User clustering method and device and electronic equipment

Abstract

The invention provides a user clustering method. The method comprises the following steps: acquiring first satisfaction evaluation data of a survey user and a first parameter value of a first satisfaction influence parameter; determining a target satisfaction influence parameter based on the first satisfaction evaluation data and the first parameter value; based on the target satisfaction influence parameter and the first satisfaction evaluation data, constructing a satisfaction prediction model; determining a satisfaction prediction value of the to-be-predicted user by using the satisfaction prediction model; and based on the satisfaction prediction value, clustering predicted unsatisfied users in the to-be-predicted users to obtain a clustering result of the predicted unsatisfied users. According to the method disclosed by the invention, by determining the target satisfaction influence parameter, the redundant first satisfaction influence parameter is eliminated, and the calculation efficiency and accuracy of the satisfaction prediction model are improved; and through the satisfaction prediction value, the predicted unsatisfied users in the to-be-predicted users are clustered, so that the accuracy of the clustering result can be improved, and the network quality of the users can be improved.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence, and in particular to a user clustering method, device and electronic device. Background Art

[0002] The traditional method of analyzing user satisfaction is often to obtain user satisfaction data based on telephone surveys and complaints. This method is slow to react and can only analyze network conditions based on problems reported by dissatisfied users, and cannot effectively guarantee user service quality.

[0003] At the same time, in order to further strengthen the perception-oriented analysis capabilities and ensure that perception problems are solved quickly and fundamentally, the network should be customer-centric, build a perception-based wireless network quality optimization system, change from "performance-centric" to "perception-centric", and change the solution granularity from "single problem cell" to "key areas and key scenarios", and strengthen the effective transmission of real customer experience to network operations. Therefore, a method is needed to pre-analyze the time and space of dissatisfied users, implement network optimization and adjustment in advance, and improve customer perception. Summary of the invention

[0004] The present disclosure provides a user clustering method, device and electronic device to solve the problems in the related art, realize the prediction of user satisfaction, and improve the accuracy of clustering results, so as to improve the network quality of users.

[0005] The first aspect of the embodiment of the present disclosure proposes a user clustering method, including: obtaining first satisfaction evaluation data of surveyed users and a first parameter value of a first satisfaction influencing parameter; determining a target satisfaction influencing parameter based on the first satisfaction evaluation data and the first parameter value; constructing a satisfaction prediction model based on the target satisfaction influencing parameter and the first satisfaction evaluation data; determining a satisfaction prediction value of a user to be predicted using the satisfaction prediction model; and clustering predicted dissatisfied users among the users to be predicted based on the satisfaction prediction values ​​to obtain a clustering result of the predicted dissatisfied users.

[0006] In some embodiments of the present disclosure, determining the target satisfaction influencing parameter based on the first satisfaction evaluation data and the first parameter value includes: performing a Boolean conversion on the first satisfaction evaluation data to obtain a first target variable; determining a survey user whose first satisfaction evaluation data is greater than a first preset value as a high-satisfaction user; determining a second parameter value corresponding to the high-satisfaction user from the first parameter value; determining a first parameter threshold of the first satisfaction influencing parameter based on the second parameter value; determining the second target variable based on the first parameter value and the first parameter threshold; and determining the target satisfaction influencing parameter based on the first target variable and the second target variable.

[0007] In some embodiments of the present disclosure, determining the second target variable based on the first parameter value and the parameter threshold includes: determining the first satisfaction influencing parameter corresponding to the second parameter value exceeding the parameter threshold as the first quality difference parameter; determining the first satisfaction influencing parameter of the second parameter value not exceeding the parameter threshold as the second quality difference parameter; constructing a Boolean matrix based on the first quality difference parameter and the second quality difference parameter; and determining the Boolean matrix as the second target variable.

[0008] In some embodiments of the present disclosure, determining the target satisfaction influencing parameter based on the first target variable and the second target variable includes: determining the first contribution weight of the first satisfaction influencing parameter based on the first target variable and the second target variable using a random forest algorithm; determining the parameter correlation between the first satisfaction influencing parameters to obtain a correlation coefficient matrix; determining the second satisfaction influencing parameter from the first satisfaction influencing parameter based on the correlation coefficient matrix, wherein the correlation between the second satisfaction influencing parameters is less than or equal to a preset correlation threshold parameter; determining the second contribution weight of the second satisfaction influencing parameter based on the correlation coefficient matrix and the first contribution weight; determining the target satisfaction influencing parameter based on the second satisfaction influencing parameter and the second contribution weight.

[0009] In some embodiments of the present disclosure, constructing a satisfaction prediction model based on the target satisfaction influencing parameter and the first satisfaction evaluation data includes:

[0010] Based on the surveyed users and the first satisfaction evaluation data, a third target variable is constructed; the first time parameter corresponding to the first parameter value is obtained; based on the first parameter value and the preset parameter threshold, a third quality difference parameter of the surveyed user is determined; based on the third quality difference parameter and the first time parameter, the total number of quality differences of the surveyed user within the preset time period is determined; based on the preset time period and the total number of quality differences, the first time period in which the surveyed user has quality difference events is determined to construct a fourth target variable; the third target variable and the fourth target variable are used as training data to train the model to be trained to obtain a satisfaction prediction model.

[0011] In some embodiments of the present disclosure, using a satisfaction prediction model to determine a satisfaction prediction value of a user to be predicted includes: based on a target satisfaction influencing parameter, obtaining a third parameter value of the user to be predicted, and a second time parameter corresponding to the third parameter value; based on the third parameter value and the second time parameter, determining a second time period in which a quality difference event occurs for the user to be predicted; and based on the second time period, determining a satisfaction prediction value using the satisfaction prediction model.

[0012] In some embodiments of the present disclosure, based on the satisfaction prediction value, clustering the predicted dissatisfied users among the predicted users to obtain the clustering results of the predicted dissatisfied users includes: determining the predicted users whose satisfaction prediction values ​​are less than a second preset value as predicted dissatisfied users; obtaining a network parameter set of the predicted dissatisfied users and a target poor quality event of the predicted dissatisfied users; determining the service grid of the predicted dissatisfied users based on the network parameter set; determining a user scenario library of the predicted dissatisfied users based on the residence time of the predicted dissatisfied users in the service grid, the user scenario library including at least one of the following: a permanent workplace, a permanent residence and a user activity route; marking scenario attributes for the target poor quality event based on the user scenario library; and clustering at least one of the scenario attributes, the target poor quality event and the residence event using a linear regression algorithm to obtain the clustering results of the predicted dissatisfied users.

[0013] In some embodiments of the present disclosure, the network parameter set includes at least one of the following: predicted serving cell of dissatisfied users, predicted neighboring cell signal strength of dissatisfied users, and timing advance TA of signals in the serving cell.

