A method and related device for quantitatively scoring and predicting satisfaction of a mobile network

Abstract

The application discloses a mobile network satisfaction quantitative score prediction method and related equipment, which can be applied to the field of communication technology. According to the current operation domain data and the current service domain data, the current perception feature wide table data is constructed, all target users are classified and grouped in combination with the current user package information, the current user traffic information and the current user call frequency information, then the current data subset corresponding to each current user subgroup is obtained through screening, the current satisfaction prediction sub-score of the target user in each category is predicted, the current satisfaction prediction sub-score of the target user in all categories is subjected to regression processing to obtain the current satisfaction prediction score of the target user, thereby the influence of single model prediction can be effectively reduced, and then the current satisfaction prediction score is adjusted based on a hierarchical rejection method to obtain the satisfaction target score of the target user, so that the influence of a black box can be reduced, and the accuracy of the satisfaction prediction result can be effectively improved.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a quantitative scoring prediction method and related equipment for mobile network satisfaction. Background Technology

[0002] In related technologies, the role of satisfaction prediction models in communication processes is to predict user satisfaction levels in advance by analyzing network metrics and user behavior data, thereby guiding network optimization and operation and maintenance decisions. Existing satisfaction prediction models suffer from low accuracy due to incomplete characterization of user behavior features corresponding to the labels used. Furthermore, the modeling process of existing satisfaction prediction models is based on mathematical calculations, resulting in a black-box nature. This weakens the actual effectiveness of some strongly correlated indicators based on expert experience during training, leading to low accuracy in satisfaction predictions during application. In addition, existing satisfaction prediction processes use a single satisfaction prediction model to predict satisfaction for all users, but significant differences exist between different user groups; therefore, the accuracy of using a single satisfaction prediction model to predict satisfaction for all users is low.

[0003] In summary, the technical problems existing in the relevant technologies need to be improved. Summary of the Invention

[0004] The main objective of this application is to propose a quantitative scoring prediction method and related equipment for mobile network satisfaction, which can effectively improve the accuracy of satisfaction prediction results.

[0005] To achieve the above objectives, one aspect of this application proposes a method for quantitatively predicting mobile network satisfaction scores, the method comprising the following steps: Construct a wide table of current perception features based on current operational domain data and current business domain data; Based on the current user perception feature wide table data, combined with the current user package information, current user data traffic information, and current user call frequency information, all target users are categorized and grouped to obtain several current user subgroups under each category; The current data subset corresponding to each current user subgroup is obtained by filtering the current perceptual feature wide table data; Each of the current data subsets is input into the pre-trained satisfaction prediction model to predict the score, thereby obtaining the current satisfaction prediction sub-score. Regression processing is performed on the current satisfaction prediction sub-scores of each target user across all categories to obtain the current satisfaction prediction score of the target user. The current predicted satisfaction score is adjusted based on a hierarchical veto method to obtain the target satisfaction score for the target user.

[0006] In some embodiments, the training process of the satisfaction prediction model includes the following steps: Training and test sets were constructed based on historical outbound call data related to satisfaction levels. A wide table of historical user perception characteristics is constructed based on historical operational domain data and historical business domain data. Based on the historical user perception feature wide table data, combined with historical user package information, historical user traffic information, and historical user call frequency information, all historical users are categorized and grouped to obtain several historical user subgroups under each category; A subset of historical data for each historical user subgroup is obtained by filtering from the historical user perception feature wide table data and the training set; Each subset of historical data is input into the satisfaction prediction model to be trained to predict the score, and the historical satisfaction prediction sub-score is obtained. Regression processing is performed on the historical satisfaction prediction sub-scores of each historical user under all categories to obtain the historical satisfaction prediction score of the historical user. The satisfaction prediction model trained in the current round is tested and verified based on the test set and the historical satisfaction prediction scores.

[0007] In some embodiments, constructing the current perception feature wide table data based on the current operational domain data and the current business domain data includes: Retrieve current operational domain data and current business domain data; Construct a mapping relationship between the current operational domain data and the current business domain data to obtain a current perception profile integrated data list; The data in the current perception profile integrated data list is filtered to obtain a filtered perception profile integrated data list; Based on the filtered perception profile integrated data list, the tag data is expanded to obtain the expanded perception profile integrated data list; The current perception feature wide table data is constructed based on the data in the expanded perception profile integrated data list.

[0008] In some embodiments, filtering the data in the current perception profile integrated data list to obtain a filtered perception profile integrated data column includes: Based on the usefulness of the data in the current perception profile integrated data list, data is selected to obtain the first perception profile integrated data column; Based on the correlation between the data in the first perception profile integrated data column, data merging and data filtering are performed to obtain the filtered perception profile integrated data column.

[0009] In some embodiments, the step of expanding the label data based on the filtered perception profile integrated data list to obtain the expanded perception profile integrated data list includes: Obtain the current raw depth packet detection information; The target sector data is calculated based on the current original depth packet detection information; The terminal profile is obtained by querying the current original deep packet inspection information; Based on the current original deep packet inspection information, obtain the mobile phone number to be processed, the WeChat identifier to be processed, or the broadband account to be processed. The target WeChat identifier is obtained by querying the mobile phone number to be processed or the broadband account to be processed. The target broadband account is obtained by querying the mobile phone number to be processed and the WeChat identifier to be processed. The target sector data, the terminal profile, the target WeChat identifier, and the target broadband account are filled into the filtered perception profile integrated data list to obtain the expanded perception profile integrated data list.

