Prediction method, device and computing equipment for main factors affecting user satisfaction

Abstract

Embodiments of the present invention relate to the field of wireless communication technology and disclose a method, apparatus, and computing device for predicting the main factors affecting user satisfaction. The method comprises: obtaining user consumption data and user network data and extracting features related to satisfaction; predicting satisfaction based on the features and outputting the user's satisfaction; for dissatisfied users, calculating the correlation coefficient between each feature and the satisfaction level and the distance between the correlation coefficient and the individual central mean; and outputting the features that influence user satisfaction based on the distance. Through the above-described approach, embodiments of the present invention can scientifically assess the impact of factors such as consumption behavior and network quality on user satisfaction.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of wireless communication technology, and in particular to a method, apparatus, and computing device for predicting main factors affecting user satisfaction. Background Art

[0002] User satisfaction is a combination of a customer's personal service needs, past service experiences, and the reputation of a company's services among their surroundings. In current mobile networks, predictions of dissatisfied users are primarily based on their perception of network signaling services. The primary drivers of user dissatisfaction remain abnormal signaling on the network side.

[0003] Satisfaction is a subjective feeling of customers and is easily affected by various factors. Currently, the causes of user dissatisfaction, which are based on network signaling anomalies, are relatively single and one-sided, and cannot scientifically assess the impact of factors such as consumer behavior and network quality on user satisfaction. Summary of the Invention

[0004] In view of the above problems, embodiments of the present invention provide a method, apparatus, and computing device for predicting the main factors affecting user satisfaction, which overcome the above problems or at least partially solve the above problems.

[0005] According to one aspect of an embodiment of the present invention, a method for predicting the main factors affecting user satisfaction is provided, the method comprising: obtaining user consumption data and user network data, and extracting features related to satisfaction; performing satisfaction prediction based on the features, and outputting the user's satisfaction; for dissatisfied users, calculating the correlation coefficient between each feature and the satisfaction and the distance from the correlation coefficient to the individual central mean; and outputting the features affecting user satisfaction based on the distance.

[0006] In an optional manner, the extracting features related to satisfaction includes: applying a chi-square distribution algorithm to extract a first preset number of features related to satisfaction.

[0007] In an optional manner, performing satisfaction prediction based on the features and outputting the user's satisfaction includes: obtaining past satisfaction survey data samples and preprocessing them; selecting an optimal model for satisfaction prediction based on the past satisfaction survey data samples; inputting the features into the optimal model for prediction, and outputting the user's satisfaction.

[0008] In an optional manner, obtaining past satisfaction survey data samples and performing preprocessing includes: obtaining past satisfaction survey data samples; for the past satisfaction survey data samples, filling empty values ​​with the mean, eliminating abnormal values, performing binary classification according to the satisfaction scores, marking those with satisfaction scores less than 7 points with 1 as positive samples, and marking the rest with 0 as negative samples; performing sample optimization on the past satisfaction survey data samples, filtering out frequently erroneous samples to obtain optimal samples with a balance of positive and negative samples.

[0009] In an optional manner, the selecting of the optimal model for satisfaction prediction based on the past satisfaction survey data samples includes: training multiple models based on the past satisfaction survey data samples to obtain corresponding recall rates and F1 values, wherein the models include logistic regression, classification tree, Gaussian process, support vector machine, and gradient boosting decision tree; and selecting the gradient boosting decision tree as the optimal model based on the recall rate and the F1 value.

[0010] In an optional manner, for dissatisfied users, the correlation coefficient between each feature and the satisfaction level and the distance from the correlation coefficient to the individual central mean are calculated, including: standardizing and null-value processing all the features of the dissatisfied users, and removing the dimensions to obtain the standardized features of the dissatisfied users; calculating the Spearman correlation coefficient between each feature and the satisfaction level to obtain the correlation coefficient between each feature and the satisfaction level; and calculating the distance between the correlation coefficient and the individual central mean, wherein the individual central mean is the average value of the correlation coefficients of the individual users.

[0011] In an optional manner, outputting the features that affect user satisfaction based on the distance includes: arranging in reverse order according to the distance, where the feature with the largest distance has the greatest impact on user satisfaction; and extracting a second preset number of features that affect user satisfaction based on the arrangement result.

[0012] According to another aspect of an embodiment of the present invention, a device for predicting the main factors affecting user satisfaction is provided, the device comprising: a data acquisition unit for acquiring user consumption data and user network data, and extracting features related to satisfaction; a satisfaction prediction unit for performing satisfaction prediction based on the features, and outputting the user's satisfaction; a distance calculation unit for calculating, for dissatisfied users, the correlation coefficient between each feature and the satisfaction and the distance from the correlation coefficient to the individual central mean; and a feature output unit for outputting features affecting user satisfaction based on the distance.