[0014] The second aspect of the embodiment of the present disclosure proposes a user clustering device, which includes: an acquisition unit, which is used to acquire first satisfaction evaluation data of surveyed users and a first parameter value of a first satisfaction influencing parameter; a first determination unit, which is used to determine a target satisfaction influencing parameter based on the first satisfaction evaluation data and the first parameter value; a construction unit, which is used to construct a satisfaction prediction model based on the target satisfaction influencing parameter and the first satisfaction evaluation data; a second determination unit, which is used to use the satisfaction prediction model to determine the satisfaction prediction value of the user to be predicted; and a clustering unit, which is used to cluster target quality poor events of predicted dissatisfied users among the users to be predicted based on the satisfaction prediction value, so as to obtain a clustering result of predicted dissatisfied users.

[0015] The third aspect embodiment of the present disclosure proposes an electronic device, comprising: a processor and a memory for storing a computer program that can be run on the processor, wherein the processor, when used to run the computer program, executes the method described in the first aspect embodiment of the present disclosure.

[0016] In summary, according to a user clustering method proposed in the present disclosure, the method includes: obtaining the first satisfaction evaluation data of the surveyed user and the first parameter value of the first satisfaction influencing parameter; determining the target satisfaction influencing parameter based on the first satisfaction evaluation data and the first parameter value; constructing a satisfaction prediction model based on the target satisfaction influencing parameter and the first satisfaction evaluation data; determining the satisfaction prediction value of the user to be predicted using the satisfaction prediction model; clustering the predicted dissatisfied users among the users to be predicted based on the satisfaction prediction value to obtain the clustering results of the predicted dissatisfied users. The method disclosed in the present disclosure, by determining the target satisfaction influencing parameter, eliminates the redundant first satisfaction influencing parameter, and improves the computational efficiency and accuracy of the satisfaction prediction model; and by clustering the predicted dissatisfied users among the users to be predicted through the satisfaction prediction value, the accuracy of the clustering results can be improved, which is convenient for improving the network quality of the users.

[0017] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute improper limitations on the present disclosure.

[0019] Figure 1 A flowchart of a user clustering method provided by an embodiment of the present disclosure;

[0020] Figure 2 A flowchart of another user clustering method provided by an embodiment of the present disclosure;

[0021] Figure 3 A flowchart of another user clustering method provided by an embodiment of the present disclosure;

[0022] Figure 4 A flowchart of another user clustering method provided by an embodiment of the present disclosure;

[0023] Figure 5 A flowchart of a user clustering method provided by an embodiment of the present disclosure;

[0024] Figure 6 A schematic diagram of the structure of a user clustering device provided in an embodiment of the present disclosure;

[0025] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0026] Embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, and should not be construed as limiting the present disclosure.

[0027] The following is a brief introduction to the background technology involved in this disclosure:

[0028] The traditional method of analyzing user satisfaction is often to obtain user satisfaction data based on telephone surveys and complaints. This method is slow to react and can only analyze network conditions based on problems reported by dissatisfied users, and cannot effectively guarantee user service quality.

[0029] The existing technology is mainly a user satisfaction judgment method based on resident problem cells. It is a user satisfaction determination technology based on resident problem cells. Starting from the two aspects of user and cell dimensions, resident problem cells are screened out through correlation analysis of specific indicators. For the perception indicators related to user satisfaction, the model indicator set is established. The user's resident problem cells are taken as the research object. Through data acquisition and machine learning of the data, the quantitative scoring of customer perception is achieved, and the user satisfaction of resident problem cells is finally determined. The correlation analysis between users and resident cells is carried out, and the resident cells that affect the network quality are found in combination with the satisfaction indicators, which are used as the basis for analysis and processing to improve the network quality.

[0030] Therefore, the existing technology mainly takes the resident cells as the object and is used to analyze the perception of the user's resident cells. This technology cannot reflect the actual perception of the user. It cannot meet the needs of the solution granularity change from "single problem cell, single complaint" to "key areas, key scenarios, key customers".

[0031] The present disclosure is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0032] Figure 1 A flowchart of a user clustering method provided by an embodiment of the present disclosure. Figure 1 As shown, the method includes steps 101-105.

[0033] Step 101: Obtain first satisfaction evaluation data of a surveyed user and a first parameter value of a first satisfaction influencing parameter.

[0034] In some embodiments, the first satisfaction evaluation data may include a satisfaction score and a satisfaction result (ie, whether the user is satisfied), such as shown in the following table:

[0035]

[0036] In some embodiments, the first satisfaction influencing parameter refers to a parameter that affects the first satisfaction evaluation data. In other words, the first satisfaction influencing parameter refers to a parameter that can affect the satisfaction of the surveyed user.

[0037] In some embodiments, the first satisfaction influencing parameter is, for example, a user-level KQI (Key Quality Indicator) indicator. In other words, the first satisfaction influencing parameter may include: Hypertext Transfer Protocol (HTTP) delay, traffic, and rate; Transmission Control Protocol (TCP) establishment delay, success rate, and retransmission rate; Round-Trip Time (RTT) delay; video freeze delay, buffering delay, upload, and download rate; the number of times audio / picture / text is sent / received, success rate, etc.

[0038] In some embodiments, the first parameter value is the value of the first satisfaction influencing parameter of the surveyed user obtained.

[0039] Step 102: Determine a target satisfaction influencing parameter based on the first satisfaction evaluation data and the first parameter value.

[0040] In some embodiments, since the first satisfaction influencing parameters include many parameters, some of which have little or no impact on user satisfaction, in order to improve the accuracy and speed of user satisfaction prediction, it is necessary to eliminate some parameters in the first satisfaction influencing parameters to obtain the target satisfaction influencing parameters.

[0041] In some optional embodiments, the first target variable can be obtained by performing a Boolean conversion on the first satisfaction evaluation data; the survey users whose first satisfaction evaluation data is greater than a first preset value are determined as high-satisfaction users; the second parameter value corresponding to the high-satisfaction users is determined from the first parameter value; based on the second parameter value, the first parameter threshold of the first satisfaction influencing parameter is determined; based on the first parameter value and the first parameter threshold, the second target variable is determined; based on the first target variable and the second target variable, the target satisfaction influencing parameter is determined.

[0042] In some optional embodiments, the target satisfaction influencing parameter can also be determined from the first satisfaction influencing parameters through the correspondence between multiple groups of first satisfaction evaluation data and first parameter values. (For example, when the first parameter value of parameter A increases by only 0.1, but the first satisfaction evaluation data increases significantly, the parameter A can be determined as the target satisfaction parameter).

[0043] Step 103: construct a satisfaction prediction model based on the target satisfaction influencing parameter and the first satisfaction evaluation data.