[0010] In some embodiments, adjusting the current satisfaction prediction score based on the hierarchical veto method includes: Obtain the wide table data of application perception features for the preset time period before the current application prediction node; The dynamic threshold of the target user is determined based on the wide table data of the perceptual features to be applied, and the dynamic threshold includes a first threshold, a second threshold, or a third threshold. If there is data to be processed in the current perceptual feature wide table data that is less than the dynamic threshold, a preset score is obtained as the target deduction value; otherwise, a dynamic weight is calculated based on the current perceptual feature wide table data, and a target deduction value is calculated based on the dynamic weight and the preset score. The current satisfaction prediction score is adjusted based on the target deduction value.

[0011] In some embodiments, the step of calculating dynamic weights based on the current perceptual feature wide table data, and calculating a target deduction value based on the dynamic weights and the preset score, includes: Calculate the network quality coefficient based on the current perceptual feature wide table data; The dynamic weights are obtained by adjusting the base weights based on the network quality coefficients. The target deduction value for the corresponding category is calculated based on the dynamic weight and the preset score.

[0012] To achieve the above objectives, another aspect of this application proposes a quantitative scoring prediction device for mobile network satisfaction, the device comprising: The building module is used to construct the current perception feature wide table data based on the current operation domain data and the current business domain data; The category grouping module is used to group all target users into categories based on the current user perception feature wide table data combined with the current user package information, current user traffic information and current user call frequency information, to obtain several current user subgroups under each category; The filtering module is used to filter the current data subset corresponding to each current user subgroup from the current perception feature wide table data; The prediction module is used to input each of the current data subsets into the pre-trained satisfaction prediction model to predict the score and obtain the current satisfaction prediction sub-score. The regression processing module is used to perform regression processing on the current satisfaction prediction sub-score of each target user under all categories to obtain the current satisfaction prediction score of the target user. The adjustment module is used to adjust the current satisfaction prediction score based on the hierarchical veto method to obtain the target satisfaction score of the target user.

[0013] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0014] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the methods described above.

[0015] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer program product, including a computer program that, when executed by a processor, implements the aforementioned method.

[0016] The embodiments of this application include at least the following beneficial effects: This application provides a method and related equipment for quantitatively predicting mobile network satisfaction scores. This scheme constructs a wide table of current perceived features based on current operating domain data and current service domain data. Then, based on the wide table of current user perceived features, combined with current user package information, current user traffic information, and current user call frequency information, all target users are categorized into several current user subgroups under each category. Next, a subset of current data corresponding to each current user subgroup is selected from the wide table of current perceived features. Each current data subset is then input into a pre-trained satisfaction prediction model for score prediction, thereby obtaining the current satisfaction prediction sub-score for each target user in each category. Regression processing is then performed on the current satisfaction prediction sub-scores for each target user across all categories to obtain the current satisfaction prediction score for the target user. This effectively reduces the impact of a single model prediction. Finally, a hierarchical rejection method is used to adjust the current satisfaction prediction score to obtain the target user's satisfaction target score, thereby reducing the black-box effect and effectively improving the accuracy of the satisfaction prediction results. Attached Figure Description

[0017] Figure 1 This is a flowchart of a method for quantitatively predicting mobile network satisfaction provided in an embodiment of this application; Figure 2 This is a flowchart of the training process for the satisfaction prediction model provided in the embodiments of this application; Figure 3 This is a flowchart illustrating the acquisition of perceptual feature wide table data provided in an embodiment of this application; Figure 4 This is a flowchart illustrating the expansion of target sector data provided in an embodiment of this application; Figure 5 This is a flowchart illustrating the process of obtaining a terminal profile as provided in an embodiment of this application. Figure 6 This is a schematic diagram of the grouping categories provided in the embodiments of this application; Figure 7 This is a flowchart of adjusting the current satisfaction prediction score provided in an embodiment of this application; Figure 8 This is a schematic diagram of the structure of a mobile network satisfaction quantification scoring prediction device provided in an embodiment of this application; Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application.

[0019] It is understood that the terms “first,” “second,” etc., used in this application may be used herein to describe various concepts, but unless otherwise stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the words “if,” “when,” or “in response to a determination” as used herein may be interpreted as “when…” or “when…” or “in response to a determination.”

[0020] As used in this application, the terms "at least one", "multiple", "each", "any", etc., "at least one" includes one, two or more, "multiple" includes two or more, "each" refers to each of the corresponding multiples, and "any" refers to any one of the multiples.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0022] Before providing a detailed description of the embodiments of this application, some of the nouns and terms used in the embodiments of this application will be explained first. The nouns and terms used in the embodiments of this application shall be interpreted as follows: LightGBM (Light Gradient Boosting Machine) is a gradient boosting framework that uses decision trees as base learners. LightGBM is designed for efficient parallel computing, and its "Light" aspect is reflected in several areas: faster training speed; lower memory usage; support for single-machine multi-threading, multi-machine parallel computing, and GPU training; and the ability to handle large-scale data.

[0023] VoNR (Voice over New Radio) is the target voice solution for 5G networks. The radio access portion of 5G is called NR (New Radio), and voice services based on 5G are called VoNR. This means that NR users can directly conduct voice services on the NR network without falling back to the LTE network, thus obtaining a higher quality voice service experience and a higher data service speed.

[0024] CEM (Consumer Satisfaction), also known as the Customer Satisfaction Index, is a shorthand for a customer satisfaction survey system in the service industry. It measures the degree of match between customer expectations and customer experience. In other words, it's an index derived by comparing a customer's perceived experience with their expectations of a product.

[0025] In the telecommunications field, the Operation Domain (O) primarily focuses on network operation and resource management, ensuring the stable operation of the communication network. Its main functions include network monitoring, fault management, performance optimization, and resource scheduling. The Business Domain (B) primarily focuses on user and business operations, responsible for customer management, billing, and marketing. Its functions include customer management, calculation and settlement, marketing and customer service, and service activation. The two domains work closely together through interfaces (such as APIs and message queues) to jointly support the operator's end-to-end services.

