Methods, apparatus, equipment, and computer storage media for determining user network quality.

Abstract

This application discloses a method, apparatus, device, and computer storage medium for determining user network quality. This disclosure acquires user communication parameter data within a target time period, trains on historical user communication parameter data, and obtains a network quality determination model based on the correspondence between historical traffic data and tag data. Based on date information and time sequence, it acquires the target similarity of traffic data within the same time period, thereby determining the periodicity of traffic data within the same time period. The target similarity is then compared with a first similarity threshold to obtain a comparison result. The target dataset corresponding to the comparison result is determined and input into the network quality determination model to determine the target prediction result. This disclosure, based on the pre-set correspondence information between historical traffic data and prediction results in the network quality determination model, determines the target prediction result corresponding to the target dataset, improving the accuracy of predicting users with poor network quality.

Description

Technical Field

[0001] This disclosure relates to the field of communication technology, and in particular to a method, apparatus, device, computer storage medium, and computer program product for determining user network quality. Background Technology

[0002] With the widespread use of the internet, more and more users are obtaining information or engaging in other activities online. However, due to various objective factors, slow internet speeds are a frequent occurrence.

[0003] To provide timely support services to users with slow internet speeds, related technologies typically use threshold settings to predict users with poor network quality (e.g., slow internet speeds) and then implement support measures accordingly. However, these threshold settings rely excessively on the experience of administrators, resulting in low accuracy in predicting users with poor network quality. Summary of the Invention

[0004] This disclosure provides a method, apparatus, device, computer storage medium, and computer program product for determining user network quality, which can improve the accuracy of predicting users with poor network quality.

[0005] In a first aspect, embodiments of this disclosure provide a method for determining user network quality, the method comprising:

[0006] Obtain historical user communication parameter data, which includes historical traffic data and tag data;

[0007] A network quality determination model is obtained by training on historical user communication parameter data. The network quality determination model includes the correspondence information between historical traffic data and tag data.

[0008] Acquire user communication parameter data within the target time period, including traffic data;

[0009] Based on date information and time order, obtain the target similarity of traffic data within the same time period;

[0010] The target similarity is compared with the first similarity threshold to obtain the comparison result;

[0011] Determine the target dataset corresponding to the comparison results;

[0012] Input the target dataset into the network quality determination model to determine the target prediction results;

[0013] The process involves training a network quality determination model using historical user communication parameter data. This includes: processing the historical user communication parameter data according to time sequence or the size of historical traffic data to obtain a first dataset containing historical traffic data; dividing the first dataset into a training set and a test set based on a preset partitioning ratio, with the training set including historical traffic data and the test set including label data, respectively; training the model based on the training set to obtain multiple network quality determination models, each including the correspondence information between historical traffic data and label data; inputting the historical traffic data from the test set into the multiple network quality determination models to obtain a first prediction result; comparing the label data from the test set with the first prediction result to obtain a comparison result; for each network quality determination model, counting the number of first prediction results in the comparison result that match the label data from the test set; and obtaining a relatively large number of network quality determination models.

[0014] In one feasible implementation, based on date information and time order, the target similarity of traffic data within the same time period is obtained, including:

[0015] Based on date information, traffic data within the same time period is divided into sub-traffic data;

[0016] The sub-flow data are sorted according to time order to obtain the first target sequence;

[0017] Based on multiple first target sequences with the same date information, the similarity between any two first target sequences with the same date information is calculated using a preset similarity calculation method, and a similarity set of the first target sequences is obtained.

[0018] Based on the data information of the similarity set of the first target sequence, the average value of the similarity set of the first target sequence is calculated, and the average value is determined as the target similarity of the traffic data.

[0019] In one feasible implementation, determining the target dataset corresponding to the comparison result includes:

[0020] If the target similarity is greater than the first similarity threshold, obtain the traffic data that is greater than the first preset threshold of traffic data within the same time period to obtain the target dataset;

[0021] If the target similarity is less than or equal to the first similarity threshold, a preset number of traffic data points are randomly obtained from the traffic data within the same time period to obtain the target dataset.

[0022] In one feasible implementation, when the target similarity is greater than a first similarity threshold, traffic data exceeding a first preset threshold within the same time period is acquired to obtain the target dataset, including:

[0023] If the target similarity is greater than the first similarity threshold, the median of the traffic data of the first target sequence is calculated based on multiple sub-traffic data of the first target sequence, and the median is used as the first preset threshold.

[0024] By comparing the sub-flow data with the median of the flow data in the first target sequence, the comparison result of the second target is obtained;

[0025] If the second target comparison result indicates that the sub-flow data is greater than the median of the flow data in the first target sequence, the sub-flow data is grouped to obtain the grouped sub-flow data.

[0026] The sub-traffic data after grouping is filtered according to a preset duration threshold to obtain the filtered sub-traffic data.

[0027] Based on the first target duration corresponding to the filtered sub-traffic data, obtain the number of sub-traffic data corresponding to the first target duration;

[0028] The target dataset is obtained from multiple sub-flow data.

[0029] In one feasible implementation, the method further includes:

[0030] If the second target comparison result indicates that the sub-traffic data is less than or equal to the median of the traffic data in the first target sequence, the corresponding number of first target user sub-traffic data within the first target duration is obtained as the target dataset.

[0031] In one feasible implementation, the target dataset is input into the network quality determination model to determine the target prediction result, including:

[0032] The traffic data in the target dataset is input into the network quality determination model to obtain multiple initial prediction results corresponding to the traffic data. The initial prediction results include the initial prediction target result and the initial prediction non-target result.

[0033] The initial prediction ratio is obtained by dividing the number of multiple initial prediction target results by the number of traffic data in the target dataset;

[0034] The initial prediction ratio is compared with the preset prediction threshold to obtain the target comparison result;

[0035] If the target comparison result indicates that the initial prediction ratio is greater than the preset prediction threshold, the target prediction result corresponding to the target dataset is determined.

[0036] In one feasible implementation, users include target users and non-target users, and historical user communication parameter data includes historical target user traffic data and historical non-target user traffic data. The historical user communication parameter data is processed according to time sequence or the size of the traffic data to obtain a first dataset including traffic data, comprising:

[0037] Based on the target filtering threshold, historical target user traffic data is filtered to obtain the first historical target user traffic data that is greater than the target filtering threshold; based on the target filtering threshold, historical non-target user traffic data is filtered to obtain the first historical non-target user traffic data that is greater than the target filtering threshold.

[0038] Obtain the first similarity of the first historical target user traffic data; obtain the second similarity of the first historical non-target user traffic data;

[0039] Compare the first similarity with a preset similarity threshold to obtain a first comparison result; compare the second similarity with a preset similarity threshold to obtain a second comparison result;

[0040] If the first comparison result indicates that the first similarity is greater than the preset similarity threshold, the first historical target user traffic data with larger traffic data is obtained according to the size of the first historical target user traffic data, and a second dataset is obtained; if the second comparison result indicates that the second similarity is greater than the preset similarity threshold, the first historical non-target user traffic data with larger traffic data is obtained according to the size of the first historical non-target user traffic data, and a third dataset is obtained.

[0041] The second and third datasets are combined to obtain the first dataset.

[0042] In one feasible implementation, the first historical target user traffic data includes first historical target user sub-traffic data, and the first historical non-target user traffic data includes first historical non-target user sub-traffic data. A first similarity of the first historical target user traffic data is obtained; a second similarity of the first historical non-target user traffic data is obtained, including:

[0043] Based on date information, the traffic data of the first historical target user is divided into first historical target user sub-traffic data; based on date information, the traffic data of the first historical non-target user is divided into first historical non-target user sub-traffic data.

[0044] Sort the first historical target user sub-traffic data in chronological order to obtain the first sequence; sort the first historical non-target user sub-traffic data in chronological order to obtain the second sequence.

[0045] Based on multiple first sequences with identical date information, the similarity between any two first sequences with identical date information is calculated using a preset similarity calculation method, resulting in a similarity set for the first sequences; based on multiple second sequences with identical date information, the similarity between any two second sequences with identical date information is calculated using the preset similarity calculation method, resulting in a similarity set for the second sequences.

[0046] Based on the data information of the similarity set of the first sequence, the median and mean of the similarity set of the first sequence are calculated; based on the data information of the similarity set of the second sequence, the median and mean of the similarity set of the second sequence are calculated.

[0047] The third comparison result is obtained by comparing the mean of the similarity set of the first sequence with the median of the similarity set of the second sequence; the fourth comparison result is obtained by comparing the mean of the similarity set of the second sequence with the median of the similarity set of the second sequence.

[0048] If the third comparison result indicates that the average value of the similarity set of the first sequence is greater than the median of the similarity set of the first sequence, the average value of the similarity set of the first sequence is obtained as the first similarity of the first historical target user traffic data; if the fourth comparison result indicates that the average value of the similarity set of the second sequence is greater than the median of the similarity set of the second sequence, the average value of the similarity set of the second sequence is obtained as the second similarity of the first historical non-target user traffic data.

[0049] In one feasible implementation, if the first comparison result indicates that the first similarity is greater than a preset similarity threshold, then based on the size of the first historical target user traffic data, the first historical target user traffic data with larger traffic data is obtained to obtain a second dataset; if the second comparison result indicates that the second similarity is greater than a preset similarity threshold, then based on the size of the first historical non-target user traffic data, the first historical non-target user traffic data with larger traffic data is obtained to obtain a third dataset, including:

[0050] The median of the traffic data for the first sequence is calculated based on multiple first historical target user sub-traffic data for the first sequence; the median of the traffic data for the second sequence is calculated based on the first historical non-target user sub-traffic data for the second sequence.