[0044] In some embodiments, the target satisfaction parameters and the first satisfaction evaluation data can be used to perform machine learning on the model to be trained, so that the model to be trained can understand the correlation between the target satisfaction influencing parameters and the first satisfaction evaluation data, thereby determining the trained model to be trained as a satisfaction prediction model.

[0045] In some optional embodiments, a first time parameter corresponding to a first parameter value can be obtained; based on the first parameter value and a preset parameter threshold, a third quality difference parameter of the surveyed user can be determined; based on the third quality difference parameter and the first time parameter, the total number of quality differences of the surveyed user within a preset time period can be determined; based on the preset time period and the total number of quality differences, the first time period in which the surveyed user has quality difference events can be determined to construct a fourth target variable; the third target variable and the fourth target variable are used as training data to train the model to be trained to obtain a satisfaction prediction model.

[0046] Step 104: using the satisfaction prediction model, determine the satisfaction prediction value of the user to be predicted.

[0047] In some embodiments, the third parameter value of the target satisfaction influencing parameter of the user to be predicted is obtained and input into the satisfaction prediction, thereby using the satisfaction prediction model to determine the satisfaction prediction value to be predicted.

[0048] In some optional embodiments, based on the target satisfaction influencing parameter, a third parameter value of the user to be predicted and a second time parameter corresponding to the third parameter value are obtained; based on the third parameter value and the second time parameter, a second time period in which a quality-poor event occurs for the user to be predicted is determined; based on the second time period, a satisfaction prediction model is used to determine a satisfaction prediction value.

[0049] The poor quality event refers to an event corresponding to the third parameter exceeding a preset threshold value.

[0050] Step 105: cluster the predicted dissatisfied users among the users to be predicted based on the satisfaction prediction values ​​to obtain a clustering result of the predicted dissatisfied users.

[0051] In some embodiments, predicted dissatisfied users among the users to be predicted may be determined according to the satisfaction prediction values, and then the predicted dissatisfied users may be clustered by obtaining a network parameter set of the predicted dissatisfied users.

[0052] In some optional embodiments, a predicted user whose predicted satisfaction value is less than a second preset value is determined as a predicted dissatisfied user; a network parameter set of the predicted dissatisfied user and a target poor quality event of the predicted dissatisfied user are obtained; based on the network parameter set, a service grid of the predicted dissatisfied user is determined; based on the residence time of the predicted dissatisfied user in the service grid, a user scenario library of the predicted dissatisfied user is determined, the user scenario library including at least one of the following: a permanent workplace, a permanent residence, and a user activity route; based on the user scenario library, scenario attributes are marked for the target poor quality event; and a linear regression algorithm is used to cluster at least one of the scenario attributes, the target poor quality event, and the residence event to obtain a clustering result of the predicted dissatisfied user.

[0053] Furthermore, in some embodiments, the network parameter set includes at least one of the following: predicted serving cell of the dissatisfied user, predicted neighboring cell signal strength of the dissatisfied user, and timing advance TA of the signal in the serving cell.

[0054] In summary, according to the user clustering method proposed in the present disclosure, the method includes: obtaining the first satisfaction evaluation data of the surveyed user and the first parameter value of the first satisfaction influencing parameter; determining the target satisfaction influencing parameter based on the first satisfaction evaluation data and the first parameter value; constructing a satisfaction prediction model based on the target satisfaction influencing parameter and the first satisfaction evaluation data; determining the satisfaction prediction value of the user to be predicted using the satisfaction prediction model; clustering the predicted dissatisfied users among the users to be predicted based on the satisfaction prediction value to obtain the clustering results of the predicted dissatisfied users. The method disclosed in the present disclosure, by determining the target satisfaction influencing parameter, eliminates the redundant first satisfaction influencing parameter, and improves the computational efficiency and accuracy of the satisfaction prediction model; and by clustering the predicted dissatisfied users among the users to be predicted through the satisfaction prediction value, the accuracy of the clustering results can be improved, which is convenient for improving the network quality of the users.

[0055] Figure 2 A flowchart of a user clustering method proposed in the present disclosure is further shown. Figure 1 The embodiment shown further explains step 102. Figure 2 The following steps are included.

[0056] Step 201, performing Boolean transformation on the first satisfaction evaluation data to obtain the first target variable.

[0057] In some embodiments, since the first satisfaction evaluation data can be converted into satisfied and dissatisfied data, in order to facilitate calculation, the satisfaction results in the first satisfaction evaluation data can be subjected to a Boolean conversion. For example, the satisfied data can be set to 1, and the dissatisfied data can be set to 0. The converted Boolean value can be combined with the surveyed user to generate the first target variable.

[0058] Among them, the first target variable can be expressed as follows:

[0059]

[0060] Step 202: Determine the surveyed users whose first satisfaction evaluation data is greater than a first preset value as high-satisfaction users.

[0061] In some embodiments, users with high satisfaction scores are determined from the surveyed users, so that the parameters that have a greater impact on user satisfaction among the first satisfaction influencing parameters can be determined based on the users with high satisfaction scores.

[0062] In some embodiments, survey users whose satisfaction scores in the first satisfaction evaluation data are greater than a first preset value may be determined as high-satisfaction score users.

[0063] Step 203: Determine, from the first parameter values, a second parameter value corresponding to the user with a high satisfaction score.

[0064] In some embodiments, the parameter value corresponding to the user with a high satisfaction score in the first parameter value can be determined based on the user with a high satisfaction score, and the corresponding parameter value can be determined as the second parameter value.

[0065] Step 204: Determine a first parameter threshold of a first satisfaction influencing parameter based on the second parameter value.

[0066] In some embodiments, a confidence interval of the first satisfaction influencing parameter may be determined based on the second parameter value, and then the first parameter threshold may be determined according to the confidence interval.

[0067] For example, taking the existence of m first satisfaction influencing parameters as an example, the second parameter value can be used to calculate the confidence interval of the 95% confidence level of the above m first satisfaction influencing parameters, and this interval is used as the threshold of the first satisfaction influencing parameter indicator.

[0068] Furthermore, the first parameter threshold may be indicated by the following formula:

[0069]

[0070] Among them, THR represents the first parameter threshold, KQI_Low represents the lower limit value of the first parameter threshold, and KQI_High represents the upper limit value of the first parameter threshold.

[0071] Step 205: Determine a second target variable based on the first parameter value and the first parameter threshold.

[0072] In some embodiments, the first satisfaction influencing parameter corresponding to the second parameter value exceeding the parameter threshold can be determined as the first quality difference parameter, and the first satisfaction influencing parameter corresponding to the second parameter value not exceeding the parameter threshold can be determined as the second quality difference parameter; then, a Boolean matrix is ​​constructed using the first quality difference parameter and the second quality difference parameter to obtain the second target variable.