[0026] User perception and profiling data integration refers to the cross-domain correlation and fusion analysis of user perception data (such as network experience metrics) and user profile data (such as consumption habits and behavioral preferences). The aim is to break down data silos between the O domain (network domain) and the B domain (business domain), enabling more precise service optimization and marketing decisions. Perception data (O domain) includes KQIs (Key Quality Indicators) and KPIs (Key Performance Indicators) collected from the network side, such as webpage loading latency, video buffering rate, signal strength, and number of dropped calls. Profiling data (B domain) includes user attributes (age, data plan), consumption behavior (ARPU value, data usage habits), terminal information (phone model, operating system), and location trajectory.

[0027] The 4G / 5G user perception profile wide table refers to a wide table structure used by operators or data analysis platforms to store user network experience and behavioral characteristics. This type of wide table integrates data from the O domain (network domain) and B domain (service domain), organizing multi-dimensional user information in a "one user per line" format, facilitating subsequent profile analysis, model training, and business decision-making.

[0028] Wireless DPI (Deep Packet Inspection) is a technology deployed in mobile communication networks to perform in-depth parsing and analysis of data packets passing through the wireless network. Unlike shallow inspection based on the IP 5-tuple (source / destination IP, port, protocol), DPI can delve into the application layer of data packets to identify specific service types, protocol characteristics, and even content, providing a data foundation for network optimization, user services, security protection, and business innovation.

[0029] DPI (Deep Packet Inspection) user resident scenarios refer to the main activity areas or typical behavioral environments of users identified by using Deep Packet Inspection (DPI) technology combined with long-term, stable network behavior data.

[0030] User-sector occurrence days refer to the number of unique days within a given statistical period (usually one month) during which a user has generated network activity (such as attach, service requests, location updates) in a specific base station sector (Cell).

[0031] The sector TOPI scenario refers to the TOPI (Top Indicator) analysis scenario based on the sector (Cell) dimension. A sector is the smallest logical unit of a base station's coverage (a base station typically contains 3 sectors, each covering a 120° direction). The TOPI scenario identifies the sectors that perform best (Top N) or require the most attention in a specific indicator through sorting or threshold filtering, and then performs specific analysis or optimization on these sectors.

[0032] Effective user-sector refers to the user-sector pair that has analytical value in communication network data analysis, selected through specific rules. The aim is to eliminate invalid or noisy data and focus on the combination that truly reflects user behavior and network status.

[0033] The number of days a sector appears in a month refers to the number of unique days in a month when a user has generated network activity (such as attach, service request, location update) in a certain sector under a certain sector in communication network data analysis.

[0034] The 50th percentile, also known as the median, is a key indicator in statistics used to describe the central tendency of data. It represents the value in the middle when a set of data is arranged in ascending order.

[0035] In related technologies, existing satisfaction prediction models suffer from low accuracy due to incomplete characterization of user behavior features corresponding to the labels used. Furthermore, the modeling process of existing satisfaction prediction models is based entirely on mathematical calculations, resulting in a black-box nature. This weakens the actual effectiveness of some strongly correlated indicators based on expert experience during training, leading to low accuracy in application. Additionally, existing satisfaction prediction processes use a single model to predict satisfaction for all users, but significant differences exist between different user groups; therefore, using a single model to predict satisfaction for all users results in low accuracy.

[0036] In view of this, this application provides a method and related equipment for quantitatively predicting mobile network satisfaction, which can effectively improve the accuracy of satisfaction prediction results.

[0037] This application provides a method for quantitatively predicting mobile network satisfaction scores, relating to the field of communication technology. This method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or in-vehicle terminal, but is not limited to these. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application implementing the method for quantitatively predicting mobile network satisfaction scores, but is not limited to the above forms.

[0038] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0039] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.

[0040] The embodiments of this application are described below with reference to the accompanying drawings: Figure 1 This is an optional flowchart for predicting a quantitative score of mobile network satisfaction, provided in an embodiment of this application. Figure 1 The method may include, but is not limited to, steps S110 to S160: Step S110: Construct the current perception feature wide table data based on the current operation domain data and the current business domain data; Step S120: Based on the current user perception feature wide table data, combined with the current user package information, current user data traffic information and current user call frequency information, all target users are categorized and grouped to obtain several current user subgroups under each category; Step S130: Filter the current data subset corresponding to each current user subgroup from the current perceptual feature wide table data; Step S140: Input each current data subset into the pre-trained satisfaction prediction model to predict the score and obtain the current satisfaction prediction sub-score. Step S150: Perform regression processing on the current satisfaction prediction sub-scores of each target user under all categories to obtain the current satisfaction prediction score of the target user; Step S160: Adjust the current satisfaction prediction score based on the hierarchical veto method to obtain the target satisfaction score for the target user.

[0041] It is understandable that, such as Figure 2 As shown, the training process of the satisfaction prediction model involved in the embodiments of this application includes, but is not limited to, the following steps: Step S210: Construct training and testing sets based on historical outbound call data related to satisfaction levels; Step S220: Construct a wide table of historical user perception characteristics based on historical operation domain data and historical business domain data; Step S230: Based on the historical user perception feature wide table data, combined with historical user package information, historical user traffic information and historical user call frequency information, all historical users are categorized and grouped to obtain several historical user subgroups under each category; Step S240: Select historical data subsets for each historical user subgroup from the historical user perception feature wide table data and the training set; Step S250: Input each historical data subset into the satisfaction prediction model to be trained to predict the score and obtain the historical satisfaction prediction sub-score. Step S260: Perform regression processing on the historical satisfaction prediction sub-scores of each historical user under all categories to obtain the historical satisfaction prediction score of the historical user. Step S270: Test and verify the satisfaction prediction model trained in the current round based on the test set and historical satisfaction prediction scores.