[0051] The fifth comparison result is obtained by comparing the median of the traffic data of the first historical target user sub-data with the median of the traffic data of the first sequence; the sixth comparison result is obtained by comparing the median of the traffic data of the first historical non-target user sub-data with the median of the traffic data of the second sequence.

[0052] If the fifth comparison result indicates that the first historical target user sub-traffic data is greater than the median of the first sequence of traffic data, the first historical target user sub-traffic data is grouped to obtain the grouped first historical target user sub-traffic data; if the sixth comparison result indicates that the first historical non-target user sub-traffic data is greater than the median of the second sequence of traffic data, the first historical non-target user sub-traffic data is grouped to obtain the grouped first historical non-target user sub-traffic data.

[0053] The first historical target user sub-traffic data after grouping is filtered according to a preset duration threshold to obtain the filtered first historical target user sub-traffic data; the first historical non-target user sub-traffic data after grouping is filtered according to a preset duration threshold to obtain the filtered first historical non-target user sub-traffic data.

[0054] Based on the first duration corresponding to the filtered first historical target user sub-traffic data, obtain the first historical target user sub-traffic data of a quantity corresponding to the first duration; based on the second duration corresponding to the filtered first historical non-target user sub-traffic data, obtain the first historical non-target user sub-traffic data of a quantity corresponding to the second duration.

[0055] A second dataset is obtained based on multiple first historical target user sub-traffic data; a third dataset is obtained based on multiple first historical non-target user sub-traffic data.

[0056] In one feasible implementation, the method further includes:

[0057] If the fifth comparison result indicates that the first historical target user sub-traffic data is less than or equal to the median of the first sequence of traffic data, the corresponding number of first historical target user sub-traffic data within the third time period is obtained as the second dataset; if the sixth comparison result indicates that the first historical non-target user sub-traffic data is less than or equal to the median of the second sequence of traffic data, the corresponding number of first historical non-target user sub-traffic data within the third time period is obtained as the third dataset.

[0058] In one feasible implementation, after comparing a first similarity with a preset similarity threshold to obtain a first comparison result, and comparing a second similarity with a preset similarity threshold to obtain a second comparison result, the method further includes:

[0059] If the first comparison result indicates that the first similarity is less than or equal to a preset similarity threshold, the first historical target user traffic data is divided according to the date information to obtain the first historical target user sub-traffic data; if the second comparison result indicates that the second similarity is less than or equal to a preset similarity threshold, the first historical non-target user traffic data is divided according to the date information to obtain the first historical non-target user sub-traffic data.

[0060] Obtain the first historical target user sub-traffic data corresponding to the fourth time period as the second dataset; obtain the first historical non-target user sub-traffic data corresponding to the fourth time period as the third dataset.

[0061] In one feasible implementation, the first dataset is divided according to a preset partitioning ratio to obtain a training set and a test set, including:

[0062] The fourth dataset is obtained by arbitrarily sorting the data information of the first dataset.

[0063] The non-empty data in the fourth dataset are filtered out to obtain the fifth dataset;

[0064] The fifth dataset is divided into training and test sets based on a preset partitioning ratio.

[0065] In one feasible implementation, the users include target users and non-target users, and the method further includes, before obtaining historical user communication parameter data:

[0066] Based on the complaint information sent by the target user, obtain the target user's account list information;

[0067] Obtain a list of accounts belonging to non-target users;

[0068] Retrieve historical user communication parameter data, including:

[0069] Based on the account list information of the target users, obtain historical communication parameter data of the target users; based on the account list information of the non-target users, obtain historical communication parameter data of the non-target users.

[0070] Secondly, embodiments of this disclosure provide an apparatus for determining user network quality, the apparatus comprising:

[0071] The acquisition module is used to acquire historical user communication parameter data, which includes historical traffic data and tag data.

[0072] The training module is used to train on historical user communication parameter data to obtain a network quality determination model. The network quality determination model includes the correspondence information between historical traffic data and tag data.

[0073] The acquisition module is also used to acquire user communication parameter data within a target time period, including traffic data.

[0074] The acquisition module is also used to obtain the target similarity of traffic data within the same time period based on date information and time order;

[0075] The comparison module is used to compare the target similarity with a first similarity threshold and obtain the comparison result;

[0076] The determination module is used to determine the target dataset corresponding to the comparison results;

[0077] The input module is used to input the target dataset into the network quality determination model to determine the target prediction result;

[0078] The process involves training a network quality determination model using historical user communication parameter data. This includes: processing the historical user communication parameter data according to time sequence or the size of historical traffic data to obtain a first dataset containing historical traffic data; dividing the first dataset into a training set and a test set based on a preset partitioning ratio, with the training set including historical traffic data and the test set including label data, respectively; training the model based on the training set to obtain multiple network quality determination models, each including the correspondence information between historical traffic data and label data; inputting the historical traffic data from the test set into the multiple network quality determination models to obtain a first prediction result; comparing the label data from the test set with the first prediction result to obtain a comparison result; for each network quality determination model, counting the number of first prediction results in the comparison result that match the label data from the test set; and obtaining a relatively large number of network quality determination models.

[0079] Thirdly, embodiments of this disclosure provide a device for determining user network quality. The device includes a processor and a memory storing computer program instructions. The processor reads and executes the computer program instructions to implement the method for determining user network quality as described in any of the first aspects.

[0080] Fourthly, embodiments of this disclosure provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement a method for determining user network quality as described in any of the first aspects.

[0081] Fifthly, embodiments of this disclosure provide a computer program product, including a computer program that, when executed by a processor, implements a method for determining user network quality as described in any of the first aspects.

[0082] The network quality prediction method, apparatus, device, computer storage medium, and computer program product provided in this disclosure acquire user communication parameter data within a target time period, train on historical user communication parameter data, and obtain a network quality determination model based on the correspondence between historical traffic data and tag data. Based on date information and time sequence, target similarity of traffic data within the same time period is obtained, thereby determining the periodicity of traffic data within the same time period. The target similarity is then compared with a first similarity threshold to obtain a comparison result. The target dataset corresponding to the comparison result is determined and input into the network quality determination model to determine the target prediction result. This disclosure improves the accuracy of predicting users with poor network quality by determining the target prediction result corresponding to the target dataset based on preset correspondence information between historical traffic data and prediction results. Attached Figure Description

[0083] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments of the present invention will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0084] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments of the present invention will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0085] Figure 1 This is a schematic diagram of a method for determining user network quality provided in an embodiment of this disclosure;

[0086] Figure 2 This is a flowchart illustrating a method for determining user network quality provided in an embodiment of this disclosure;

[0087] Figure 3 This is a schematic diagram of a training network quality determination model provided in an embodiment of this disclosure;

[0088] Figure 4 This is a schematic diagram illustrating an embodiment of the present disclosure for obtaining sub-traffic data of multiple first historical target users;

[0089] Figure 5 This is a schematic diagram of the structure of a device for determining user network quality provided in an embodiment of this disclosure;

[0090] Figure 6 This is a schematic diagram of the structure of a device for determining user network quality provided in an embodiment of this disclosure. Detailed Implementation

[0091] The features and exemplary embodiments of various aspects of the present invention will now be described in detail. To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely intended to explain the present invention and not to limit the present invention. For those skilled in the art, the present invention can be practiced without some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present invention by illustrating examples of the invention.

[0092] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0093] Before describing the technical solutions provided by the embodiments of this disclosure, in order to facilitate understanding of the embodiments of this disclosure, this disclosure will specifically explain the problems existing in the related technologies:

[0094] Currently, an increasing number of users are using the internet for activities such as playing games and watching videos to meet their needs. However, due to objective factors such as physical latency, users' internet quality, such as internet speed, is often not high. To provide timely services to users with poor internet quality, related technologies employ thresholding to predict such users; or they collect daily-level user data and use machine learning algorithms to predict users with poor internet quality.

[0095] However, the threshold-based approach used in these technologies heavily relies on the business experience of administrators and fails to consider the varying network demands arising from different internet browsing behaviors. Furthermore, the method of collecting daily user data and using machine learning algorithms to predict internet quality, because it analyzes data from the entire day, neglects the objective fact that users are unlikely to be online all day, resulting in most of the data metrics used being from invalid time periods. In conclusion, the accuracy of these technologies in predicting users with poor network quality is not high.

[0096] In view of the inventors’ above-mentioned research findings, this disclosure provides a method, apparatus, device, computer storage medium and computer program product for determining user network quality.

[0097] The technical concept of this disclosure is to: reasonably collect user network quality-related data and use machine learning algorithms to predict it, thereby predicting the user's internet access quality. In one embodiment, the steps are as follows:

[0098] Identify users with fast and slow internet speeds. Since determining internet speed based on thresholds has significant drawbacks, identifying slow-connecting users can be achieved by gathering complaints from them. Similarly, since users receiving VIP (Very Important Person) services generally have higher internet quality, VIP users can be considered as having fast internet speeds.

[0099] Then, the performance data of users with slow internet speeds and those with fast internet speeds are analyzed based on their internet usage time periods to determine if user internet usage is periodic, while minimizing the impact of special dates such as holidays on data collection. By determining the periodicity of user internet usage, sampled traffic data is obtained by sampling the performance data of users with slow internet speeds and the performance data of users with fast internet speeds.

[0100] The sampled traffic data of users with slow internet speeds are then merged with those of users with fast internet speeds to form a dataset, which is used to train and adjust the parameters of machine learning algorithms such as extreme gradient boosting (XGBoost). Finally, the optimally optimized XGBoost algorithm is used to obtain real-time user communication parameter data to predict user network quality. Therefore, the accuracy of predicting users with poor network quality is improved.