[0073] In some embodiments, a Boolean matrix is ​​constructed using the first quality difference parameter and the second quality difference parameter, and the first quality difference parameter can be marked as 1 and the second quality difference parameter can be marked as 0; or, the first quality difference parameter can be marked as 0 and the second quality difference parameter can be marked as 1, thereby representing the above-mentioned 0 and 1 values ​​in the form of a matrix.

[0074] Furthermore, the second target variable can be expressed by the following formula:

[0075]

[0076] in, Label_i,j represents the second parameter value.

[0077] Step 206: Determine a target satisfaction influencing parameter based on the first target variable and the second target variable.

[0078] In some embodiments, the first contribution weight of the first satisfaction influencing parameter can be determined by using the first target variable and the second target variable and utilizing the random forest algorithm; the parameter correlation between the first satisfaction influencing parameters is determined to obtain a correlation coefficient matrix; based on the correlation coefficient matrix, the second satisfaction influencing parameter is determined from the first satisfaction influencing parameter, wherein the correlation between the second satisfaction influencing parameters is less than or equal to a preset correlation threshold parameter; based on the correlation coefficient matrix and the first contribution weight, the second contribution weight of the second satisfaction influencing parameter is determined; based on the second satisfaction influencing parameter and the second contribution weight, the target satisfaction influencing parameter is determined.

[0079] In an optional embodiment, a plurality of data pairs can be randomly selected from the first target variable and the second target variable with replacement as subsets by the Bootstrap sampling method, and each subset is used to train a decision tree. When constructing each decision tree, only some features are randomly selected for split selection for each split. This step is implemented by the "random subspace method", that is, a subset is randomly selected from all available features for each split, and then the optimal feature is selected on this subset for splitting. A decision tree is constructed using the selected feature set and the training data set. The decision tree construction process adopts the "recursive splitting" method, that is, the data set is divided into two or more subsets until the samples in the subset belong to the same category or a predetermined stopping condition is reached. By repeating the above-mentioned sample sampling and feature sampling steps, as well as the decision tree construction steps, multiple decision trees are constructed, and finally the importance of the feature is output through multiple decision trees, where the importance is the first contribution weight.

[0080] Among them, the first contribution weight can be expressed by the following formula:

[0081] Feature_importance=[Importance_KQI1...Importance_KQIm] (Formula 4)

[0082] Among them, Feature_importance represents the first contribution weight, and Importance_KQIm represents the first contribution weight of the first satisfaction influencing parameter m.

[0083] In some embodiments, the correlation between the two first satisfaction influencing parameters can be determined by means of chi-square test, calculation of information entropy, etc.

[0084] Among them, the correlation coefficient matrix can be expressed by the following formula:

[0085]

[0086] Among them, Corr represents the correlation coefficient matrix, and Corrmm represents the parameter correlation between the first satisfaction influencing parameter m and the first satisfaction influencing parameter m.

[0087] In some embodiments, whether the first satisfaction influencing parameters are strongly correlated can be determined based on a preset correlation threshold parameter. When two first satisfaction influencing parameters are greater than or equal to the preset correlation threshold parameter, the two first satisfaction influencing parameters can be considered to be strongly correlated. At this time, the first contribution weights of the two first satisfaction influencing parameters are compared, the first satisfaction influencing parameters with larger first contribution weight values ​​are retained, and the first satisfaction influencing parameters with smaller first contribution weight values ​​are eliminated, and the retained first satisfaction influencing parameters are determined as the second satisfaction influencing parameters. At the same time, the first contribution weights corresponding to the two first satisfaction influencing parameters are summed to determine the summation result as the second contribution weight. It should be understood that the second contribution weight can also be determined by weighted summation and other methods, and the present disclosure is not limited to this.

[0088] The second satisfaction influence parameter and the second contribution weight can be expressed by the following formula:

[0089] NewKQI=[KQI1...KQIk] (Formula 6)

[0090] New_importance=[Importance_KQI1...Importance_KQIm] (Formula 7)

[0091] Among them, NewKQI represents the second satisfaction influence parameter, and New_importance represents the second contribution weight.

[0092] In some embodiments, the second satisfaction influencing parameters may be sorted in descending order according to the size of the second contribution weight, thereby determining the target satisfaction influencing parameters.

[0093] Optionally, the first i values ​​of the second satisfaction influence parameter may be determined as the target satisfaction influence parameter, wherein the sum of the first i values ​​is greater than a preset value (eg, 0.8).

[0094] For example, the following table is used as an example:

[0095]

[0096]

[0097] Among them, the first i parameters are taken as target satisfaction influencing parameters, and the sum of the second contribution weights of the first i parameters is greater than 0.8.

[0098] Then the target satisfaction influence weight can be determined as:

[0099]

[0100] In summary, the user clustering method proposed in the present disclosure includes: performing a Boolean transformation on the first satisfaction evaluation data to obtain a first target variable; determining the survey users whose first satisfaction evaluation data is greater than a first preset value as high-satisfaction users; determining the second parameter value corresponding to the high-satisfaction users from the first parameter value; determining the first parameter threshold of the first satisfaction influencing parameter based on the second parameter value; determining the second target variable based on the first parameter value and the first parameter threshold; determining the target satisfaction influencing parameter based on the first target variable and the second target variable. The method disclosed in the present disclosure improves the computational efficiency and accuracy of the satisfaction prediction model by determining the parameter correlation and the first contribution weight in the first satisfaction influencing parameter, thereby eliminating the redundant data and the first satisfaction influencing parameter.

[0101] Figure 3 A flowchart of a user clustering method proposed in the present disclosure is further shown. Figure 1 The embodiment shown further explains step 103. Figure 3 The following steps are included.

[0102] Step 301: construct a third target variable based on the surveyed users and the first satisfaction evaluation data.

[0103] In some embodiments, the satisfaction scores of the surveyed users and the first satisfaction evaluation data may be sorted to construct a third target variable.

[0104] In some embodiments, the third target variable may be determined by the following formula:

[0105]

[0106] Among them, Score_i={1,2,3,4,5,6,7,8,9,10}.

[0107] Step 302: Obtain a first time parameter corresponding to a first parameter value.

[0108] In some embodiments, the first time parameter corresponding to the first parameter value is obtained so as to determine the number of times the poor quality events occur to the surveyed user in each time period.

[0109] In some embodiments, the first time parameter is the time at which the first parameter value is generated.

[0110] Step 303: Determine a third quality difference parameter of the surveyed user based on the first parameter value and a preset parameter threshold.

[0111] In some embodiments, the preset parameter threshold may be the same as or different from the first parameter threshold in step 205, which is not limited in the present disclosure.