[0013] According to another aspect of an embodiment of the present invention, there is provided a computing device, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus;

[0014] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute the steps of the above-mentioned method for predicting the main factors affecting user satisfaction.

[0015] According to another aspect of an embodiment of the present invention, a computer storage medium is provided, wherein the storage medium stores at least one executable instruction, and the executable instruction enables the processor to execute the steps of the above-mentioned method for predicting the main factors affecting user satisfaction.

[0016] The embodiment of the present invention obtains user consumption data and user network data and extracts features related to satisfaction; performs satisfaction prediction based on the features and outputs the user's satisfaction; calculates, for dissatisfied users, the correlation coefficient between each feature and the satisfaction and the distance between the correlation coefficient and the individual central mean; outputs features that affect user satisfaction based on the distance, and can scientifically evaluate the impact of factors such as consumption behavior and network quality on user satisfaction.

[0017] The above description is only an overview of the technical solutions of the embodiments of the present invention. In order to more clearly understand the technical means of the embodiments of the present invention, they can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiments of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0019] Figure 1 A schematic flow chart of a method for predicting the main factors affecting user satisfaction provided by an embodiment of the present invention is shown;

[0020] Figure 2 Shown Figure 1 Schematic diagram of the process of step S12;

[0021] Figure 3 A schematic diagram showing the structure of a device for predicting the main factors affecting user satisfaction provided by an embodiment of the present invention is shown;

[0022] Figure 4 A schematic structural diagram of a computing device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0023] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0024] Figure 1 The flowchart of the method for predicting the main factors affecting user satisfaction provided by the embodiment of the present invention is shown. The method for predicting the main factors affecting user satisfaction is applied to the operator server side, such as Figure 1 As shown in Figure 2, the prediction methods for the main factors affecting user satisfaction include:

[0025] Step S11: Obtain user consumption data and user network data, and extract features related to satisfaction.

[0026] In this embodiment of the present invention, two categories of user data, domain B consumption data and domain O user network data, are obtained to establish a user feature database. A chi-square distribution algorithm is applied to the user consumption data and user network data to extract a first predetermined number of features that are highly correlated with satisfaction. For example, the user consumption data and user network data contain a total of 345 feature fields. A chi-square analysis is performed using the chi-square distribution algorithm. After extracting the top 150 features, the remaining 134 features with the largest correlation coefficients are selected and synthesized into 284 optimal features, ultimately yielding the features that are highly correlated with satisfaction.

[0027] Step S12: performing satisfaction prediction based on the features and outputting the user's satisfaction.

[0028] In the embodiment of the present invention, the optimal prediction model is first selected and the satisfaction prediction is performed using the optimal model. Figure 2 As shown, including:

[0029] Step S121: Obtain samples of previous satisfaction survey data and perform preprocessing.

[0030] Obtain samples of past satisfaction survey data; for the past satisfaction survey data samples, perform mean filling on empty values, remove abnormal values, perform binary classification based on satisfaction scores, mark 1 for satisfaction scores less than 7 as positive samples, and mark the rest as negative samples; perform sample optimization on the past satisfaction survey data samples, filter out frequently erroneous samples to obtain optimal samples with a balance between positive and negative samples. When performing sample optimization, use logistic regression to judge the samples. After 1000 logistic regression model calculations, filter out frequently erroneous samples to obtain the optimal sample. If there is an imbalance between positive and negative samples, perform sample resampling. The past satisfaction survey data samples in the embodiment of the present invention are 4500.

[0031] In step S121, past user consumption data and past user network data corresponding to past satisfaction survey data samples are simultaneously obtained, and the same method as in step S11 is applied to extract corresponding feature data.

[0032] Step S122: selecting an optimal model for satisfaction prediction based on the previous satisfaction survey data samples.

[0033] In an embodiment of the present invention, multiple models are trained based on the previous satisfaction survey data samples to obtain corresponding recall rates and F1 values, wherein the models include logistic regression, classification tree, Gaussian process, support vector machine, gradient boosting decision tree (Gradient Boosting Decision Tree, GBDT) and other models; the gradient boosting decision tree is selected as the optimal model based on the recall rate and the F1 value. Specifically, the feature data extracted based on the previous user consumption data and the previous user network data are input into the logistic regression, classification tree, Gaussian process, support vector machine, GBDT and other models for training, and the predicted satisfaction output by each model after training is compared with the actual satisfaction of the users in the previous satisfaction survey data samples to obtain the recall rate and F1 value corresponding to each model, and then the optimal model is selected based on the recall rate and F1 value corresponding to each model. The optimal model finally selected in the embodiment of the present invention is the GBDT model.