[0101] In one example, such as Figure 1 As shown, a list of slow-connection user accounts can be determined based on historical complaint data, and then historical performance data for these users can be derived from this list. Simultaneously, historical performance data for fast-connection users can be determined from the list of fast-connection user accounts. Further analysis of user internet usage time periods is then performed on both sets of historical performance data, resulting in separate datasets for fast and slow-connection users. A training dataset is then constructed based on these datasets. The datasets are then trained, and a model is trained using these datasets to obtain a model file. Finally, the model file and real-time user internet performance data are used for prediction, ultimately yielding a list of users with slow internet connections.

[0102] To address the problems of the prior art, embodiments of this disclosure provide a method, apparatus, device, computer storage medium, and computer program product for determining user network quality.

[0103] The method for determining user network quality provided in the embodiments of this disclosure will be introduced first.

[0104] Figure 2 A flowchart illustrating a method for determining user network quality according to an embodiment of this disclosure is shown. Figure 2 As shown, the method may include the following steps S110-S170.

[0105] S110: Obtain historical user communication parameter data.

[0106] Among them, user communication parameter data is data characterizing the user's network quality. Historical user communication parameter data includes historical traffic data and tag data.

[0107] In one example, the target time period could be from 0:00 to 24:00 on Tuesday, or from 0:00 to 24:00 on Thursday.

[0108] In one example, user communication parameter data may include: gateway model, gateway manufacturer, Wireless Fidelity (WIFI) mode, duty cycle of current channel interference signal, gateway runtime, Central Processing Unit (CPU) utilization, memory utilization, main chip temperature, downlink traffic, uplink traffic, average uplink traffic per cycle, average downlink traffic per cycle, peak uplink traffic per cycle, peak downlink traffic per cycle, number of connected devices, Wireless Local Area Network (WLAN) negotiation receive rate, WLAN negotiation transmit rate, WLAN signal strength, WLAN access type, number of surrounding WIFI networks, number of downlink Transmission Control Protocol (TCP) connections, number of uplink TCP connections, number of successful TCP connections, average interval between the first and second handshakes, average interval between the second and third handshakes, downlink TCP retransmission rate, uplink TCP retransmission rate, downlink Time To Live (TTL) information, uplink TTL information, Domain Name Server (DNS) request time, DNS response time, Hypertext Transfer Protocol (HTP) data. HTTP (HTTP) response latency, HTTP error codes, HTTP download speed, number of HTTP requests, and number of failed HTTP requests.

[0109] In one embodiment, user communication parameters can be acquired at fixed intervals, such as acquiring user communication parameter data every 10 minutes.

[0110] S120: Train the network quality determination model using historical user communication parameter data.

[0111] After acquiring historical user communication parameter data, the network quality determination model is trained using this data. This model includes the correspondence between historical traffic data and label data.

[0112] In one embodiment, the step of training a network quality determination model using historical user communication parameter data includes: processing the historical user communication parameter data according to time sequence or the size of historical traffic data to obtain a first dataset including historical traffic data; dividing the first dataset according to a preset partitioning ratio to obtain a training set and a test set, wherein the training set and the test set respectively include historical traffic data and label data; training the model based on the training set to obtain multiple network quality determination models, each network quality determination model including the correspondence information between historical traffic data and label data; inputting the historical traffic data in the test set into the multiple network quality determination models to obtain a first prediction result; comparing the label data in the test set with the first prediction result to obtain a comparison result; for each network quality determination model, counting the number of first prediction results in the comparison result that are consistent with the label data in the test set; and obtaining a relatively large number of network quality determination models.

[0113] S130: Obtain user communication parameter data within the target time period.

[0114] After obtaining the network quality determination model, user communication parameter data for the target time period is acquired. This user communication parameter data includes traffic data.

[0115] S140: Based on date information and time order, obtain the target similarity of traffic data within the same time period.

[0116] After acquiring user communication parameter data for the target time period, the target similarity of traffic data within the same time period is obtained based on date information and time order. Target similarity characterizes the regularity of users' traffic usage data within the same time period.

[0117] In one example, the date information could be "Tuesday".

[0118] In one example, the traffic data from 9:00 to 12:00 on the first Friday of month P is denoted as P1, and the traffic data from 9:00 to 12:00 on the second Friday of month P is denoted as P2. Then, the target similarity of the traffic data from 9:00 to 12:00 on the first Friday of month P and the second Friday of month P can be calculated.

[0119] S150: Compare the target similarity with the first similarity threshold to obtain the comparison result.

[0120] In this embodiment of the disclosure, the first similarity threshold represents the threshold for judging the regularity of a user's traffic data usage in different time periods, and the first similarity threshold can be modified as needed.

[0121] S160: Determine the target dataset corresponding to the comparison results.

[0122] After obtaining the comparison results, the corresponding target dataset is generated based on the comparison results.

[0123] S170: Input the target dataset into the network quality determination model to determine the target prediction results.

[0124] After obtaining the target dataset, inputting the target dataset into the network quality determination model will yield the target prediction results.

[0125] In one example, if the user traffic data in the target dataset is {X, Y, Z}, multiple prediction results can be obtained based on the correspondence between the traffic data and the prediction results. The user is then identified as having slow internet access based on the information from these multiple prediction results.

[0126] This embodiment of the disclosure acquires user communication parameter data within a target time period, trains on historical user communication parameter data, and obtains a network quality determination model based on the correspondence between historical traffic data and tag data. Based on date information and time order, it acquires the target similarity of traffic data within the same time period, thereby determining the periodicity of traffic data within the same time period. The target similarity is then compared with a first similarity threshold to obtain a comparison result. The target dataset corresponding to the comparison result is determined and input into the network quality determination model to determine the target prediction result. This embodiment of the disclosure determines the target prediction result corresponding to the target dataset based on preset correspondence information between historical traffic data and prediction results, improving the accuracy of predicting users with poor network quality.

[0127] In one embodiment, step S140, which obtains the target similarity of traffic data within the same time period based on date information and time order, may include steps S141-S144.

[0128] S141: Divide the traffic data within the same time period according to the date information to obtain sub-traffic data.

[0129] In this embodiment of the disclosure, the date information represents the weekday attribute of the day, and the division method is based on the weekday attribute of the date.

[0130] In one example, traffic data for the first Wednesday of month K can be retrieved as sub-traffic data based on the date attribute information.

[0131] S142: Sort the sub-flow data according to time order to obtain the first target sequence.

[0132] In this embodiment of the disclosure, the first target sequence includes multiple sub-flow data, and the number of sub-flow data is not fixed.

[0133] In one example, the traffic data for the first Wednesday of month K is processed from 0:00 to 24:00 in hourly intervals to obtain the first target sequence.

[0134] S143: Based on multiple first target sequences with the same date information, calculate the similarity between any two first target sequences with the same date information using a preset similarity calculation method, and obtain the similarity set of the first target sequences.

[0135] In one example, the default similarity calculation method could be Dynamic Time Warping (DTW).

[0136] In one example, the DTW method can be used to calculate the similarity between the first target sequence of the first Wednesday of month K and the first target sequences of other Wednesdays of the same month or other months, thus obtaining a set of similarity scores for the first target sequences.

[0137] S144: Based on the data information of the similarity set of the first target sequence, calculate the average value of the similarity set of the first target sequence, and determine the average value as the target similarity of the traffic data.

[0138] After obtaining the data information of the similarity set of the first target sequence, the average value of the similarity set of the first target sequence is calculated, and the average value of the similarity set of the first target sequence is determined as the target similarity of the traffic data.

[0139] In this embodiment of the disclosure, traffic data is divided by date information, which can determine traffic data with the same date information as needed, and sort them according to time order to obtain the first target sequence. Finally, the similarity of different first target sequences can be effectively obtained by using a preset similarity calculation method.

[0140] In one embodiment, step S160, where the target similarity is greater than a first similarity threshold, involves acquiring traffic data that exceeds a first preset threshold within the same time period to obtain the target dataset, and may include steps S161 and S162.

[0141] S161: If the target similarity is greater than the first similarity threshold, obtain the traffic data that is greater than the first preset threshold of traffic data within the same time period to obtain the target dataset.

[0142] If the target similarity is greater than the first similarity threshold, then the traffic data that is greater than the first preset threshold of traffic data within the same time period is obtained to obtain the target dataset.

[0143] S162: If the target similarity is less than or equal to the first similarity threshold, arbitrarily obtain a preset number of traffic data from the traffic data within the same time period to obtain the target dataset.

[0144] If the target similarity is less than or equal to the first similarity threshold, then a preset number of traffic data points are randomly obtained from the traffic data within the same time period to obtain the target dataset.

[0145] The embodiments of this disclosure determine the corresponding method for obtaining the target dataset based on the comparison result between the target similarity and the first similarity threshold, which can effectively determine the target dataset.

[0146] In one embodiment, step S161, when the target similarity is greater than a first similarity threshold, acquires traffic data that is greater than a first preset threshold of traffic data within the same time period to obtain the target dataset, and may include steps S1611-S1616.

[0147] S1611: When the target similarity is greater than the first similarity threshold, the median of the traffic data of the first target sequence is calculated based on multiple sub-traffic data of the first target sequence, and the median is used as the first preset threshold.

[0148] In one example, there are multiple samples of sub-flow data in the first target sequence K1{1,2,2,2,3,4}, where 2 is the median of the flow data in K1. Then, 2 is used as the first preset threshold for the flow data.