[0112] In some embodiments, based on the first time parameter, the number of times that poor quality events occur in the surveyed users in a certain time period is determined, and the number is determined as the third poor quality parameter.

[0113] When the first parameter value exceeds the preset parameter threshold value, it is determined that there is a quality difference number.

[0114] In an optional embodiment, the relevant parameters of the surveyed user can be expressed by the following formula:

[0115]

[0116] in,

[0117] In some embodiments, the time period may be in units of hours or other time lengths, which is not limited in the present disclosure.

[0118] Step 304: Determine the total number of times the quality of the surveyed user is poor within a preset time period based on the third quality difference parameter and the first time parameter.

[0119] For example, when the third quality difference parameter is determined in hours in step 303, the total number of quality differences can be determined in days (i.e., the above-mentioned preset time period). It should be understood that the present disclosure does not limit the specific time of the preset time period.

[0120] For example, in units of hours, the number of poor quality events generated by all target satisfaction influencing parameters is summed up to obtain the total number of poor quality events in the period. For example, a table of poor quality event times with hourly granularity in units of days is generated, which can be expressed by the following formula:

[0121]

[0122] Step 305, based on the preset time period and the total number of poor quality events, determine the first time period in which the surveyed user has poor quality events, so as to construct a fourth target variable.

[0123] In some embodiments, the time period during which the poor quality events occurred for the surveyed users, ie, the first time period, can be determined based on the total number of poor quality events and a preset time period.

[0124] Then, according to the number of surveyed users and the first time period, a fourth target variable is constructed.

[0125] For example, the fourth target variable can be constructed by the following formula:

[0126]

[0127] Step 306: Use the third target variable and the fourth target variable as training data to train the model to be trained to obtain a satisfaction prediction model.

[0128] In some embodiments, the third target variable and the fourth target variable may be used as evaluation data sets to train the model to be trained to obtain a satisfaction prediction model.

[0129] For example, the third target variable and the fourth target variable are used as evaluation data sets, and a multi-layer perceptron (MLP) machine learning algorithm is constructed to perform machine learning on the training model, wherein a sequential model can be used to construct a deep learning network, which can include two hidden layers and use a ReLU activation function.

[0130] Furthermore, the Adam optimizer and the mean square error loss function can be used to compile the model. By using 20% ​​of the data (the third target variable and the fourth target variable) as the validation set, the performance of the satisfaction prediction model is evaluated and the mean square error is calculated. The relationship between user satisfaction and the number of occurrences of poor quality time periods is obtained as follows:

[0131]

[0132]

[0133] The satisfaction prediction model is determined as follows:

[0134]

[0135] In other words, the satisfaction prediction model can predict the user's satisfaction score based on the number of quality difference segments of the user.

[0136] In summary, the user clustering method proposed in the present disclosure includes: constructing a third target variable based on the surveyed users and the first satisfaction evaluation data; obtaining the first time parameter corresponding to the first parameter value; determining the third quality difference parameter of the surveyed users based on the first parameter value and the preset parameter threshold; determining the total number of quality differences of the surveyed users within the preset time period based on the third quality difference parameter and the first time parameter; determining the first time period in which the surveyed users have quality difference events based on the preset time period and the total number of quality differences to construct the fourth target variable; using the third target variable and the fourth target variable as training data to train the training model to obtain the satisfaction prediction model. The method disclosed in the present disclosure constructs a satisfaction prediction model by determining the first time period corresponding to the target satisfaction influencing parameter and the satisfaction score of the surveyed users, so that the satisfaction prediction model can deeply understand the relationship between time, parameters and satisfaction, thereby improving the accuracy of the model prediction results.

[0137] Figure 4 A flowchart of a user clustering method proposed in the present disclosure is further shown. Figure 1 The embodiment shown further explains step 104. Figure 4 The following steps are included.

[0138] Step 401: Based on the target satisfaction influencing parameter, a third parameter value of the user to be predicted and a second time parameter corresponding to the third parameter value are obtained.

[0139] In some embodiments, parameters included in the target satisfaction influencing parameters are used as a reference to obtain parameters for the user to be predicted.

[0140] In other words, the third parameter value is the parameter value of the parameter affecting the target satisfaction of the user to be predicted; the second time parameter is the time point corresponding to when the user to be predicted generates the third parameter value.

[0141] Step 402: Based on the third parameter value and the second time parameter, determine a second time period during which the user to be predicted has a poor quality event.

[0142] In this embodiment, whether a poor quality event occurs is determined by using the third parameter value and the preset parameter threshold. In other words, when the third parameter value exceeds the preset parameter threshold, it can be recorded that a poor quality event occurs for the parameter.

[0143] Furthermore, according to the second time parameter and the poor quality event, a second time period is determined. The second time period is an event period in which the poor quality event exists. For details, please refer to the relevant descriptions and embodiments in steps 303 to 305, which will not be repeated here.

[0144] Step 403: Based on the second time period, a satisfaction prediction value is determined using a satisfaction prediction model.

[0145] In some embodiments, based on the second time period, the total number of times the to-be-predicted user has poor quality time is determined, thereby using a satisfaction prediction model to determine a satisfaction prediction value.

[0146] For example, assuming that the number of poor quality time periods is 7, and the satisfaction prediction model is illustrated by taking the model in step 306 as an example, the predicted satisfaction of the user to be predicted is 4.

[0147] In summary, the user clustering method proposed in the present disclosure includes: obtaining the third parameter value of the user to be predicted and the second time parameter corresponding to the third parameter value based on the target satisfaction influencing parameter; determining the second time period in which the user to be predicted has a poor quality event based on the third parameter value and the second time parameter; and determining the satisfaction prediction value based on the second time period using the satisfaction prediction model. The method disclosed in the present disclosure can achieve the satisfaction prediction value of the user to be predicted by obtaining the third parameter value of the user to be predicted and the second time parameter corresponding to the third parameter value, and using the satisfaction prediction model, thereby achieving the prediction of user satisfaction, so as to improve the user's network before the user complains and improve the user's service quality.

[0148] Figure 5 A flowchart of a user clustering method proposed in the present disclosure is further shown. Figure 1 The embodiment shown further explains step 105. Figure 5 The following steps are included.

[0149] Step 501: Determine the predicted user whose satisfaction prediction value is less than the second preset value as a predicted dissatisfied user.

[0150] In some embodiments, the second preset value may be the same as the first preset value, or may be different from the first preset value, which is not limited in the present disclosure.

[0151] Step 502, obtaining a network parameter set for predicting dissatisfied users and a target quality-poor event for predicting dissatisfied users.

[0152] In some embodiments, the network parameter set includes at least one of the following: predicted serving cell of the dissatisfied user, predicted neighboring cell signal strength of the dissatisfied user, and timing advance TA of the signal in the serving cell.