[0042] Specifically, the historical outbound satisfaction data in this embodiment includes, but is not limited to, provincial human and AI-generated satisfaction outbound data and group-wide human-generated satisfaction outbound data. Both provincial human and AI-generated satisfaction outbound data and group-wide human-generated satisfaction outbound data can be obtained quarterly. Historical user scores (between 1 and 10 points) based on internet usage, voice communication, and overall performance are extracted, and outliers, such as outbound calls without scores, are removed. In this embodiment, 80% of all satisfaction outbound data can be used as the training set, and the remaining 20% ​​as the test set.

[0043] It is understood that the processing procedure for the user-perceived feature wide table data involved in the training and application of the satisfaction prediction model is the same in this embodiment. This embodiment uses the user-perceived feature wide table data during application as an example to illustrate the specific processing procedure. Specifically, as follows... Figure 3As shown, the process of constructing the current perception feature wide table data based on the current operation domain data and the current business domain data in this embodiment includes, but is not limited to, the following steps: Step S310: Obtain current operational domain data and current business domain data; Step S320: Construct the mapping relationship between the current operation domain data and the current business domain data to obtain the current perception profile integrated data list; Step S330: Filter the data in the current perception profile integrated data list to obtain the filtered perception profile integrated data list; Step S340: Based on the filtered perception profile integrated data list, expand the label data to obtain the expanded perception profile integrated data list; Step S350: Construct the current perceptual feature wide table data based on the data in the expanded perceptual profile data list.

[0044] Specifically, the current operating domain (B domain) data in this embodiment may include, but is not limited to, data such as package, terminal, tariff, broadband, complaint records, hotline, business hall, WIFI, age, user value, etc.; the current service domain (O domain) data may include, but is not limited to, data such as base station operating parameters, base station performance, high load, base station configuration, traffic, resident, MR (measurement report), perceived DPI (deep packet inspection), data service KQI (key quality indicator), and call data.

[0045] This embodiment, after obtaining the current operational domain data and the current business domain data, can clean the data separately before mapping and associating them. The association methods include, but are not limited to, time alignment, spatial association, or identifier unification. After completing the mapping operation for all data, this embodiment can use machine learning algorithms to transform all associated data into readable tags, thereby obtaining the current perception profile integrated data list.

[0046] Specifically, the current perception profile integrated data list in this embodiment can include at least 256 indicator data. However, the indicator data in the perception profile integrated data list may contain indicator data with poor correlation to the current business type, indicator data with poor description of the current business, or data with duplicate descriptions of the characteristics of the current business. Therefore, this embodiment can perform selection, merging, or filtering on the data in the current perception profile integrated data list to eliminate indicator data with excessive scenario fluctuations, compressing the 256 indicator data into 156 indicator data to obtain the filtered perception profile integrated data column. Among them, correlation can be characterized by Pearson correlation coefficient. The simplification and merging process can merge indicator data based on information such as significance, business interpretability, and model contribution, and the fusion process can fuse indicator data based on information such as maximum information entropy, collinearity, and cross-domain consistency.

[0047] Specifically, in this embodiment, the process of filtering data in the current perception profile integrated data list can be as follows: First, a first perception profile integrated data column is obtained by selecting data based on the usefulness of the data in the current perception profile integrated data list. Then, data merging and filtering are performed based on the correlation between data in the first perception profile integrated data column to obtain a filtered perception profile integrated data column. For example, the DPI data in the current perception profile integrated data list includes uplink TCP establishment latency percentage (<=40ms), game interaction latency percentage (<=50ms), total duration of large packet downlink time (>=500KB), and TCP packet retransmission rate; the DPI voice indicator data includes the caller's perceived connection average latency (ms) as excellent, the slice RTP packet loss rate (%) as excellent, the slice MOS total score, and the slice average MOS. In DPI data, highly accurate metrics include uplink TCP establishment latency percentage (<=40ms) and game interaction latency percentage (<=50ms). Inaccurate metrics include total duration of large packet downlink time (>=500KB) and TCP packet retransmission rate. Therefore, selection can be based on the usefulness of the metrics in describing DPI data, retaining uplink TCP establishment latency percentage (<=40ms) and game interaction latency percentage (<=50ms), while discarding total duration of large packet downlink time (>=500KB) and TCP packet retransmission rate. Similarly, in DPI voice metrics, highly accurate data includes excellent caller-perceived average connection latency (ms) and excellent slice RTP packet loss rate (%). Inaccurate data includes total slice MOS score and average slice MOS. Therefore, selection can be based on the usefulness of the DPI voice metrics, retaining excellent caller-perceived average connection latency (ms) and excellent slice RTP packet loss rate (%), while discarding total slice MOS score and average slice MOS.

[0048] Specifically, after selecting data from the integrated data list of the perception profile, a first integrated data list of perception profiles is obtained. Within this first integrated data list, there may be data with similar indicator meanings or overly granular breakdowns. Therefore, this embodiment can analyze the correlations among all data in the first integrated data list of perception profiles and then simplify and merge the data based on these correlations. For example, assuming that the MR 4 / 5G coverage index data in the first perception profile integrated data list includes the total 4 / 5G_MR sampling points, the 4 / 5G_MR sampling points greater than -110, the 4 / 5G_MR coverage percentage (%), the 4 / 5G_sir sampling points less than 0, and the 4 / 5G_mr interference percentage (%), then the correlation of these data can be analyzed, and the relevant data can be merged. Therefore, the total 4 / 5G_MR sampling points, the 4 / 5G_MR sampling points greater than -110, and the 4 / 5G_MR coverage percentage (%) can be merged into the MR coverage percentage (%) good ratio data, and the 4 / 5G_sir sampling points less than 0 and the 4 / 5G_mr interference percentage (%) can be merged into the MR interference percentage (%) good ratio data.