[0149] S1612: Compare the sub-flow data with the median of the flow data of the first target sequence to obtain the comparison result of the second target.

[0150] In one example, each sub-flow data in the first target sequence K1 can be compared with the median of the flow data in K1, 2. If the sub-flow data in K1 is greater than 2, then step S143 is executed.

[0151] S1613: If the second target comparison result indicates that the sub-flow data is greater than the median of the flow data in the first target sequence, the sub-flow data is grouped to obtain the grouped sub-flow data.

[0152] In one example, if the sub-flow data in the first target sequence K1 is greater than the median of the flow data (2), then the sub-flow data greater than 2 are grouped together.

[0153] S1614: Filter the grouped sub-traffic data according to the preset duration threshold to obtain the filtered sub-traffic data.

[0154] In one example, if the duration of the sub-traffic data after grouping does not exceed 1 hour, then the sub-traffic data will be filtered.

[0155] S1615: Based on the first target duration corresponding to the filtered sub-traffic data, obtain the number of sub-traffic data corresponding to the first target duration.

[0156] In one example, if the duration of the sub-traffic data after grouping exceeds 1 hour, then the number of sub-traffic data corresponding to the first target duration is obtained.

[0157] S1616: Obtain the target dataset based on multiple sub-flow data.

[0158] In this embodiment of the disclosure, by comparing multiple sub-flow data of the first target sequence with the median of the flow data of the first target sequence, the grouped sub-flow data can be obtained, and the target dataset can be obtained based on the grouped sub-flow data.

[0159] In one embodiment, the method for determining user network quality further includes step S510:

[0160] S510: If the sub-traffic data represented by the second target comparison result is less than or equal to the median of the traffic data of the first target sequence, obtain the corresponding number of first target user sub-traffic data within the first target duration as the target dataset.

[0161] In one embodiment, if the sub-traffic data of the first target sequence is less than or equal to the median of the traffic data of the first target sequence, then the corresponding number of first target user sub-traffic data within the first target duration can be obtained as the target dataset in a random manner.

[0162] In this embodiment of the disclosure, the first target user sub-traffic data corresponding to a first target time period is obtained in a random manner as the target dataset, thereby enriching the ways of obtaining the target dataset according to different situations.

[0163] In one embodiment, step S170 inputs the target dataset into the network quality determination model, and determining the target prediction result may include steps S171-S174.

[0164] S171: Input the traffic data in the target dataset into the network quality determination model to obtain multiple initial prediction results corresponding to the traffic data.

[0165] The initial prediction results include the initial target prediction results and the initial non-target prediction results.

[0166] In one example, the initial predicted target result represents the users that business personnel need to care about, while the initial predicted non-target result represents the users that business personnel do not need to care about.

[0167] In one example, the target dataset includes 7 traffic data points. Inputting these 7 traffic data points into the model yields 7 initial predictions, including predictions that the user's internet connection is slow and predictions that the user's internet connection is fast.

[0168] S172: Divide the number of initial prediction target results by the number of traffic data in the target dataset to obtain the initial prediction ratio.

[0169] In one example, the initial prediction ratio represents the percentage of users whose network quality is poor based on initial predictions.

[0170] In another example, if the number of initial prediction target results is O and the number of traffic data in the target dataset is U, then the initial prediction ratio is O / U.

[0171] S173: Compare the initial prediction ratio with the preset prediction threshold to obtain the target comparison result.

[0172] After obtaining the initial prediction ratio, the initial prediction ratio is compared with the preset prediction threshold to obtain the target comparison result.

[0173] S174: If the target comparison result indicates that the initial prediction ratio is greater than the preset prediction threshold, determine the target prediction result corresponding to the target dataset.

[0174] The preset prediction threshold is used to determine the target prediction result. The value of the preset prediction threshold is not fixed and can be adjusted according to needs.

[0175] In one example, if the initial prediction ratio is 80% and the preset prediction threshold is 75%, then the target prediction result for the target dataset is determined to be that the user's internet speed is slow.

[0176] In this embodiment, by inputting traffic data from a target dataset into a model, multiple initial prediction results corresponding to the traffic data are obtained. Then, the initial prediction ratio is obtained and compared with a preset prediction threshold, which can improve the accuracy of predicting users with poor network quality.

[0177] In one embodiment, such as Figure 3 As shown, step S120 trains the network quality determination model using historical user communication parameter data, and also includes steps S121-S125.

[0178] S121: Filter historical target user traffic data according to the target filtering threshold to obtain first historical target user traffic data that is greater than the target filtering threshold; filter historical non-target user traffic data according to the target filtering threshold to obtain first historical non-target user traffic data that is greater than the target filtering threshold.

[0179] The target filtering threshold is not fixed and can be modified according to requirements. Target users represent users that business personnel need to care about, while non-target users represent users that business personnel do not need to care about.

[0180] This embodiment of the disclosure uses a target filtering threshold to filter historical target user traffic data and historical non-target user traffic data respectively, so that the filtered data is representative.

[0181] S122: Obtain the first similarity of the first historical target user traffic data; obtain the second similarity of the first historical non-target user traffic data.

[0182] In one example, the first similarity represents a parameter of the periodicity of first historical target user traffic data generated on different dates with the same date attribute, and the second similarity represents a parameter of the periodicity of first historical non-target user traffic data generated on different dates with the same date attribute.

[0183] In one example, if Y2 of month N2 is a Monday and U2 of month M2 is also a Monday, then Y2 of month N2 and U2 of month M2 can be considered to have the same date attribute, namely Monday. Specifically, O2 traffic data is generated in the morning of Y2 of month N2, but no traffic data is generated in the afternoon and evening. Similarly, O2 traffic data is generated in the morning of U2 of month M2, but no traffic data is generated in the afternoon and evening. Therefore, Y2 of month N2 and U2 of month M2 have a high degree of similarity.

[0184] S123: Compare the first similarity with the preset similarity threshold to obtain the first comparison result; compare the second similarity with the preset similarity threshold to obtain the second comparison result.

[0185] The preset similarity threshold value is not fixed and can be modified according to requirements.

[0186] S124: If the first comparison result indicates that the first similarity is greater than the preset similarity threshold, the first historical target user traffic data with larger traffic data is obtained according to the size of the first historical target user traffic data, and a second dataset is obtained; if the second comparison result indicates that the second similarity is greater than the preset similarity threshold, the first historical non-target user traffic data with larger traffic data is obtained according to the size of the first historical non-target user traffic data, and a third dataset is obtained.

[0187] The sampling method is not fixed.

[0188] In one example, the sampling operation can be performed according to a preset sampling method.

[0189] In one embodiment, if the first comparison result indicates that the first similarity is greater than a preset similarity threshold, meaning that the first historical target user traffic data generated on different dates with the same date attribute is periodic, then the larger traffic data in the first historical target user traffic data can be sampled and selected as the second dataset. If the second comparison result indicates that the second similarity is greater than a preset similarity threshold, meaning that the first historical non-target user traffic data generated on different dates with the same date attribute is periodic, then the larger traffic data in the first historical non-target user traffic data can be sampled and selected as the third dataset.

[0190] S125: Combine the second and third datasets to obtain the first dataset.

[0191] This embodiment of the disclosure obtains a first dataset by sampling and merging first historical target user traffic data and first historical non-target user traffic data with a similarity greater than a preset similarity threshold, so that the traffic data in the first dataset is representative.

[0192] In one embodiment, step S122 obtaining the first similarity of the first historical target user traffic data and obtaining the second similarity of the first historical non-target user traffic data may include steps S1221-S1226.

[0193] S1221: Divide the first historical target user traffic data according to the date information to obtain the first historical target user sub-traffic data; divide the first historical non-target user traffic data according to the date information to obtain the first historical non-target user sub-traffic data.

[0194] In one example, date information includes a date attribute.

[0195] In one embodiment, the first historical target user traffic data is divided according to date information to obtain multiple first historical target user traffic data on different dates with the same date attribute, i.e., first historical target user sub-traffic data. Similarly, the first historical non-target user traffic data is divided according to date information to obtain multiple first historical non-target user traffic data on different dates with the same date attribute, i.e., first historical non-target user sub-traffic data.

[0196] In one example, by dividing the historical target user traffic data according to date information, we can obtain historical target user traffic data for multiple Mondays.

[0197] S1222: Sort the first historical target user sub-traffic data in chronological order to obtain the first sequence; sort the first historical non-target user sub-traffic data in chronological order to obtain the second sequence.

[0198] In one example, the time order can be sorted by time increment, and the first sequence can include multiple sorted first historical target user sub-traffic data.

[0199] In another example, the traffic data of the first historical target user sub-traffic data, i.e., the traffic data of day G3, can be sorted from 00:00:00 to 23:59:59 in chronological order, and the sorting method is adopted by time point increment to obtain the first sequence.

[0200] S1223: Based on multiple first sequences with the same date information, calculate the similarity between any two first sequences with the same date information using a preset similarity calculation method to obtain a similarity set of the first sequences; based on multiple second sequences with the same date information, calculate the similarity between any two second sequences with the same date information using a preset similarity calculation method to obtain a similarity set of the second sequences.

[0201] In another example, based on multiple first sequences with identical date information, the steps for calculating the similarity between any two first sequences with identical date information using the DTW method to obtain the similarity set of the first sequences are as follows: There are F first sequences with identical date information. For any two first sequences, the similarity is calculated using the DTW method to obtain an F*F matrix. The upper triangular matrix in the F*F matrix is ​​used as the similarity set of the first sequences. Wherein, F... si The element represents the DTW similarity between the s-th time series and the i-th time series, where s is not equal to i.