[0153] In some embodiments, the target poor quality event refers to a poor quality event corresponding to a target satisfaction influencing parameter that predicts dissatisfied users.

[0154] Step 503: Determine a service grid for predicted dissatisfied users based on the network parameter set.

[0155] For example, the location of the predicted unsatisfactory user may be determined based on the service cell signal strength and the neighboring cell signal strength in the network parameter set, and the unsatisfactory service grid (micro grid) may be determined based on the location.

[0156] Step 504: Determine a user scenario library of predicted dissatisfied users based on the predicted residence time of dissatisfied users in the service grid.

[0157] In some embodiments, the user scenario library includes at least one of the following: a permanent workplace, a permanent residence, and a user activity route.

[0158] In some embodiments, the user scenario may be determined based on the predicted residence time of dissatisfied users in different service grids.

[0159] For example, the residence time of the user in each micro-grid from 9:00 to 17:00 on weekdays (Monday to Friday) is calculated. When the average residence time of the user in the micro-grid is greater than 4 hours per day, the micro-grid is the user's permanent workplace.

[0160] For example, the residence time of the user in each micro-grid from 22:00 to 6:00 the next day is calculated. When the average residence time of the user in the micro-grid is greater than 4 hours per day, the micro-grid is considered to be a permanent residence.

[0161] For example, in addition to the permanent workplace and permanent residence, the sequence of microgrids occupied by the user within a certain period of time every day is calculated. If the same sequence appears for more than 3 days a week, the sequence is determined to be a frequently used activity route.

[0162] Step 505: Mark scene attributes for the target poor quality event based on the user scene library.

[0163] In some embodiments, by determining the location where the target poor quality event occurs (ie, each scene in the user scene library), the poor quality event is assigned a scene attribute.

[0164] Step 506: cluster at least one of the scene attribute, the target poor quality event, and the dwelling event using a linear regression algorithm to obtain a clustering result for predicting dissatisfied users.

[0165] In some embodiments, the poor quality events of the predicted dissatisfied users are clustered through a linear regression algorithm to obtain cluster analysis of different dimensions: such as clustering by location scenario dimension: clustering of poor quality events at work, clustering of poor quality events at residence, clustering of poor quality events on common routes, etc.; clustering by poor quality events: clustering of poor voice quality, clustering of poor Internet access, etc.; clustering by satisfaction score; clustering by poor quality time, etc. Through clustering, the distribution of dissatisfied users in various dimensions is presented in detail, realizing the transformation of the solution granularity from "single problem cell" to "key areas and key scenarios".

[0166] In summary, the user clustering method proposed in the present disclosure includes: determining the predicted user whose satisfaction prediction value is less than the second preset value as a predicted dissatisfied user; obtaining the network parameter set of the predicted dissatisfied user and the target poor quality event of the predicted dissatisfied user; determining the service grid of the predicted dissatisfied user based on the network parameter set; determining the user scenario library of the predicted dissatisfied user based on the residence time of the predicted dissatisfied user in the service grid, the user scenario library includes at least one of the following: permanent workplace, permanent residence and user activity route; marking the scenario attribute for the target poor quality event based on the user scenario library; clustering at least one of the scenario attribute, target poor quality event and residence event using a linear regression algorithm to obtain the clustering result of the predicted dissatisfied user. After clustering by scenario, the method disclosed in the present disclosure can be optimized according to the clustered scenario, and can be cross-system, cross-frequency band, and cover all cells involved in the scenario, providing an overall solution, truly solving network problems and improving user satisfaction.

[0167] Therefore, the present disclosure has the following beneficial effects:

[0168] 1. The method disclosed in the present invention, by determining the target satisfaction influencing parameter, eliminates the redundant first satisfaction influencing parameter, thereby improving the computational efficiency and accuracy of the satisfaction prediction model; and by clustering the predicted dissatisfied users among the predicted users through the satisfaction prediction value, the accuracy of the clustering results can be improved, which is convenient for improving the user's network quality.

[0169] 2. The method disclosed in the present invention constructs a satisfaction prediction model by determining the first time period corresponding to the target satisfaction influencing parameter and investigating the user's satisfaction score, so that the satisfaction prediction model can deeply understand the relationship between time, parameters and satisfaction, thereby improving the accuracy of the model prediction results and making the model prediction results closer to the user's actual perception.

[0170] 3. The method disclosed in the present invention clusters the user quality problem points into multiple workplaces, residences, and common activity routes to form a user quality problem scenario library. The quality problems of dissatisfied users are mapped to the corresponding scenario library to obtain the overall service aggregation of dissatisfied users and the regional distribution of overall quality problems, thus realizing the transformation of the solution granularity from "single problem cell" to "key areas and key scenarios".

[0171] Figure 6 FIG. 6 is a schematic diagram of a user clustering device 600 provided in an embodiment of the present disclosure. Figure 6 As shown, the user clustering device includes:

[0172] An acquisition unit 610 is used to acquire first satisfaction evaluation data of a surveyed user and a first parameter value of a first satisfaction influencing parameter;

[0173] A first determining unit 620, configured to determine a target satisfaction influencing parameter based on the first satisfaction evaluation data and the first parameter value;

[0174] A construction unit 630, configured to construct a satisfaction prediction model based on the target satisfaction influencing parameter and the first satisfaction evaluation data;

[0175] The second determining unit 640 is used to determine the satisfaction prediction value of the user to be predicted by using the satisfaction prediction model;

[0176] The clustering unit 650 is used to cluster the target poor quality events of the predicted dissatisfied users among the predicted users based on the satisfaction prediction values, so as to obtain the clustering results of the predicted dissatisfied users.

[0177] In some embodiments of the present disclosure, the first determination unit 620 is further used to perform a Boolean conversion on the first satisfaction evaluation data to obtain a first target variable; determine a survey user whose first satisfaction evaluation data is greater than a first preset value as a high-satisfaction user; determine a second parameter value corresponding to the high-satisfaction user from the first parameter value; determine a first parameter threshold of a first satisfaction influencing parameter based on the second parameter value; determine a second target variable based on the first parameter value and the first parameter threshold; and determine a target satisfaction influencing parameter based on the first target variable and the second target variable.

[0178] In some embodiments of the present disclosure, the first determination unit 620 is also used to determine the first satisfaction influencing parameter corresponding to the second parameter value exceeding the parameter threshold as the first quality difference parameter; determine the first satisfaction influencing parameter corresponding to the second parameter value not exceeding the parameter threshold as the second quality difference parameter; construct a Boolean matrix based on the first quality difference parameter and the second quality difference parameter; and determine the Boolean matrix as the second target variable.