[0049] Meanwhile, this embodiment also analyzes the correlation between the data in the first perception profile integrated data list and the satisfaction score prediction process, thereby filtering the data in the first perception profile integrated data list. The correlation in this embodiment can be calculated using the Pearson algorithm. For example, assuming that the correlation calculated by the Pearson algorithm for "whether there is broadband under the same main package [Mobile Information Broad Table]" is 7.14%, "whether there is a bundled package [B Domain Basic Information]" is 6.82%, "shopping preference TCP average latency (ms) good / bad ratio [DPI]" is 5.35%, and "age [B Domain Basic Information]" is 4.35%, this embodiment can sort the indicator data from largest to smallest based on the calculated correlations, and select the top 30% of the indicator data to form the filtered perception profile integrated data list.

[0050] It is understandable that, since the data in the filtered perception profile integrated data list may lack some indicator data related to satisfaction scores, in order to improve the accuracy of satisfaction score prediction, this embodiment will further expand the label data after obtaining the filtered perception profile integrated data list, and then merge the expanded indicator data into the filtered perception profile integrated data list to obtain the expanded perception profile integrated data list. Specifically, this embodiment can obtain the target sector data by acquiring the current original deep packet inspection information, querying the terminal profile based on the current original deep packet inspection information, and obtaining the mobile phone number to be processed, the WeChat identifier to be processed, or the broadband account to be processed based on the current original deep packet inspection information. Then, the target WeChat identifier is queried based on the mobile phone number to be processed or the broadband account to be processed, and the target broadband account is queried based on the mobile phone number to be processed and the WeChat identifier to be processed. Finally, the target sector data, terminal profile, target WeChat identifier, and target broadband account are filled into the filtered perception profile integrated data list to obtain the expanded perception profile integrated data list.

[0051] For example, such as Figure 4 The process of acquiring target sector data, as shown in this embodiment, involves acquiring all current raw deep packet inspection (DPI) information within a certain period, cleaning all the current raw DPI information, and then calculating the user-sector occurrence days based on the cleaned current raw DPI information. When the occurrence days of a user-sector are less than or equal to 7 days, it indicates that the corresponding user may have only passed through the sector area formed by the base station. Therefore, the occurrence days of that user-sector are deleted, and only user-sector occurrence days greater than 7 days are retained, and the corresponding sector is taken as a valid user-sector. Then, the monthly occurrence days of the user's sector within the valid user-sector are counted. If the monthly occurrence days of the sector are less than 6 days, it indicates that the user may be an abnormal user. The data of abnormal users are filtered out, and valid user-sectors with monthly occurrence days greater than or equal to 6 days are retained. The TOPI scene of the retained valid user-sectors is calculated as the target sector data.

[0052] like Figure 5The process of obtaining the terminal profile shown in this embodiment involves acquiring all current raw deep packet inspection (DPI) information within a certain period, extracting the International Mobile Equipment Identity (IME) from the current raw deep packet inspection information, and then extracting the clear Signal Assignment Code (TAC) from all IMEs. The extracted TAC is then matched with the data in the TAC database. If no matching TAC is found, the TAC is marked as an abnormal TAC and an abnormal data processing prompt is given. If a matching TAC is found, the terminal information is obtained based on the TAC, and a terminal profile is constructed based on the terminal brand, terminal model, and device type in the terminal information.

[0053] Specifically, after completing the filtering and expansion of indicator data, this embodiment constructs a current perception feature wide table data based on the data in the expanded perception profile integrated data list. For example, the current perception feature wide table data in this embodiment may include user information, internet access perception information, voice perception information, wireless coverage information, mobile network service information, XDR comprehensive statistical analysis information, persistent analysis information, broadband data, WiFi information, software speed test data, and service preference information, etc. User information includes, but is not limited to, the location of the number, the city of origin, the user's mobile phone number, the user's star rating, whether the user is a 5G contracted user, whether the user is an SA network user, the number of months the user has been on the network, the package name, the package charge level, whether the package has a data cap and speed limit, whether it is a campus package, and whether it is a 5G package; Internet access perception information includes, but is not limited to, the number of times the internet access quality is poor, the percentage of excellent HTTP request success rate, the percentage of excellent DNS resolution success rate, the percentage of excellent downlink TCP establishment success rate, the percentage of excellent average TCP establishment latency (ms), and the percentage of excellent average group video download speed (Mbps); Voice perception information includes, but is not limited to, the number of times the voice quality is poor, the percentage of excellent network connection rate, the percentage of excellent average call connection latency (ms) perceived by the caller, the percentage of excellent perceived dropped calls, the percentage of excellent perceived dropped calls, the percentage of excellent average MOS per slice, and the percentage of excellent single-pass per slice; Wireless coverage Coverage information includes, but is not limited to, MR weak coverage percentage (%) and good coverage percentage, MR interference percentage (%) and good coverage percentage, etc.; mobile network service information includes, but is not limited to, the number of times the operator's APP was accessed, the number of times the operator's service number was called, and the type of mobile network service accessed, etc.; XDR comprehensive statistical analysis information includes, but is not limited to, the user plane XDR failure percentage (%), the signaling plane XDR failure percentage (%), etc.; resident analysis information includes, but is not limited to, the total resident cell coverage scenarios, the total resident cell coverage types, the total resident cell poor quality times, the total resident cell alarm times, and the district / county to which the total resident cells belong, etc.; broadband data includes, but is not limited to, broadband contract speed, urban / rural attributes, duration of mismatch between broadband terminals at the same address, broadband demand level, gateway anomaly ratio, the number of times Wi-Fi quality was poor or network cable specifications were poor in the past year, customers with low scores in historical evaluations, customers contacted by proactive services in the past, the type of the last service event, and the interval between the last service events, etc.