[0202] S1224: Based on the data information of the similarity set of the first sequence, calculate the median and average of the similarity set of the first sequence; based on the data information of the similarity set of the second sequence, calculate the median and average of the similarity set of the second sequence.

[0203] In one example, the elements in the upper triangular matrix of an F*F matrix can be statistically analyzed to obtain the median and mean of the elements in the upper triangular matrix.

[0204] S1225: Compare the mean of the similarity set of the first sequence with the median of the similarity set of the first sequence to obtain the third comparison result; compare the mean of the similarity set of the second sequence with the median of the similarity set of the second sequence to obtain the fourth comparison result.

[0205] S1226: If the average value of the similarity set of the first sequence in the third comparison result is greater than the median of the similarity set of the first sequence, the average value of the similarity set of the first sequence is obtained as the first similarity of the first historical target user traffic data; if the average value of the similarity set of the second sequence in the fourth comparison result is greater than the median of the similarity set of the second sequence, the average value of the similarity set of the second sequence is obtained as the second similarity of the first historical non-target user traffic data.

[0206] In one embodiment, if the average of the similarity set of the first sequence, as represented by the third comparison result, is greater than the median of the similarity set of the first sequence, the similarity set of the first sequence can be considered to have periodicity as a whole, and the average of the similarity set of the first sequence is taken as the first similarity of the first historical target user traffic data. Similarly, if the average of the similarity set of the second sequence, as represented by the fourth comparison result, is greater than the median of the similarity set of the second sequence, the similarity set of the second sequence can be considered to have periodicity as a whole, and the average of the similarity set of the second sequence is taken as the second similarity of the first historical non-target user traffic data.

[0207] In one example, if the average of the similarity set of the first sequence is 0.9 and the median of the similarity set of the first sequence is 0.5, then the first similarity of the first historical target user traffic data is 0.9.

[0208] In this embodiment of the disclosure, by comparing the mean and median of the similarity sets of the first sequence and the second sequence, a first similarity and a second similarity with a high degree of similarity can be obtained.

[0209] In one embodiment, step S124, if the first comparison result indicates that the first similarity is greater than a preset similarity threshold, obtains the first historical target user traffic data with larger traffic data based on the size of the first historical target user traffic data, and obtains the second dataset; if the second comparison result indicates that the second similarity is greater than a preset similarity threshold, obtains the first historical non-target user traffic data with larger traffic data based on the size of the first historical non-target user traffic data, and obtains the third dataset, may include steps S1241-S1246.

[0210] S1241: Calculate the median of the traffic data for the first sequence based on multiple first historical target user sub-traffic data for the first sequence; calculate the median of the traffic data for the second sequence based on the first historical non-target user sub-traffic data for the second sequence.

[0211] In one example, if the traffic data of multiple first historical target users in the first sequence are as follows: traffic data at 10 o'clock is g, traffic data at 11 o'clock is t, and traffic data at 12 o'clock is t, then the median of the traffic data in the first sequence is t.

[0212] S1242: Compare the median of the first historical target user sub-traffic data and the median of the first sequence traffic data to obtain the fifth comparison result; compare the median of the first historical non-target user sub-traffic data and the median of the second sequence traffic data to obtain the sixth comparison result.

[0213] S1243: If the fifth comparison result indicates that the first historical target user sub-traffic data is greater than the median of the first sequence of traffic data, the first historical target user sub-traffic data is grouped to obtain the grouped first historical target user sub-traffic data; if the sixth comparison result indicates that the first historical non-target user sub-traffic data is greater than the median of the second sequence of traffic data, the first historical non-target user sub-traffic data is grouped to obtain the grouped first historical non-target user sub-traffic data.

[0214] In one embodiment, if the fifth comparison result indicates that the first historical target user sub-traffic data is greater than the median of the first sequence of traffic data, the first historical target user sub-traffic data that is greater than the median of the first sequence of traffic data and the first historical target user sub-traffic data that is less than or equal to the median of the first sequence of traffic data can be divided into two groups. If the sixth comparison result indicates that the first historical non-target user sub-traffic data is greater than the median of the second sequence of traffic data, the first historical non-target user sub-traffic data that is greater than the median of the second sequence of traffic data and the first historical non-target user sub-traffic data that is less than or equal to the median of the second sequence of traffic data can be divided into two groups.

[0215] S1244: Filter the grouped first historical target user sub-traffic data according to a preset duration threshold to obtain filtered first historical target user sub-traffic data; filter the grouped first historical non-target user sub-traffic data according to a preset duration threshold to obtain filtered first historical non-target user sub-traffic data.

[0216] The preset duration threshold is not fixed and can be adjusted according to needs.

[0217] In one example, the preset duration threshold could be 1 hour.

[0218] In one embodiment, a preset duration threshold can be used to filter the first historical target user sub-traffic data that is greater than the median of the first sequence of traffic data, resulting in filtered first historical target user sub-traffic data. Similarly, a preset duration threshold can be used to filter the first historical non-target user sub-traffic data that is greater than the median of the second sequence of traffic data, resulting in filtered first historical non-target user sub-traffic data.

[0219] S1245: Based on the first duration corresponding to the filtered first historical target user sub-traffic data, obtain the number of first historical target user sub-traffic data corresponding to the first duration; based on the second duration corresponding to the filtered first historical non-target user sub-traffic data, obtain the number of first historical non-target user sub-traffic data corresponding to the second duration.

[0220] In one example, the length of the first duration is not fixed; it can be 2 hours, 3 hours, or other durations. The length of the second duration is also not fixed; it can be 4 hours, 5 hours, or other durations.

[0221] In another example, such as Figure 4 As shown, the steps for obtaining multiple first historical target user sub-traffic data corresponding to a first duration are as follows: There exists a time series T, where the traffic data of time series T has a median. The traffic data of time series T is compared with the median, dividing it into groups Td and Tu. If the data Ti at any time point i is less than the median, it is used as data in group Td; otherwise, it is used as data in group Tu. At this point, the data in group Tu is divided into multiple groups of continuous data by the median: t1 to t2 is one group, t3 to t4 is another. It is determined whether the duration corresponding to the multiple groups of continuous data in group Tu is greater than one hour. If it is greater than one hour, the data in that group is retained. Then, the retained groups of data are judged. If the continuous time is greater than 1 hour but less than 2 hours, one period within the duration range corresponding to that group of data is randomly selected as a data point; if it is greater than 2 hours but less than 3 hours, two periods within the duration range corresponding to that group of data are randomly selected as data points, and so on.

[0222] S1246: Obtain a second dataset based on multiple first historical target user sub-traffic data; obtain a third dataset based on multiple first historical non-target user sub-traffic data.

[0223] Having obtained multiple first-historical target user sub-traffic data, a second dataset can be obtained based on these multiple first-historical target user sub-traffic data; having obtained multiple first-historical non-target user sub-traffic data, a third dataset can be obtained based on these multiple first-historical non-target user sub-traffic data.

[0224] In this embodiment of the disclosure, a second dataset can be effectively obtained by comparing the median of the first historical target user sub-traffic data and the median of the first sequence of traffic data, and by sampling the first historical target user sub-traffic data. A third dataset can be effectively obtained by comparing the median of the first historical non-target user sub-traffic data and the median of the second sequence of traffic data, and by sampling the first historical non-target user sub-traffic data.

[0225] In one embodiment, the method for determining user network quality may include step S1247.

[0226] S1247: If the fifth comparison result indicates that the first historical target user sub-traffic data is less than or equal to the median of the first sequence of traffic data, the corresponding number of first historical target user sub-traffic data within the third time period is obtained as the second dataset; if the sixth comparison result indicates that the first historical non-target user sub-traffic data is less than or equal to the median of the second sequence of traffic data, the corresponding number of first historical non-target user sub-traffic data within the third time period is obtained as the third dataset.

[0227] In one example, the third duration is not fixed and can be adjusted according to needs. The corresponding number within the third duration is also not fixed; it can be one, two, or more.

[0228] In one example, the third duration is 6 hours, and the corresponding quantity is 6. If the fifth comparison result indicates that the first historical target user sub-traffic data is less than or equal to the median of the first sequence of traffic data, then 6 first historical target user sub-traffic data can be randomly selected within 6 hours as the second dataset. If the sixth comparison result indicates that the first historical non-target user sub-traffic data is less than or equal to the median of the second sequence of traffic data, then 6 first historical non-target user sub-traffic data can be randomly selected within 6 hours as the third dataset.

[0229] The embodiments of this disclosure obtain a more complete method for obtaining the second and third datasets by acquiring a corresponding number of first historical target user sub-traffic data within a third time period as the second dataset and acquiring a corresponding number of first historical non-target user sub-traffic data within a third time period as the third dataset.

[0230] In one embodiment, after comparing the first similarity with a preset similarity threshold in step S123 to obtain a first comparison result, and comparing the second similarity with a preset similarity threshold to obtain a second comparison result, the method further includes steps S610 and S620.

[0231] S610: If the first comparison result indicates that the first similarity is less than or equal to a preset similarity threshold, the first historical target user traffic data is divided according to the date information to obtain the first historical target user sub-traffic data; if the second comparison result indicates that the second similarity is less than or equal to a preset similarity threshold, the first historical non-target user traffic data is divided according to the date information to obtain the first historical non-target user sub-traffic data.

[0232] In one example, date information includes date attribute information and time period information.

[0233] In another example, if the first comparison result indicates that the first similarity is less than or equal to a preset similarity threshold, the first historical target user traffic data can be divided into time periods using date information, such as time period information, to obtain the first historical target user sub-traffic data. Similarly, if the second comparison result indicates that the second similarity is less than or equal to a preset similarity threshold, the first historical non-target user traffic data can be divided into time periods using date information, such as time period information, to obtain the first historical non-target user sub-traffic data.

