected user are obtained based on the data generated by the network support domain system, including: obtaining the user face call detail records of the non-disconnected user from the data generated by the network support domain system according to a first preset extraction period; obtaining the mobile internet access log data of the non-disconnected user from the user face call detail records, and performing feature extraction on the mobile internet access log data to obtain the service behavior perception characteristics of the non-disconnected user.

[0031] In this embodiment, the data generated by the network support domain system includes user face call detail records (CDRs). Obtaining the CDRs of the non-disconnected users from the data generated by the network support domain system according to a first preset extraction cycle includes: collecting the CDRs of the non-disconnected users from the O domain via DPI according to a one-day extraction cycle.

[0032] In this embodiment, due to the large variety and wide scope of various services in current practical applications, the characteristics of each service are different, and the data extraction cost and output ratio vary. Therefore, this embodiment only performs service feature cleaning on popular services, such as traffic, duration, video, web browsing, games, and instant messaging. Non-popular services undergo general feature cleaning. Specifically, user service behavior perception features can be extracted at hourly or daily granularity. Based on user service behavior, daily granularity aggregation is more helpful for user type prediction. For example, a time window of 24 hours can be set, meaning the first preset extraction period is one day.

[0033] In one embodiment, the business behavior perception features include multiple indicators. The AI ​​big data model can comprehensively perceive the user's business behavior through these multiple indicators and the user's online information. For example, it can perceive whether the user is a heavy video user, a heavy game user, whether they have visited websites of other network operators, and the duration of their visits.

[0034] Specifically, the indicators include:

[0035] The behavioral perception characteristics of video-related services include, but are not limited to, one of the following indicators: video buffering wait time, video stuttering time, video clarity, etc.

[0036] The behavioral perception characteristics of network-related businesses include, but are not limited to, one of the following indicators: traffic, duration, webpage response success rate, webpage display success rate, webpage download speed, etc.

[0037] Game-related business behavior perception characteristics include, but are not limited to, one of the following indicators: traffic, duration, TCP (Transmission Control Protocol) downlink RTT (Round-Trip Time) latency, game large packet download speed, game uplink and downlink UDP (User Datagram Protocol) jitter latency, etc.

[0038] Instant messaging business behavior perception characteristics include, but are not limited to, one of the following indicators: traffic, duration, IM (Instant Messaging) login success rate, average IM login duration, IM text sending success rate, IM text receiving success rate, IM image sending success rate, IM image receiving success rate, IM audio sending success rate, IM audio receiving success rate, IM video sending success rate, and IM video receiving success rate.

[0039] General business behavior perception characteristics include, but are not limited to, one of the following metrics: traffic, duration, TCP uplink and downlink RTT latency, download speed, etc.

[0040] In step S302 of this embodiment, the call behavior characteristics of the non-disconnected user are obtained based on the data generated by the network support domain system, including: obtaining the voice call detail records of the non-disconnected user from the data generated by the network support domain system according to a second preset extraction period; obtaining voice service data from the voice call detail records; and performing feature extraction on the voice service data of the non-disconnected user to obtain the call behavior characteristics of the non-disconnected user.

[0041] In this embodiment, the data generated by the O domain system includes voice call detail records (CDRs). According to a second preset extraction cycle, the voice call detail records of the non-disconnected user are obtained from the data generated by the network support domain system, including: collecting the voice call detail records of the non-disconnected user from the O domain system through DPI according to a one-day extraction cycle.

[0042] In one embodiment, call behavior characteristics include, but are not limited to, at least one of the following indicators: call type, peer number, cell, time period, whether it is the customer service number of this operator, number of calls, affiliated operator, and call duration.

[0043] In this method, embodiments of the present invention also employ business profiling technology, namely, training a model based on mobile communication O-domain and B-domain data to obtain user profiles and determine user types. Specifically, after business cleaning of O-domain data, business characteristics are extracted, and key user information is extracted from B-domain data. The two domain data are correlated through users to generate basic user characteristics. Then, based on known outgoing user business data and questionnaire information, an AI binary classification prediction technology is used to train a model to predict potential outgoing user types.

[0044] Step S304: Based on the business behavior perception features and call behavior features, determine the business behavior profile and voice service profile corresponding to each non-disconnected user through the user type prediction model, and predict potential porting-out users from the non-disconnected users based on the business behavior profile and voice service profile.