[0054] It is understood that this embodiment, after obtaining the current perceptual feature wide table data, combines the current user's plan information, current user's data usage information, and current user's call frequency information to group all target users into categories, resulting in several current user subgroups under each category. Specifically, this embodiment can construct multiple categories, build different groups under each category, and then, based on the current user perceptual feature wide table data of each target user, assign the corresponding target user to the corresponding group under each category. After completing the grouping of all target users under all categories, several current user subgroups are obtained. For example, as shown... Figure 6 As shown, the categories constructed in this embodiment include, but are not limited to, packages, touchpoints, networks, WiFi, districts / counties, age, data usage, calls, terminals, and permanent residence. Under the package category, there are groups under 30 years old, [30-99) groups, [99-129) groups, [129-169) groups, [169-199) groups, and [199 and above) groups; under the touchpoint category, there are contact groups and contactless groups; under the network category, there are 4G groups and 5G groups; under the WiFi category, there are used WiFi groups and unused WiFi groups; under the district / county category, there are category 1 groups, category 2 groups, category 3 groups, and other groups; under the age category, there are [0-24) age groups, [24-30) age groups, [30-40) age groups, [40-50) age groups, and [50-60) age groups. Groups of 60+ G are categorized as follows: Under the traffic category, there are groups of [0-15)G, [15-30)G, [30-50)G, [50-70)G, and groups of 70G and above; under the call category, there are groups of [0-10) minutes, [10-30) minutes, [30-60) minutes, [60-100) minutes, [100-20) minutes, and 200 minutes; under the terminal category, there are groups of Brand 1, Brand 2, Brand 3, Brand 4, Brand 5, and Brand 6; under the permanent location category, there are groups of Location 1, Location 2, Location 3, Location 4, Location 5, and Location 6. In this embodiment, after completing the group division for each category, the target user's current perceived characteristic wide table data is used to determine the group they belong to within each category. After completing the group division for all target users, the target users within each group under each category are grouped into a user subgroup, thus obtaining multiple current user subgroups.

[0055] Understandably, after obtaining the user subgroups, in the training process of the satisfaction prediction model, this embodiment can calculate the importance of all data in the training set and the user perception feature wide table data to the user subgroups. Based on the calculated importance, it can select data with high correlation from the training set and the user perception feature wide table data to form a data subset. This data subset can then be used to train the satisfaction prediction model to obtain the corresponding user subgroup's satisfaction prediction sub-score. Similarly, in the application process of the satisfaction prediction model, this embodiment can calculate the importance of all data in the user perception feature wide table data to the user subgroups. Based on the calculated importance, it can select data with high correlation from the user perception feature wide table data to form a data subset. This data subset can then be input into the trained satisfaction prediction model for rating prediction to obtain the corresponding user subgroup's satisfaction prediction sub-score. For example, for non-contact groups, the data subsets selected in this embodiment and their indicator importance are shown in Table 1. Table 1

[0056] For the [0-15)G group, the data subsets obtained by this embodiment and the importance of their indicators are shown in Table 2: Table 2

[0057] Understandably, after predicting the sub-scores for each user in each category using the satisfaction prediction model, a regression tree is used to regress all predicted sub-scores for each user, thereby obtaining the user's predicted satisfaction score. Specifically, during the training process of the satisfaction prediction model, this embodiment compares the predicted satisfaction score with the actual satisfaction score in the data subset. Simultaneously, it uses a test set to test and validate the satisfaction prediction model for the current round until the difference between the predicted satisfaction score and the actual satisfaction score is less than a preset difference, at which point the model training process for the current cycle ends. In the application of the satisfaction prediction model, this embodiment uses the predicted satisfaction score processed by the regression tree as the predicted score for the corresponding user.

[0058] Specifically, in this embodiment, the satisfaction prediction model can be trained using a training set composed of provincial and group-level assessment data, and tested using a test set composed of group-level test data. This embodiment verifies the trained satisfaction prediction model by testing it in three areas: voice services, data services, and integrated services, and calculates the precision, recall, and F1 score for the overall, satisfied, and dissatisfied aspects during the testing process. The testing results for voice services are shown in Table 3. Table 3

[0059] The test results of the data service are shown in Table 4: Table 4

[0060] The test results of the integrated services are shown in Table 5: Table 5

[0061] Taking the LGBM (Light Gradient Boosting Machine) algorithm as an example for satisfaction prediction, the test results for cluster modeling in the test set are shown in Table 6: Table 6

[0062] The data verified by the above tests show that this embodiment can achieve good model training results in both cluster modeling and regression tree training.

[0063] It is understandable that in the application of the satisfaction prediction model in this embodiment, after obtaining the user's predicted satisfaction score, the current predicted satisfaction score will be adjusted based on a hierarchical veto method. Specifically, as follows... Figure 7 As shown, the process of adjusting the current satisfaction prediction score in this embodiment includes, but is not limited to, the following steps: Step S610: Obtain the wide table data of the application perception features to be predicted for the preset time before the current application prediction node; Step S620: Determine the dynamic threshold of the target user based on the wide table data of the perception features to be applied, wherein the dynamic threshold includes a first threshold, a second threshold, or a third threshold. Step S630: If there is data to be processed in the current perceptual feature wide table data that is less than the dynamic threshold, obtain the preset score as the target deduction value; otherwise, calculate the dynamic weight based on the current perceptual feature wide table data, and calculate the target deduction value based on the dynamic weight and the preset score. Step S640: Adjust the current satisfaction prediction score according to the target deduction value.