[0179] In some embodiments of the present disclosure, the first determination unit 620 is also used to determine the first contribution weight of the first satisfaction influencing parameter based on the first target variable and the second target variable using a random forest algorithm; determine the parameter correlation between the first satisfaction influencing parameters to obtain a correlation coefficient matrix; determine the second satisfaction influencing parameter from the first satisfaction influencing parameters based on the correlation coefficient matrix, wherein the correlation between the second satisfaction influencing parameters is less than or equal to a preset correlation threshold parameter; determine the second contribution weight of the second satisfaction influencing parameter based on the correlation coefficient matrix and the first contribution weight; and determine the target satisfaction influencing parameter based on the second satisfaction influencing parameter and the second contribution weight.

[0180] In some embodiments of the present disclosure, the construction unit 630 is also used to construct a third target variable based on the surveyed user and the first satisfaction evaluation data; obtain the first time parameter corresponding to the first parameter value; determine the third quality difference parameter of the surveyed user based on the first parameter value and the preset parameter threshold; determine the total number of quality differences of the surveyed user within a preset time period based on the third quality difference parameter and the first time parameter; determine the first time period in which the surveyed user has quality difference events based on the preset time period and the total number of quality differences to construct a fourth target variable; use the third target variable and the fourth target variable as training data to train the model to be trained to obtain a satisfaction prediction model.

[0181] In some embodiments of the present disclosure, the second determination unit 640 is also used to obtain a third parameter value of the user to be predicted and a second time parameter corresponding to the third parameter value based on the target satisfaction influencing parameter; determine a second time period in which a quality difference event occurs for the user to be predicted based on the third parameter value and the second time parameter; and determine a satisfaction prediction value based on the second time period using a satisfaction prediction model.

[0182] In some embodiments of the present disclosure, the clustering unit 650 is further used to determine a predicted user whose satisfaction prediction value is less than a second preset value as a predicted dissatisfied user; obtain a network parameter set of the predicted dissatisfied user and a target poor quality event of the predicted dissatisfied user; determine a service grid of the predicted dissatisfied user based on the network parameter set; determine a user scenario library of the predicted dissatisfied user based on the predicted residence time of the dissatisfied user in the service grid, the user scenario library including at least one of the following: a permanent workplace, a permanent residence, and a user activity route; based on the user scenario library, mark scenario attributes for the target poor quality event; and cluster at least one of the scenario attributes, the target poor quality event, and the residence event using a linear regression algorithm to obtain a clustering result of the predicted dissatisfied user.

[0183] In some embodiments of the present disclosure, the network parameter set includes at least one of the following: predicted serving cell of dissatisfied users, predicted neighboring cell signal strength of dissatisfied users, and timing advance TA of signals in the serving cell.

[0184] In summary, the user clustering device proposed in the present disclosure includes: an acquisition unit for acquiring the first satisfaction evaluation data of the surveyed user and the first parameter value of the first satisfaction influencing parameter; a first determination unit for determining the target satisfaction influencing parameter based on the first satisfaction evaluation data and the first parameter value; a construction unit for constructing a satisfaction prediction model based on the target satisfaction influencing parameter and the first satisfaction evaluation data; a second determination unit for determining the satisfaction prediction value of the user to be predicted using the satisfaction prediction model; and a clustering unit for clustering the target poor quality events of the predicted dissatisfied users among the predicted users based on the satisfaction prediction value to obtain the clustering results of the predicted dissatisfied users. The device of the present disclosure, by determining the target satisfaction influencing parameter, eliminates the redundant first satisfaction influencing parameter, and improves the computational efficiency and accuracy of the satisfaction prediction model; and by clustering the predicted dissatisfied users among the predicted users through the satisfaction prediction value, the accuracy of the clustering results can be improved, which is convenient for improving the network quality of the users.

[0185] It should be noted that: when the user clustering device provided in the above embodiment performs user clustering, it only uses the division of the above-mentioned program modules as an example. In actual applications, the above-mentioned processing can be assigned to different program modules as needed, that is, the internal structure of the user clustering device is divided into different program modules to complete all or part of the processing described above.

[0186] Since the device provided in the embodiment of the present disclosure corresponds to the methods provided in the above-mentioned embodiments, the implementation of the method is also applicable to the device provided in the embodiment and will not be described in detail in this embodiment.

[0187] In the embodiments provided in the present application, the methods and devices provided in the embodiments of the present application are introduced. In order to implement the functions in the methods provided in the embodiments of the present application, the electronic device may include a hardware structure and a software module, and implement the functions in the form of a hardware structure, a software module, or a hardware structure plus a software module. A function of the functions may be executed in the form of a hardware structure, a software module, or a hardware structure plus a software module.

[0188] Figure 7 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present disclosure is shown in FIG. Figure 7As shown, the electronic device 700 includes at least one processor 702; and a memory 701 communicatively connected to the at least one processor 702; wherein the memory 701 stores instructions executable by the at least one processor 702, and the instructions are executed by the at least one processor 702 to implement the steps of the user clustering method described in the embodiment of the present disclosure; or, the instructions are executed by the at least one processor 702 to implement the steps of the user clustering method described in the embodiment of the present disclosure.

[0189] It is understood that the electronic device also includes a communication interface. The various components in the electronic device are coupled together through a bus system. It is understood that the bus system is used to achieve connection and communication between these components. In addition to the data bus, the bus system also includes a power bus, a control bus and a status signal bus.

[0190] It can be understood that the memory 701 can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory can be a disk memory or a tape memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct RAMbus random access memory (DRRAM).The memory 701 described in this embodiment of the present solution is intended to include, but is not limited to, these and any other suitable types of memories.

[0191] The method disclosed in the above embodiment of the present disclosure may be applied to the processor 702, or implemented by the processor 702. The processor 702 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method may be completed by an integrated logic circuit of hardware in the processor 702 or an instruction in the form of software.

[0192] The processor 702 described above may be a general-purpose processor, a DSP, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 702 may implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this solution. A general-purpose processor may be a microprocessor or any conventional processor, etc. In combination with the steps of the method disclosed in the embodiments of this solution, it may be directly embodied as being executed by a hardware decoding processor, or executed by a combination of hardware and software modules in a decoding processor. The software module may be located in a storage medium, which is located in the memory 701, and the processor 702 reads the information in the memory 701 and completes the steps of the aforementioned method in combination with its hardware.

[0193] In an exemplary embodiment, the electronic device may be implemented by one or more application specific integrated circuits (ASIC), DSP, programmable logic device (PLD), complex programmable logic device (CPLD), FPGA, general purpose processor, controller, MCU, microprocessor, or other electronic components to execute the aforementioned method.

[0194] The embodiment of the present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to implement the steps of the user clustering method described in the embodiment of the present solution when the computer instructions are executed.