[0045] By adding voice service profiles and business behavior profiles, the network needs of users can be determined more accurately. The voice service profile includes, but is not limited to, the following information: whether the user's call recipient is a customer of the operator, whether the user on the other end of the call is a user of another network and the proportion of users of other networks, call frequency, etc. The business behavior profile includes, but is not limited to, the following information: user's APP application preferences, such as users who prefer video / games.

[0046] Before step S304 in this embodiment, the method further includes: obtaining the personal key information characteristics of the non-disconnected user based on the data generated by the service support domain system; and obtaining the network performance index characteristics and network coverage quality characteristics of the non-disconnected user based on the data generated by the network support domain system.

[0047] In one embodiment, obtaining the personal key information feature data of the non-disconnected user based on the data generated by the business support domain system includes: obtaining the personal information data of the non-disconnected user from the data generated by the business support domain system according to a second preset extraction period; and performing feature extraction on the personal information data of the non-disconnected user to obtain the personal key information feature of the non-disconnected user.

[0048] In this embodiment, the data generated by the business support domain system includes personal (critical) information data. According to a second preset extraction cycle, the personal information data of the non-disconnected user is obtained from the data generated by the business support domain system. This includes collecting the personal information data of the non-disconnected user from domain B according to a one-month extraction cycle. The personal information data includes, but is not limited to, one of the following categories: gender, age, address, communication billing package, package fee, user payment mode, package type, network duration, marital status, job category, and whether it is a broadband user. The above-mentioned personal information data are attributes closely related to disconnection.

[0049] In one embodiment, obtaining the network performance index characteristics and network coverage quality characteristics of the non-disconnected user based on the data generated by the network support domain system includes: obtaining the control plane signaling data based on the control plane call detail records (CDRs) in the data generated by the network support domain system; and extracting features from the control plane signaling data to obtain the network performance index characteristics; obtaining the coverage service data based on the coverage-related CDRs in the data generated by the network support domain system; and extracting features from the coverage service data to obtain the network coverage quality information characteristics.

[0050] In this embodiment, the data generated in the network support domain system also includes control plane call detail records (CDRs). Obtaining control plane CDRs includes: collecting control plane CDRs of non-disconnected users from the O domain via DPI (Data Point Indicator) according to a daily extraction cycle; and then obtaining network performance indicator characteristics of non-disconnected users based on the collected control plane CDRs. These network performance indicators include: attach / registration success rate, handover success rate, frequent handover status, and service request success rate.

[0051] In this embodiment, the data generated in the network support domain system also includes coverage-related call detail records (CDRs). Obtaining coverage-related CDRs includes: collecting coverage-related CDRs of users who have not left the network from the O domain via DPI according to a daily extraction cycle. Then, network coverage quality information features are obtained from the collected coverage-related CDRs. This network coverage quality information includes, but is not limited to, at least one of the following categories: base station identifier, cell identifier, total number of valid Measurement Report (MR) sampling points for Reference Signal Received Power (RSRP), whether it is a weak coverage cell, whether it is an overlapping coverage cell, whether it is an over-coverage cell, RSRP sum, RSRP mean, total number of valid MR sampling points for Uplink Signal to Interference plus Noise Ratio (ULSINR), ULSINR sum, ULSINR mean, whether it is a cell with poor uplink SINR, whether it is a cell with modulo-3 interference, whether it is a cell with weak uplink coverage, whether it is a cell with insufficient deep coverage, whether it is an over-coverage cell, and the mean of Reference Signal Received Quality (RSRQ), etc.

[0052] In one embodiment, predicting potential porting-out users from the non-disconnected users based on the business behavior profile and voice service profile includes: predicting potential porting-out users from the non-disconnected users using the user type prediction model based on the business behavior profile and the voice service profile, combined with the personal key information features, the network performance indicator features, and the network coverage quality features.

[0053] By extracting user profile features from multiple aspects, such as business behavior perception features, call behavior features, personal key information features, network performance index features, and network coverage quality features, this technology overcomes the barrier of predicting outgoing users based solely on basic user data and business data, thereby improving the accuracy of predicting potential outgoing users.