[0064] Specifically, after obtaining the current perceptual feature wide table data, this embodiment can select numbers whose satisfaction prediction model is degraded (i.e., the model predicts results with negative impacts) and whose one-way network indicators are below the threshold (e.g., data limit, service suspension due to overdue payment, or service suspension due to fraud) based on the test result list numbers. Then, the indicator data corresponding to the number with a precision rate of more than 50% in the current perceptual feature wide table data can be included in the veto deduction items to improve the accuracy of the satisfaction prediction score.

[0065] Specifically, in this embodiment, after determining the phone number corresponding to the user whose satisfaction prediction score needs adjustment, a network quality coefficient can be calculated based on the current perceptual feature wide table data. After adjusting the base weights according to the network quality coefficient to obtain dynamic weights, the target deduction value for the corresponding category is calculated based on the dynamic weights and preset scores. For example, taking network coverage as an example, the network quality coefficient in this embodiment is 1.2 - 0.4. Coverage; Dynamic weight = Base weight Network quality coefficient (Qnk).

[0066] Understandably, taking network connectivity rate, MR coverage percentage, and data throttling from the wide-scale data table of the perception features to be applied as examples, when the network connectivity rate is less than the first threshold (e.g., 30%), the MR coverage percentage is less than the second threshold (e.g., 30%), or the number of data throttling incidents in the past 3 months equals the third threshold, the pre-scored value corresponding to the relevant indicator data is directly rejected. That is, the preset score is selected as the target deduction value to be deducted from the current satisfaction prediction score. Conversely, after calculating the dynamic weights based on the current wide-scale data table of perception features, the target deduction value is calculated based on the dynamic weights and the preset score. That is, the target deduction value = the preset score. The target user's satisfaction target score is obtained by dynamically weighting the target score and then subtracting the target deduction value from the current satisfaction prediction score.

[0067] In some embodiments, the complete implementation process of the method of this embodiment is as follows: Step 1: Construct training and testing sets based on historical outbound call data related to satisfaction levels; Step 2: After constructing a unified profile list based on the O and B domain data, perform data processing to obtain historical user perception feature wide table data; Step 3: The historical user perception feature wide table data is used to divide historical users into groups under each category, resulting in historical user subgroups under each category; Step 4: Select the data corresponding to each historical user subgroup from the training set and the historical user perception feature wide table data to form a historical data subset, and use the historical data subset to perform cluster training on the satisfaction prediction model to obtain the historical satisfaction prediction sub-score for each subgroup. Step 5: Perform regression fusion on all historical satisfaction prediction sub-scores to obtain the historical satisfaction prediction score; Step 6: Test and validate the satisfaction prediction model based on the test set and historical satisfaction prediction scores; Step 7: After constructing the integrated profile list based on the O-domain and B-domain data, perform data processing to obtain the current user perception feature wide table data; Step 8: Determine the group to which the current user belongs in each category, resulting in several subgroups of the current user; Step 9: Filter the current data subset corresponding to the current user subgroup from the current perceptual feature wide table data; Step 10: Input the current data subset into the pre-trained satisfaction prediction model to predict the score and obtain the current satisfaction prediction sub-score; Step 11: Perform regression processing on the current satisfaction prediction sub-scores of the target user under all categories to obtain the current satisfaction prediction score of the target user; Step 12: Adjust the current satisfaction prediction score based on the hierarchical veto method to obtain the target satisfaction score for the target user.

[0068] As can be seen from the above, the method of this application embodiment can effectively improve the accuracy of satisfaction prediction by using group training or application prediction during the training and application process, then combining regression algorithm to regress and fuse the scores of all groups, and adjusting the predicted scores by combining hierarchical rejection method during the application process.

[0069] Please see Figure 8 This application also provides a quantitative scoring prediction device for mobile network satisfaction, the device comprising: The building module is used to construct the current perception feature wide table data based on the current operation domain data and the current business domain data; The category grouping module is used to group all target users into categories based on the current user perception feature wide table data combined with the current user package information, current user data traffic information and current user call frequency information, resulting in several current user subgroups under each category; The filtering module is used to filter the current data subset corresponding to each current user subgroup from the current perceptual feature wide table data; The prediction module is used to input each current data subset into the pre-trained satisfaction prediction model to predict the score and obtain the current satisfaction prediction sub-score. The regression processing module is used to perform regression processing on the current satisfaction prediction sub-scores of each target user under all categories to obtain the current satisfaction prediction score of the target user. The adjustment module is used to adjust the current satisfaction prediction score based on the hierarchical veto method to obtain the target satisfaction score for the target user.

[0070] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0071] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0072] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0073] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 810 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 820 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 820 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 820 and is called and executed by the processor 810 using the methods described in the embodiments of this application. The input / output interface 830 is used to implement information input and output; The communication interface 840 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 850 transmits information between various components of the device (e.g., processor 810, memory 820, input / output interface 830, and communication interface 840); The processor 810, memory 820, input / output interface 830 and communication interface 840 are connected to each other within the device via bus 850.

[0074] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0075] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0076] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0077] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0078] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0079] This application provides a method and related equipment for quantitatively predicting mobile network satisfaction scores. This method constructs a wide table of current user perception features based on current operating domain data and current service domain data. Then, it groups all target users into categories based on the wide table of current user perception features, combined with current user package information, current user data traffic information, and current user call frequency information. Several current user subgroups are obtained under each category. Next, a subset of current data corresponding to each current user subgroup is selected from the wide table of current user perception features. Each current data subset is then input into a pre-trained satisfaction prediction model for score prediction, resulting in a predicted sub-score of the target user's current satisfaction in each category. Regression processing is then performed on the predicted sub-scores of the target user's current satisfaction across all categories to obtain the target user's current satisfaction prediction score. This effectively reduces the impact of a single model prediction. Finally, a hierarchical rejection method is used to adjust the predicted current satisfaction score to obtain the target user's target satisfaction score, thereby reducing the black-box effect and effectively improving the accuracy of the satisfaction prediction results.