[0234] S620: Obtain the corresponding number of first historical target user sub-traffic data within the fourth time period as the second dataset; obtain the corresponding number of first historical non-target user sub-traffic data within the fourth time period as the third dataset.

[0235] In one example, the fourth duration is not fixed and can be adjusted according to needs. The corresponding number within the third duration is also not fixed; it can be one, two, or more.

[0236] In another example, the fourth time period is 8 hours, and the corresponding quantity within the fourth time period is 8. Eight first historical target user sub-traffic data points are randomly selected within the 8 hours as the second dataset; eight first historical non-target user sub-traffic data points are randomly selected within the 8 hours as the third dataset.

[0237] The embodiments of this disclosure provide a more complete way to obtain the second and third datasets by using first historical target user sub-traffic data and first historical non-target user sub-traffic data.

[0238] In one embodiment, step S120, which trains historical user communication parameter data to obtain a network quality determination model, may include steps S126-S128.

[0239] S126: Arbitrarily sort the data information of the first dataset to obtain the fourth dataset.

[0240] In this embodiment of the disclosure, the data information of the first dataset is arbitrarily sorted for data preprocessing.

[0241] In one example, the order of the data information in the first dataset is {(A1, fast internet), (B1, slow internet), (C1, fast internet)}. After arbitrarily sorting the order of the data information in the first dataset, the order of the data information in the first dataset becomes {(C1, fast internet), (A1, fast internet), (B1, slow internet)}.

[0242] S127: Filter out the non-empty data from the fourth dataset to obtain the fifth dataset.

[0243] In this embodiment of the disclosure, the missing data represents the partial data information of the fourth dataset, which includes missing items.

[0244] S128: Divide the fifth dataset according to the preset partitioning ratio to obtain the training set and the test set.

[0245] After obtaining the fifth dataset, it is divided into training and test sets based on a preset partitioning ratio.

[0246] In this embodiment of the disclosure, by filtering out the non-empty data in the fourth dataset, the data information in the fifth dataset is complete. Dividing the first dataset into a training set and a test set can avoid overfitting during model training.

[0247] In one embodiment, before obtaining historical user communication parameter data in step S110, steps S310-S330 are also included.

[0248] S310: Obtain the target user's account list information based on the complaint information sent by the target user.

[0249] In this embodiment of the disclosure, the user sending a complaint message indicates that the user is a user that business personnel need to care about.

[0250] In one embodiment, the target user's account list information, including communication parameter information, can be obtained based on the complaint information sent by the target user.

[0251] S320: Obtain a list of accounts not belonging to the target user.

[0252] In one example, a non-target user could be a VIP user.

[0253] In one embodiment, an account list information, including communication parameter information, of a known VIP user can be obtained.

[0254] S330: Obtain historical communication parameter data of target users based on the account list information of target users; obtain historical communication parameter data of non-target users based on the account list information of non-target users.

[0255] This embodiment of the disclosure obtains the target user's account list information based on the complaint information sent by the target user, which can effectively obtain historical communication parameter data of the target user. Based on the account list information of non-target users, historical communication parameter data of non-target users can also be effectively obtained.

[0256] like Figure 5 As shown in the embodiments of this disclosure, a user network quality determination apparatus 400 is also provided, the user network quality determination apparatus 400 comprising:

[0257] The acquisition module 401 is used to acquire historical user communication parameter data, which includes historical traffic data and tag data.

[0258] Training module 402 is used to train on historical user communication parameter data to obtain a network quality determination model. The network quality determination model includes the correspondence information between historical traffic data and label data.

[0259] The acquisition module 401 is also used to acquire user communication parameter data within a target time period, including traffic data;

[0260] The acquisition module 401 is also used to acquire the target similarity of traffic data within the same time period based on date information and time order.

[0261] The comparison module 403 is used to compare the target similarity with a first similarity threshold to obtain the comparison result;

[0262] Module 404 is used to determine the target dataset corresponding to the comparison result;

[0263] Input module 405 is used to input the target dataset into the network quality determination model to determine the target prediction result;

[0264] The process involves training a network quality determination model using historical user communication parameter data. This includes: processing the historical user communication parameter data according to time sequence or the size of historical traffic data to obtain a first dataset containing historical traffic data; dividing the first dataset into a training set and a test set based on a preset partitioning ratio, with the training set including historical traffic data and the test set including label data, respectively; training the model based on the training set to obtain multiple network quality determination models, each including the correspondence information between historical traffic data and label data; inputting the historical traffic data from the test set into the multiple network quality determination models to obtain a first prediction result; comparing the label data from the test set with the first prediction result to obtain a comparison result; for each network quality determination model, counting the number of first prediction results in the comparison result that match the label data from the test set; and obtaining a relatively large number of network quality determination models.

[0265] In some embodiments, the device for determining user network quality includes a partitioning module, which is used to partition traffic data within the same time period according to date information to obtain sub-traffic data;

[0266] The sub-flow data are sorted according to time order to obtain the first target sequence;

[0267] Based on multiple first target sequences with the same date information, the similarity between any two first target sequences with the same date information is calculated using a preset similarity calculation method, and a similarity set of the first target sequences is obtained.

[0268] Based on the data information of the similarity set of the first target sequence, the average value of the similarity set of the first target sequence is calculated, and the average value is determined as the target similarity of the traffic data.

[0269] In some embodiments, the acquisition module 401 is specifically used for:

[0270] If the target similarity is greater than the first similarity threshold, obtain the traffic data that is greater than the first preset threshold of traffic data within the same time period to obtain the target dataset;

[0271] If the target similarity is less than or equal to the first similarity threshold, a preset number of traffic data points are randomly obtained from the traffic data within the same time period to obtain the target dataset.

[0272] In some embodiments, the device for determining user network quality includes a calculation module, which is configured to calculate the median of traffic data of the first target sequence based on multiple sub-traffic data of the first target sequence when the target similarity is greater than a first similarity threshold, and use the median as a first preset threshold.

[0273] By comparing the sub-flow data with the median of the flow data in the first target sequence, the comparison result of the second target is obtained;

[0274] If the second target comparison result indicates that the sub-flow data is greater than the median of the flow data in the first target sequence, the sub-flow data is grouped to obtain the grouped sub-flow data.

[0275] The sub-traffic data after grouping is filtered according to a preset duration threshold to obtain the filtered sub-traffic data.

[0276] Based on the first target duration corresponding to the filtered sub-traffic data, obtain the number of sub-traffic data corresponding to the first target duration;

[0277] The target dataset is obtained from multiple sub-flow data.

[0278] In some embodiments, the acquisition module 401 is specifically used for:

[0279] If the second target comparison result indicates that the sub-traffic data is less than or equal to the median of the traffic data in the first target sequence, the corresponding number of first target user sub-traffic data within the first target duration is obtained as the target dataset.

[0280] In some embodiments, the input module 405 is specifically used for:

[0281] The traffic data in the target dataset is input into the network quality determination model to obtain multiple initial prediction results corresponding to the traffic data. The initial prediction results include the initial prediction target result and the initial prediction non-target result.

[0282] The initial prediction ratio is obtained by dividing the number of multiple initial prediction target results by the number of traffic data in the target dataset;

[0283] The initial prediction ratio is compared with the preset prediction threshold to obtain the target comparison result;

[0284] If the target comparison result indicates that the initial prediction ratio is greater than the preset prediction threshold, the target prediction result corresponding to the target dataset is determined.

[0285] In some embodiments, the means for determining user network quality includes a filtering module, which is configured to: filter historical target user traffic data according to a target filtering threshold to obtain first historical target user traffic data that is greater than the target filtering threshold; and filter historical non-target user traffic data according to the target filtering threshold to obtain first historical non-target user traffic data that is greater than the target filtering threshold.

[0286] Obtain the first similarity of the first historical target user traffic data; obtain the second similarity of the first historical non-target user traffic data;

[0287] Compare the first similarity with a preset similarity threshold to obtain a first comparison result; compare the second similarity with a preset similarity threshold to obtain a second comparison result;

[0288] If the first comparison result indicates that the first similarity is greater than the preset similarity threshold, the first historical target user traffic data with larger traffic data is obtained according to the size of the first historical target user traffic data, and a second dataset is obtained; if the second comparison result indicates that the second similarity is greater than the preset similarity threshold, the first historical non-target user traffic data with larger traffic data is obtained according to the size of the first historical non-target user traffic data, and a third dataset is obtained.

[0289] The second and third datasets are combined to obtain the first dataset.

[0290] In some embodiments, the device for determining user network quality includes a partitioning module, which is configured to: partition first historical target user traffic data according to date information to obtain first historical target user sub-traffic data; and partition first historical non-target user traffic data according to date information to obtain first historical non-target user sub-traffic data.

[0291] Sort the first historical target user sub-traffic data in chronological order to obtain the first sequence; sort the first historical non-target user sub-traffic data in chronological order to obtain the second sequence.

[0292] Based on multiple first sequences with identical date information, the similarity between any two first sequences with identical date information is calculated using a preset similarity calculation method, resulting in a similarity set for the first sequences; based on multiple second sequences with identical date information, the similarity between any two second sequences with identical date information is calculated using the preset similarity calculation method, resulting in a similarity set for the second sequences.

[0293] Based on the data information of the similarity set of the first sequence, the median and mean of the similarity set of the first sequence are calculated; based on the data information of the similarity set of the second sequence, the median and mean of the similarity set of the second sequence are calculated.