[0054] In step S304 of this embodiment, the following steps are included: based on the business behavior profile and voice service profile, and combined with the questionnaire information of users who have left the network and the predefined leaving network tags, the user type prediction model is trained using artificial intelligence binary classification technology, so as to predict users with potential leaving types through the trained user type prediction model.

[0055] In one embodiment, a survey report on users who have left the network is compiled and output. The survey is based on a summary of the reasons for leaving the network, such as: basic personal information of users, current network quality satisfaction, whether they have left the network before, whether they currently have the intention to leave the network, and the main reasons for leaving the network.

[0056] Figure 4 This is a schematic diagram illustrating the training of a user type prediction model according to an embodiment of the present invention. Figure 4 As shown, using the month as the unit, based on the data of users who have already migrated out of the O and B domain systems, and combined with survey questionnaire information and predefined churn labels, an AI migration model is trained and migration prediction is performed. Specifically, the user type prediction model is trained after data preprocessing and feature engineering using AI binary classification technology, so as to predict users with potential migration types through the trained user type prediction model.

[0057] Following step S304 in this embodiment, the method further includes: predicting whether a user of the potential porting type will port to another network based on the user's mobile internet access log data and voice service data, wherein the user's mobile internet access log data includes data on accessing websites of other network operators; and the user's voice service data includes operator customer service number data.

[0058] In this embodiment, a user type prediction model is used to assess user porting in the future based on the mobile internet access log data and voice service data of users with potential porting types, predict whether users with potential porting types will port, and output the user assessment results.

[0059] In this embodiment of the invention, by adding a user potential porting prediction survey report, adding user voice service profiles (e.g., whether the user's call recipient is a carrier customer, whether the called user is a user on another network, the proportion of users on other networks, and the frequency of calls), and adding user service behavior profiles (e.g., user APP application preferences, such as video / game preferences, and user network needs), the user's service characteristics can be extracted more accurately. At the same time, by using AI technology for data preprocessing and feature engineering, this embodiment of the invention can be executed more completely and efficiently, improving the accuracy of user type prediction.

[0060] This invention also provides an AI model training method. Figure 5 This is a flowchart of an AI model training method according to an embodiment of the present invention, such as... Figure 5 As shown, the process includes four steps: data preprocessing, data feature extraction, model training and inference, and outputting prediction results.

[0061] like Figure 6 As shown, Figure 6 This is a schematic diagram of the AI ​​model training process according to an embodiment of the present invention. O-domain data, B-domain data, historical data, and survey reports are input into VMAX for data cleaning and business feature extraction. Then, AI data preprocessing, AI data feature extraction, AI model training and inference are performed sequentially, and a network migration prediction list (prediction results) is output.

[0062] The overall business process of the user type prediction method includes the data cleaning stage and model training and inference.

[0063] Data cleaning phase: extraction of key personal information of users, cleaning of key information of user control face call details, cleaning of user mobile Internet access log information, cleaning of user network coverage data, cleaning of user call information, and data standardization of user survey questionnaire reports.

[0064] Model training and inference: Preprocessing business data, extracting business data features through engineering, training the user potential carryover prediction model, and evaluating the user potential carryover inference.

[0065] The data input to VMAX data cleaning and feature extraction requires complete basic user information, such as gender, age, address, job type, communication package, payment cycle, and payment mode. Furthermore, the Deep Packet Inspection (DPI) data collection is complete, including both software-collected wireless data and hardware-collected core network data. This comprehensive data collection facilitates accurate predictions of potential user outflows. The entire prediction process places high demands on servers. Because DPI data is massive, both data preprocessing and model training are computationally expensive, requiring servers with high CPU capacity, large memory, and ample storage space.

[0066] Through the above steps, based on the data generated by the network support domain system, business behavior perception features and call behavior features that better characterize user behavior are obtained. Thus, based on the business behavior perception features and call behavior features, potential porting-out users can be predicted more accurately. Therefore, this can solve the problem of inaccurate prediction of porting-out users based on user basic information and whether the user is online in related technologies, thereby improving the accuracy of prediction of potential porting-out users.

[0067] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0068] It should be noted that the above modules can be implemented by software or hardware. For the latter, they can be implemented in the following ways, but are not limited to: all the above modules are located in the same processor; or, the above modules are located in different processors in any combination.

[0069] Embodiments of the present invention also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to perform the steps in any of the above method embodiments when executed.