[0080] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0081] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0082] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0083] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0084] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0085] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0086] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0087] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0088] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0089] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0090] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for quantitatively predicting mobile network satisfaction, characterized in that, The method includes the following steps: Construct a wide table of current perception features based on current operational domain data and current business domain data; Based on the current user perception feature wide table data, combined with the current user package information, current user data traffic information, and current user call frequency information, all target users are categorized and grouped to obtain several current user subgroups under each category; The current data subset corresponding to each current user subgroup is obtained by filtering the current perceptual feature wide table data; Each of the current data subsets is input into the pre-trained satisfaction prediction model to predict the score, thereby obtaining the current satisfaction prediction sub-score. Regression processing is performed on the current satisfaction prediction sub-scores of each target user across all categories to obtain the current satisfaction prediction score of the target user. The current predicted satisfaction score is adjusted based on a hierarchical veto method to obtain the target satisfaction score for the target user.

2. The method according to claim 1, characterized in that, The training process of the satisfaction prediction model includes the following steps: Training and test sets were constructed based on historical outbound call data related to satisfaction levels. A wide table of historical user perception characteristics is constructed based on historical operational domain data and historical business domain data. Based on the historical user perception feature wide table data, combined with historical user package information, historical user traffic information, and historical user call frequency information, all historical users are categorized and grouped to obtain several historical user subgroups under each category; A subset of historical data for each historical user subgroup is obtained by filtering from the historical user perception feature wide table data and the training set; Each subset of historical data is input into the satisfaction prediction model to be trained to predict the score, and the historical satisfaction prediction sub-score is obtained. Regression processing is performed on the historical satisfaction prediction sub-scores of each historical user under all categories to obtain the historical satisfaction prediction score of the historical user. The satisfaction prediction model trained in the current round is tested and verified based on the test set and the historical satisfaction prediction scores.

3. The method according to claim 1, characterized in that, The construction of the current perception feature wide table data based on the current operational domain data and the current business domain data includes: Retrieve current operational domain data and current business domain data; Construct a mapping relationship between the current operational domain data and the current business domain data to obtain a current perception profile integrated data list; The data in the current perception profile integrated data list is filtered to obtain a filtered perception profile integrated data list; Based on the filtered perception profile integrated data list, the tag data is expanded to obtain the expanded perception profile integrated data list; The current perception feature wide table data is constructed based on the data in the expanded perception profile integrated data list.

4. The method according to claim 3, characterized in that, The step of filtering the data in the current perception profile integrated data list to obtain the filtered perception profile integrated data column includes: Based on the usefulness of the data in the current perception profile integrated data list, data is selected to obtain the first perception profile integrated data column; Based on the correlation between the data in the first perception profile integrated data column, data merging and data filtering are performed to obtain the filtered perception profile integrated data column.

5. The method according to claim 3, characterized in that, The expanded perceptual profile integrated data list, obtained by expanding the tag data based on the filtered perceptual profile integrated data list, includes: Obtain the current raw depth packet detection information; The target sector data is calculated based on the current original depth packet detection information; The terminal profile is obtained by querying the current original deep packet inspection information; Based on the current original deep packet inspection information, obtain the mobile phone number to be processed, the WeChat identifier to be processed, or the broadband account to be processed. The target WeChat identifier is obtained by querying the mobile phone number to be processed or the broadband account to be processed. The target broadband account is obtained by querying the mobile phone number to be processed and the WeChat identifier to be processed. The target sector data, the terminal profile, the target WeChat identifier, and the target broadband account are filled into the filtered perception profile integrated data list to obtain the expanded perception profile integrated data list.

6. The method according to claim 1, characterized in that, The adjustment of the current satisfaction prediction score based on the hierarchical veto method includes: Obtain the wide table data of application perception features for the preset time period before the current application prediction node; The dynamic threshold of the target user is determined based on the wide table data of the perceptual features to be applied, and the dynamic threshold includes a first threshold, a second threshold, or a third threshold. If there is data to be processed in the current perceptual feature wide table data that is less than the dynamic threshold, a preset score is obtained as the target deduction value; otherwise, a dynamic weight is calculated based on the current perceptual feature wide table data, and a target deduction value is calculated based on the dynamic weight and the preset score. The current satisfaction prediction score is adjusted based on the target deduction value.

7. The method according to claim 6, characterized in that, The step of calculating dynamic weights based on the current perceptual feature wide table data, and calculating the target deduction value based on the dynamic weights and the preset score, includes: Calculate the network quality coefficient based on the current perceptual feature wide table data; The dynamic weights are obtained by adjusting the base weights based on the network quality coefficients. The target deduction value for the corresponding category is calculated based on the dynamic weight and the preset score.

8. A quantitative scoring and prediction device for mobile network satisfaction, characterized in that, The device includes: The building module is used to construct the current perception feature wide table data based on the current operation domain data and the current business domain data; The category grouping module is used to group all target users into categories based on the current user perception feature wide table data combined with the current user package information, current user traffic information and current user call frequency information, to obtain several current user subgroups under each category; The filtering module is used to filter the current data subset corresponding to each current user subgroup from the current perception feature wide table data; The prediction module is used to input each of the current data subsets into the pre-trained satisfaction prediction model to predict the score and obtain the current satisfaction prediction sub-score. The regression processing module is used to perform regression processing on the current satisfaction prediction sub-score of each target user under all categories to obtain the current satisfaction prediction score of the target user. The adjustment module is used to adjust the current satisfaction prediction score based on the hierarchical veto method to obtain the target satisfaction score of the target user.

9. An electronic device, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images