[0294] The third comparison result is obtained by comparing the mean of the similarity set of the first sequence with the median of the similarity set of the second sequence; the fourth comparison result is obtained by comparing the mean of the similarity set of the second sequence with the median of the similarity set of the second sequence.

[0295] If the third comparison result indicates that the average value of the similarity set of the first sequence is greater than the median of the similarity set of the first sequence, the average value of the similarity set of the first sequence is obtained as the first similarity of the first historical target user traffic data; if the fourth comparison result indicates that the average value of the similarity set of the second sequence is greater than the median of the similarity set of the second sequence, the average value of the similarity set of the second sequence is obtained as the second similarity of the first historical non-target user traffic data.

[0296] In some embodiments, the means for determining user network quality includes a calculation module, which is configured to: calculate the median of traffic data for a first sequence based on a plurality of first historical target user sub-traffic data for a first sequence; and calculate the median of traffic data for a second sequence based on a first historical non-target user sub-traffic data for a second sequence.

[0297] The fifth comparison result is obtained by comparing the median of the traffic data of the first historical target user sub-data with the median of the traffic data of the first sequence; the sixth comparison result is obtained by comparing the median of the traffic data of the first historical non-target user sub-data with the median of the traffic data of the second sequence.

[0298] If the fifth comparison result indicates that the first historical target user sub-traffic data is greater than the median of the first sequence of traffic data, the first historical target user sub-traffic data is grouped to obtain the grouped first historical target user sub-traffic data; if the sixth comparison result indicates that the first historical non-target user sub-traffic data is greater than the median of the second sequence of traffic data, the first historical non-target user sub-traffic data is grouped to obtain the grouped first historical non-target user sub-traffic data.

[0299] The first historical target user sub-traffic data after grouping is filtered according to a preset duration threshold to obtain the filtered first historical target user sub-traffic data; the first historical non-target user sub-traffic data after grouping is filtered according to a preset duration threshold to obtain the filtered first historical non-target user sub-traffic data.

[0300] Based on the first duration corresponding to the filtered first historical target user sub-traffic data, obtain the first historical target user sub-traffic data of a quantity corresponding to the first duration; based on the second duration corresponding to the filtered first historical non-target user sub-traffic data, obtain the first historical non-target user sub-traffic data of a quantity corresponding to the second duration.

[0301] A second dataset is obtained based on multiple first historical target user sub-traffic data; a third dataset is obtained based on multiple first historical non-target user sub-traffic data.

[0302] In some embodiments, the acquisition module 401 is specifically used for:

[0303] If the fifth comparison result indicates that the first historical target user sub-traffic data is less than or equal to the median of the first sequence of traffic data, the corresponding number of first historical target user sub-traffic data within the third time period is obtained as the second dataset; if the sixth comparison result indicates that the first historical non-target user sub-traffic data is less than or equal to the median of the second sequence of traffic data, the corresponding number of first historical non-target user sub-traffic data within the third time period is obtained as the third dataset.

[0304] In some embodiments, the device for determining user network quality includes a segmentation module, which is configured to, when a first comparison result indicates that a first similarity is less than or equal to a preset similarity threshold, segment first historical target user traffic data according to date information to obtain first historical target user sub-traffic data; and when a second comparison result indicates that a second similarity is less than or equal to a preset similarity threshold, segment first historical non-target user traffic data according to date information to obtain first historical non-target user sub-traffic data.

[0305] Obtain the first historical target user sub-traffic data corresponding to the fourth time period as the second dataset; obtain the first historical non-target user sub-traffic data corresponding to the fourth time period as the third dataset.

[0306] In some embodiments, the device for determining user network quality includes a sorting module, which is used to: arbitrarily sort the data information of the first dataset to obtain a fourth dataset;

[0307] The non-empty data in the fourth dataset are filtered out to obtain the fifth dataset;

[0308] The fifth dataset is divided into training and test sets based on a preset partitioning ratio.

[0309] In some embodiments, the acquisition module 401 is specifically used for:

[0310] Based on the complaint information sent by the target user, obtain the target user's account list information;

[0311] Obtain a list of accounts belonging to non-target users;

[0312] Retrieve historical user communication parameter data, including:

[0313] Based on the account list information of the target users, obtain historical communication parameter data of the target users; based on the account list information of the non-target users, obtain historical communication parameter data of the non-target users.

[0314] Figure 5 Each module / unit in the illustrated device has the ability to implement Figure 2 and Figure 3The functions of each step in the process and their corresponding technical effects are described in detail here for the sake of brevity.

[0315] Figure 6 A schematic diagram of the hardware structure for determining user network quality provided in an embodiment of this disclosure is shown.

[0316] The device used to determine the quality of the user's network may include a processor 501 and a memory 502 storing computer program instructions.

[0317] Specifically, the processor 501 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0318] Memory 502 may include mass storage for data or instructions. For example, and not limitingly, memory 502 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 502 may include removable or non-removable (or fixed) media. Where appropriate, memory 502 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 502 is non-volatile solid-state memory.

[0319] Memory 502 may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to one aspect of this disclosure.

[0320] The processor 501 implements any of the user network quality determination methods in the above embodiments by reading and executing computer program instructions stored in the memory 502.

[0321] In one example, the device for determining user network quality may also include a communication interface 503 and a bus 504. Wherein, as Figure 6As shown, the processor 501, memory 502, and communication interface 503 are connected through bus 504 and complete communication with each other.

[0322] The communication interface 503 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0323] Bus 504 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 504 may include one or more buses. Although specific buses are described and illustrated in embodiments of the invention, the invention contemplates any suitable bus or interconnect. Additionally, in conjunction with the user network quality determination method in the above embodiments, embodiments of this application also provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when these computer program instructions are executed by the processor, they implement any of the methods for determining user network quality in the above embodiments.

[0324] This application also provides a computer program product, including a computer program that, when executed by a processor, implements any of the methods for determining user network quality in the above embodiments.

[0325] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0326] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the required tasks. The programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, read-only memory (ROM), flash memory, erasable read-only memory (EROM), floppy disks, compact disc read-only memory (CD-ROM), optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0327] It should also be noted that the exemplary embodiments mentioned in this invention describe methods or systems based on a series of steps or apparatus. However, this invention is not limited to the order of the steps described above; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0328] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0329] The above description is merely a specific embodiment of the present invention. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the protection scope of the present invention.

Claims

1. A method for determining user network quality, characterized in that, include: Acquire historical user communication parameter data, which includes historical traffic data and tag data; The network quality determination model is obtained by training the historical user communication parameter data, and the network quality determination model includes the correspondence information between the historical traffic data and the tag data; Acquire user communication parameter data within a target time period, wherein the user communication parameter data includes traffic data; Based on date information and time order, obtain the target similarity of the traffic data within the same time period; The target similarity is compared with a first similarity threshold to obtain a comparison result; Determine the target dataset corresponding to the comparison result; The target dataset is input into the network quality determination model to determine the target prediction result; The step of training the historical user communication parameter data to obtain a network quality determination model includes: processing the historical user communication parameter data according to time sequence or the size of the historical traffic data to obtain a first dataset including the historical traffic data; dividing the first dataset according to a preset partitioning ratio to obtain a training set and a test set, wherein the training set and the test set respectively include the historical traffic data and the label data; training the model based on the training set to obtain multiple network quality determination models, each network quality determination model including the correspondence information between the historical traffic data and the label data; inputting the historical traffic data in the test set into the multiple network quality determination models to obtain a first prediction result; comparing the label data in the test set with the first prediction result to obtain a comparison result; for each network quality determination model, counting the number of first prediction results in the comparison result that are consistent with the label data in the test set; and obtaining the network quality determination models with a relatively large number of results.

2. The method according to claim 1, characterized in that, The step of obtaining the target similarity of the traffic data within the same time period based on date information and time order includes: Based on date information, the traffic data within the same time period is divided to obtain sub-traffic data; The sub-flow data are sorted according to the time sequence to obtain the first target sequence; Based on multiple first target sequences with the same date information, the similarity between any two first target sequences with the same date information is calculated using a preset similarity calculation method to obtain a similarity set of the first target sequences; Based on the data information of the similarity set of the first target sequence, the average value of the similarity set of the first target sequence is calculated, and the average value is determined as the target similarity of the traffic data.

3. The method according to claim 2, characterized in that, Determining the target dataset corresponding to the comparison result includes: If the target similarity is greater than the first similarity threshold, traffic data that is greater than a first preset threshold of traffic data within the same time period is obtained to obtain the target dataset; If the target similarity is less than or equal to the first similarity threshold, a preset number of traffic data points are randomly obtained from the traffic data within the same time period to obtain the target dataset.

4. The method according to claim 3, characterized in that, When the target similarity is greater than the first similarity threshold, the traffic data within the same time period that exceeds a first preset threshold is obtained to acquire the target dataset, including: If the target similarity is greater than the first similarity threshold, the median of the traffic data of the first target sequence is calculated based on multiple sub-traffic data of the first target sequence, and the median is used as the first preset threshold. The median of the sub-flow data and the flow data of the first target sequence are compared to obtain the second target comparison result; If the second target comparison result indicates that the sub-flow data is greater than the median of the flow data in the first target sequence, the sub-flow data is grouped to obtain grouped sub-flow data. The grouped sub-traffic data is filtered according to a preset duration threshold to obtain filtered sub-traffic data. Based on the first target duration corresponding to the filtered sub-traffic data, obtain the number of sub-traffic data corresponding to the first target duration; The target dataset is obtained based on multiple sub-flow data.