[0070] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0071] Embodiments of the present invention also provide an electronic device including a memory and a processor, the memory storing a computer program and the processor being configured to run the computer program to perform the steps in any of the above method embodiments.

[0072] In one exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor and the input / output device is connected to the processor.

[0073] According to yet another embodiment of the present invention, a computer program product is also provided, comprising a computer program that, when executed by a processor, implements the steps of the methods described in various embodiments of the present invention.

[0074] Specific examples in this embodiment can be found in the examples described in the above embodiments and exemplary implementations, and will not be repeated here.

[0075] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0076] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A user type prediction method, characterized in that, include: Based on the data generated by the network support domain system, obtain the business behavior perception characteristics and call behavior characteristics of users who have not left the network; Based on the business behavior perception characteristics and call behavior characteristics, the business behavior profile and voice service profile corresponding to each non-disconnected user are determined by the user type prediction model, and users with potential porting-out types are predicted from the non-disconnected users based on the business behavior profile and voice service profile.

2. The method according to claim 1, characterized in that, Based on the data generated by the network support domain system, obtain the perceived characteristics of business behavior of users who have not left the network, including: According to the first preset extraction cycle, the user face call detail records of the non-disconnected users are obtained from the data generated by the network support domain system; The mobile internet access log data of the non-disconnected user is obtained from the user face call detail record (CDR), and features are extracted from the mobile internet access log data to obtain the service behavior perception features of the non-disconnected user.

3. The method according to claim 1, characterized in that, Based on the data generated by the network support domain system, the call behavior characteristics of users who have not left the network are obtained, including: According to the second preset extraction cycle, the voice call details of the users who have not left the network are obtained from the data generated by the network support domain system; Voice service data is obtained from the voice call detail records, and features are extracted from the voice service data of the users who have not left the network to obtain the call behavior features of the users who have not left the network.

4. The method according to claim 1, characterized in that, Before determining the business behavior profile and voice service profile corresponding to each user who has not left the network through the user type prediction model, the method further includes: Based on the data generated by the business support domain system, the key personal information characteristics of the users who have not left the network are obtained; Based on the data generated by the network support domain system, obtain the network performance index characteristics and network coverage quality characteristics of the users who have not left the network.

5. The method according to claim 4, characterized in that, Based on the data generated by the business support domain system, obtain the key personal information features of the users who have not disconnected from the network, including: According to the second preset extraction cycle, the personal information data of the users who have not left the network is obtained from the data generated by the business support domain system; Feature extraction is performed on the personal information data of the users who have not left the network to obtain the key personal information features of the users who have not left the network.

6. The method according to claim 4, characterized in that, Based on the data generated by the network support domain system, obtain the network performance index characteristics and network coverage quality characteristics of the users who have not left the network, including: Based on the control plane call detail records (CDRs) generated by the network support domain system, obtain the control plane signaling data; and extract features from the control plane signaling data to obtain the network performance indicator features. Based on the coverage-related call detail records (CDRs) generated by the network support domain system, the coverage-related service data is obtained; and features are extracted from the coverage-related service data to obtain the network coverage quality information features.

7. The method according to claim 4, characterized in that, Based on the business behavior profile and voice service profile, users with potential migration types are predicted from the non-disconnected users, including: Based on the business behavior profile and the voice service profile, and combined with the personal key information features, the network performance index features, and the network coverage quality features, the user type prediction model predicts potential porting-out users from the users who have not left the network.

8. The method according to claim 1, characterized in that, Based on the business behavior profile and voice service profile, users with potential migration types are predicted from the non-disconnected users, including: Based on the business behavior profile and voice business profile, and combined with the questionnaire information of users who have left the network and the predefined leaving tags, the user type prediction model is trained using artificial intelligence binary classification technology, so as to predict users with potential leaving types through the trained user type prediction model.

9. The method according to claim 1, characterized in that, After predicting potential porting-out users from the non-disconnected users based on the business behavior profile and voice service profile, the method further includes: Based on the mobile internet access log data and voice service data of the potential porting-out users, it is predicted whether the potential porting-out users will port themselves. The mobile internet access log data of the potential porting-out users includes data on visits to websites of other network operators; the voice service data of the potential porting-out users includes operator customer service number data.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the steps of the method according to any one of claims 1 to 9.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 9.

12. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method described in any one of claims 1-9.

Images