[0195] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0196] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples" or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiments or examples are included in at least one embodiment or example of the present solution. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0197] Any process or method description in the flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiment of the present scheme includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present scheme belong.

[0198] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processing module, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (control method), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or otherwise processing in a suitable manner if necessary, and then stored in a computer memory.

[0199] It should be understood that the various parts of the implementation of the present solution can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned implementation, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another implementation, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0200] A person skilled in the art may understand that all or part of the steps in the above-mentioned embodiment method may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0201] In addition, each functional unit in each embodiment of the present solution can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into one module. The above-mentioned integrated module can be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. The above-mentioned storage medium can be a read-only memory, a disk or an optical disk, etc.

[0202] Although the implementation methods of the present scheme have been shown and described above, it can be understood that the above implementation methods are exemplary and cannot be understood as limitations on the present scheme. Ordinary technicians in this field can change, modify, replace and modify the above implementation methods within the scope of the present scheme.

Claims

1. A user clustering method, characterized in that: The method comprises: Obtaining first satisfaction evaluation data of surveyed users and a first parameter value of a first satisfaction influencing parameter; Determining a target satisfaction influencing parameter based on the first satisfaction evaluation data and the first parameter value; Constructing a satisfaction prediction model based on the target satisfaction influencing parameter and the first satisfaction evaluation data; Determine the predicted satisfaction value of the user to be predicted by using the satisfaction prediction model; Based on the predicted satisfaction value, the predicted dissatisfied users among the users to be predicted are clustered to obtain a clustering result of the predicted dissatisfied users.

2. The method according to claim 1, characterized in that The determining of the target satisfaction influencing parameter based on the first satisfaction evaluation data and the first parameter value includes: Performing Boolean transformation on the first satisfaction evaluation data to obtain a first target variable; Determine the surveyed user whose first satisfaction evaluation data is greater than a first preset value as a high-satisfaction user; Determine, from the first parameter values, a second parameter value corresponding to the user with a high satisfaction score; Determining a first parameter threshold of the first satisfaction influencing parameter based on the second parameter value; Determining a second target variable based on the first parameter value and the first parameter threshold; Based on the first target variable and the second target variable, the target satisfaction influencing parameter is determined.

3. The method according to claim 2, characterized in that The determining the second target variable based on the first parameter value and the parameter threshold comprises: Determine the first satisfaction influencing parameter corresponding to the second parameter value exceeding the parameter threshold as the first quality difference parameter; Determine the first satisfaction influencing parameter of the second parameter value that does not exceed the parameter threshold as the second quality difference parameter; constructing a Boolean matrix based on the first quality difference parameter and the second quality difference parameter; The Boolean matrix is ​​determined as the second target variable.

4. The method according to claim 2, characterized in that: The determining the target satisfaction influencing parameter based on the first target variable and the second target variable includes: Based on the first target variable and the second target variable, using a random forest algorithm, determining a first contribution weight of the first satisfaction influencing parameter; determining parameter correlations between the first satisfaction influencing parameters to obtain a correlation coefficient matrix; Based on the correlation coefficient matrix, determining a second satisfaction influencing parameter from the first satisfaction influencing parameters, wherein the correlation between the second satisfaction influencing parameters is less than or equal to a preset correlation threshold parameter; Determining a second contribution weight of the second satisfaction influencing parameter based on the correlation coefficient matrix and the first contribution weight; The target satisfaction influence parameter is determined based on the second satisfaction influence parameter and the second contribution weight.

5. The method according to claim 1, characterized in that The constructing of a satisfaction prediction model based on the target satisfaction influencing parameter and the first satisfaction evaluation data includes: Constructing a third target variable based on the surveyed users and the first satisfaction evaluation data; Obtaining a first time parameter corresponding to the first parameter value; Determining a third quality difference parameter of the surveyed user based on the first parameter value and a preset parameter threshold; Based on the third quality difference parameter and the first time parameter, determining the total number of quality differences of the surveyed user within a preset time period; Based on the total number of poor quality events in the preset time period, determining a first time period in which the surveyed user has poor quality events, so as to construct a fourth target variable; The third target variable and the fourth target variable are used as training data to train the model to be trained, so as to obtain the satisfaction prediction model.

6. The method according to claim 1, characterized in that Determining the predicted satisfaction value of the user to be predicted by using the satisfaction prediction model includes: Based on the target satisfaction influencing parameter, obtaining a third parameter value of the to-be-predicted user and a second time parameter corresponding to the third parameter value; Based on the third parameter value and the second time parameter, determine a second time period in which the to-be-predicted user has a poor quality event; Based on the second time period, the satisfaction prediction value is determined using the satisfaction prediction model.

7. The method according to claim 1, characterized in that The clustering of predicted dissatisfied users among the users to be predicted based on the satisfaction prediction value to obtain the clustering result of the predicted dissatisfied users includes: Determine the predicted user whose satisfaction prediction value is less than the second preset value as a predicted dissatisfied user; Acquire the network parameter set of the predicted dissatisfied user and the target poor quality event of the predicted dissatisfied user; Determining a service grid for the predicted dissatisfied user based on the network parameter set; Determine a user scenario library of the predicted dissatisfied user based on the predicted residence time of the dissatisfied user in the service grid, wherein the user scenario library includes at least one of the following: a permanent workplace, a permanent residence, and a user activity route; Based on the user scenario library, marking scenario attributes for the target poor quality event; A linear regression algorithm is used to cluster at least one of the scene attribute, the target poor quality event, and the resident event to obtain a clustering result of the predicted dissatisfied users.

8. The method according to claim 7, characterized in that The network parameter set includes at least one of the following: the serving cell of the predicted dissatisfied user, the signal strength of the neighboring cell of the predicted dissatisfied user, and the time advance TA of the signal in the serving cell.

9. A user clustering device, characterized in that: The device comprises: An acquisition unit, used for acquiring first satisfaction evaluation data of a surveyed user and a first parameter value of a first satisfaction influencing parameter; A first determining unit, configured to determine a target satisfaction influencing parameter based on the first satisfaction evaluation data and the first parameter value; A construction unit, configured to construct a satisfaction prediction model based on the target satisfaction influencing parameter and the first satisfaction evaluation data; A second determining unit, configured to determine a satisfaction prediction value of a user to be predicted by using the satisfaction prediction model; A clustering unit is used to cluster the target poor quality events of the predicted dissatisfied users among the users to be predicted based on the satisfaction prediction value, so as to obtain the clustering results of the predicted dissatisfied users.

10. An electronic device, characterized in that: include: a processor and a memory for storing a computer program capable of being executed on the processor, Wherein, when the processor is used to run the computer program, it executes the method described in any one of claims 1-8.

Images