5. The method according to claim 4, characterized in that, The method further includes: If the second target comparison result indicates that the sub-traffic data is less than or equal to the median of the traffic data in the first target sequence, the corresponding number of first target user sub-traffic data within the first target duration is obtained as the target dataset.

6. The method according to claim 1, characterized in that, The step of inputting the target dataset into the network quality determination model to determine the target prediction result includes: The traffic data in the target dataset is input into the network quality determination model to obtain multiple initial prediction results corresponding to the traffic data. The initial prediction results include initial target prediction results and initial non-target prediction results. The initial prediction ratio is obtained by dividing the number of multiple initial prediction target results by the number of traffic data in the target dataset; The initial prediction ratio is compared with the preset prediction threshold to obtain the target comparison result; If the target comparison result indicates that the initial prediction ratio is greater than the preset prediction threshold, a target prediction result corresponding to the target dataset is determined.

7. The method according to claim 1, characterized in that, The users include target users and non-target users. The historical user communication parameter data includes historical target user traffic data and historical non-target user traffic data. The step of processing the historical user communication parameter data according to time order or the size of the historical traffic data to obtain a first dataset including the historical traffic data includes: According to the target filtering threshold, the historical target user traffic data is filtered to obtain first historical target user traffic data that is greater than the target filtering threshold; according to the target filtering threshold, the historical non-target user traffic data is filtered to obtain first historical non-target user traffic data that is greater than the target filtering threshold. Obtain the first similarity of the first historical target user traffic data; obtain the second similarity of the first historical non-target user traffic data; The first similarity is compared with a preset similarity threshold to obtain a first comparison result; the second similarity is compared with the preset similarity threshold to obtain a second comparison result. If the first comparison result indicates that the first similarity is greater than the preset similarity threshold, then based on the size of the first historical target user traffic data, the first historical target user traffic data with larger traffic data is obtained to obtain a second dataset; if the second comparison result indicates that the second similarity is greater than the preset similarity threshold, then based on the size of the first historical non-target user traffic data, the first historical non-target user traffic data with larger traffic data is obtained to obtain a third dataset. The second dataset and the third dataset are combined to obtain the first dataset.

8. The method according to claim 7, characterized in that, The first historical target user traffic data includes first historical target user sub-traffic data, the first historical non-target user traffic data includes first historical non-target user sub-traffic data, and the first similarity of the first historical target user traffic data is obtained. Obtaining the second similarity of the first historical non-target user traffic data includes: Based on the date information, the first historical target user traffic data is divided to obtain the first historical target user sub-traffic data; based on the date information, the first historical non-target user traffic data is divided to obtain the first historical non-target user sub-traffic data. The first historical target user sub-traffic data is sorted according to the time order to obtain a first sequence; the first historical non-target user sub-traffic data is sorted according to the time order to obtain a second sequence. Based on multiple first sequences with the same date information, the similarity between any two first sequences with the same date information is calculated using a preset similarity calculation method to obtain a similarity set of the first sequences; based on multiple second sequences with the same date information, the similarity between any two second sequences with the same date information is calculated using the preset similarity calculation method to obtain a similarity set of the second sequences; Based on the data information of the similarity set of the first sequence, the median and average of the similarity set of the first sequence are calculated; based on the data information of the similarity set of the second sequence, the median and average of the similarity set of the second sequence are calculated. A third comparison result is obtained by comparing the average value of the similarity set of the first sequence with the median value of the similarity set of the first sequence; a fourth comparison result is obtained by comparing the average value of the similarity set of the second sequence with the median value of the similarity set of the second sequence. If the third comparison result indicates that the average value of the similarity set of the first sequence is greater than the median of the similarity set of the first sequence, the average value of the similarity set of the first sequence is obtained as the first similarity of the first historical target user traffic data; if the fourth comparison result indicates that the average value of the similarity set of the second sequence is greater than the median of the similarity set of the second sequence, the average value of the similarity set of the second sequence is obtained as the second similarity of the first historical non-target user traffic data.

9. The method according to claim 8, characterized in that, When the first comparison result indicates that the first similarity is greater than the preset similarity threshold, the first historical target user traffic data with larger traffic data is obtained according to the size of the first historical target user traffic data to obtain the second dataset; If the second comparison result indicates that the second similarity is greater than the preset similarity threshold, based on the size of the first historical non-target user traffic data, the first historical non-target user traffic data with larger traffic data is obtained to obtain a third dataset, including: The median of the traffic data for the first sequence is calculated based on multiple first historical target user sub-traffic data for the first sequence; the median of the traffic data for the second sequence is calculated based on the first historical non-target user sub-traffic data for the second sequence. A fifth comparison result is obtained by comparing the first historical target user sub-traffic data with the median of the traffic data in the first sequence; a sixth comparison result is obtained by comparing the first historical non-target user sub-traffic data with the median of the traffic data in the second sequence. If the fifth comparison result indicates that the first historical target user sub-traffic data is greater than the median of the traffic data in the first sequence, the first historical target user sub-traffic data is grouped to obtain the grouped first historical target user sub-traffic data; if the sixth comparison result indicates that the first historical non-target user sub-traffic data is greater than the median of the traffic data in the second sequence, the first historical non-target user sub-traffic data is grouped to obtain the grouped first historical non-target user sub-traffic data. The first historical target user sub-traffic data after grouping is filtered according to a preset duration threshold to obtain filtered first historical target user sub-traffic data; the first historical non-target user sub-traffic data after grouping is filtered according to the preset duration threshold to obtain filtered first historical non-target user sub-traffic data. Based on the first duration corresponding to the filtered first historical target user sub-traffic data, obtain the number of first historical target user sub-traffic data corresponding to the first duration; based on the second duration corresponding to the filtered first historical non-target user sub-traffic data, obtain the number of first historical non-target user sub-traffic data corresponding to the second duration. The second dataset is obtained based on multiple first historical target user sub-traffic data; the third dataset is obtained based on multiple first historical non-target user sub-traffic data.

10. The method according to claim 9, characterized in that, The method further includes: If the fifth comparison result indicates that the first historical target user sub-traffic data is less than or equal to the median of the traffic data in the first sequence, the corresponding number of first historical target user sub-traffic data within the third time period is obtained as the second dataset; if the sixth comparison result indicates that the first historical non-target user sub-traffic data is less than or equal to the median of the traffic data in the second sequence, the corresponding number of first historical non-target user sub-traffic data within the third time period is obtained as the third dataset.

11. The method according to claim 7, characterized in that, By comparing the first similarity with a preset similarity threshold, a first comparison result is obtained; After comparing the second similarity with the preset similarity threshold to obtain a second comparison result, the method further includes: If the first comparison result indicates that the first similarity is less than or equal to the preset similarity threshold, the first historical target user traffic data is divided according to the date information to obtain the first historical target user sub-traffic data; if the second comparison result indicates that the second similarity is less than or equal to the preset similarity threshold, the first historical non-target user traffic data is divided according to the date information to obtain the first historical non-target user sub-traffic data. Obtain the corresponding number of first historical target user sub-traffic data within the fourth time period as the second dataset; obtain the corresponding number of first historical non-target user sub-traffic data within the fourth time period as the third dataset.

12. The method according to claim 1, characterized in that, The process of dividing the first dataset into a training set and a test set based on a preset partitioning ratio includes: The data information of the first dataset is arbitrarily sorted to obtain the fourth dataset; The non-empty data in the fourth dataset are filtered out to obtain the fifth dataset; The fifth dataset is divided according to a preset partitioning ratio to obtain the training set and the test set.

13. The method according to claim 1, characterized in that, The users include target users and non-target users. Before obtaining historical user communication parameter data, the method further includes: Based on the complaint information sent by the target user, obtain the account list information of the target user; Obtain the account list information of the non-target users; The acquisition of historical user communication parameter data includes: Based on the account list information of the target users, obtain historical communication parameter data of the target users; based on the account list information of the non-target users, obtain historical communication parameter data of the non-target users.

14. An apparatus for determining user network quality, characterized in that, The device includes: The acquisition module is used to acquire historical user communication parameter data, which includes historical traffic data and tag data. The training module is used to train the historical user communication parameter data to obtain a network quality determination model, which includes the correspondence information between the historical traffic data and the tag data. The acquisition module is also used to acquire user communication parameter data within a target time period, the user communication parameter data including traffic data; The acquisition module is also used to acquire the target similarity of the traffic data within the same time period based on date information and time order; The comparison module is used to compare the target similarity with a first similarity threshold to obtain a comparison result; The determination module is used to determine the target dataset corresponding to the comparison result; The input module is used to input the target dataset into the network quality determination model to determine the target prediction result; The step of training the historical user communication parameter data to obtain a network quality determination model includes: processing the historical user communication parameter data according to time sequence or the size of the historical traffic data to obtain a first dataset including the historical traffic data; dividing the first dataset according to a preset partitioning ratio to obtain a training set and a test set, wherein the training set and the test set respectively include the historical traffic data and the label data; training the model based on the training set to obtain multiple network quality determination models, each network quality determination model including the correspondence information between the historical traffic data and the label data; inputting the historical traffic data in the test set into the multiple network quality determination models to obtain a first prediction result; comparing the label data in the test set with the first prediction result to obtain a comparison result; for each network quality determination model, counting the number of first prediction results in the comparison result that are consistent with the label data in the test set; and obtaining the network quality determination models with a relatively large number of results.

15. A device for determining user network quality, characterized in that, The device includes: a processor and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement the method for determining user network quality as described in any one of claims 1-13.

16. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the method for determining user network quality as described in any one of claims 1-13.

17. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method for determining user network quality as described in any one of claims 1-13.

Images