[0034] Step S123: input the features into the optimal model for prediction, and output the user's satisfaction.

[0035] After obtaining the optimal GBDT model, the features are input into the GBDT model, and the GBDT model is used to predict and output the user's satisfaction. Satisfied users with a satisfaction score less than 7 points are marked as satisfied users, and the rest are marked as dissatisfied users.

[0036] Step S13: For dissatisfied users, calculate the correlation coefficient between each feature and the satisfaction level and the distance from the correlation coefficient to the individual central mean.

[0037] After prediction by the GBDT model, the user's satisfaction can be obtained. However, it is impossible to directly determine which factors lead to user dissatisfaction, whether they are subjective factors or objective reasons, based on current means. The embodiment of the present invention gives the impact of each feature on the individual based on the correlation coefficient between the feature and satisfaction and the user's own data.

[0038] Specifically, all features of the dissatisfied users are normalized, null-valued, and dimensionless to obtain the standardized features of the dissatisfied users. For example:

[0039] df_std_ns=df_no_satisfy[[c for c in df_no_satisfy.columns if c notin['USER_ID']]]

[0040] Standard_scaler=StandardScaler()

[0041] Standard_scaler=Standard_scaler.fit(df_std_ns)

[0042] df_std_ns=pd.DataFrame(Standard_scaler.transform(df_std_ns), index=df_std_ns.indes, columns=df_std_ns.columns)

[0043] Among them, df_no_satisfy is the dissatisfied user set, and df_std_ns is the dissatisfied user set after feature standardization.

[0044] After standardizing the features of dissatisfied users, calculate the Spearman correlation coefficient between each feature and the satisfaction score to obtain the correlation coefficient x between each feature and the satisfaction score. Then calculate the distance between the correlation coefficient and the individual central mean, where the individual central mean is the average value of the correlation coefficients of individual users. For example:

[0045] deffun_corr_x_v(df):

[0046] c=df_corr3[df_corr3.index==df.name]['ABS'][0]

[0047] returndf.abs()*abs(c)

[0048] df_std_ns2=df_std_na.apply(fun_corr_x_v)

[0049] Among them, df_corr3 is the correlation coefficient set between features and satisfaction, and df_std_ns2 is the distance table from the correlation coefficient x to the individual central mean.

[0050] Step S14: outputting features that affect user satisfaction according to the distance.

[0051] In this embodiment of the present invention, by calculating the distance from the correlation coefficient of each feature to the central mean, the degree of influence of each feature on a single user can be obtained. The feature with the largest distance is the reason that has the greatest impact on user satisfaction. The importance of the impact of other features on user satisfaction can be obtained by ranking them. Therefore, in step S14, the features are sorted in reverse order according to the distance, and the feature with the largest distance has the greatest impact on user satisfaction. Based on the sorting results, a second preset number of features that affect user satisfaction are extracted. For example, as shown in Table 1, the top 10 main (TOP10) features that affect a certain user's satisfaction are obtained.

[0052] Table 1 Top 10 features that affect user satisfaction

[0053] feature corr_x_v Customer Level 0.499860 Number of intra-province calls 0.497440 Uplink packet discard rate 0.496560 The number of times users conduct business through self-service terminals 0.495520 Terminal Value 0.494030 Total number of voice call transfers 0.492060 Switching success rate 0.491880 2G call times 0.489270 Package Type 13 0.487380 10086 call duration 0.486840

[0054] The embodiment of the present invention establishes a user feature database by combining user personal consumption attribute feature data (B domain data) with network data (O domain data), predicts user satisfaction through a gradient boosting tree, and uses the distance between the feature and the satisfaction correlation coefficient and the individual central mean to obtain the main factors affecting user satisfaction. This solves the problem that the previous method can only manually judge and classify the causes of user dissatisfaction based on the user's service abnormal signaling features, resulting in single and inaccurate results.

[0055] The embodiment of the present invention obtains user consumption data and user network data and extracts features related to satisfaction; performs satisfaction prediction based on the features and outputs the user's satisfaction; calculates, for dissatisfied users, the correlation coefficient between each feature and the satisfaction and the distance between the correlation coefficient and the individual central mean; outputs features that affect user satisfaction based on the distance, and can scientifically evaluate the impact of factors such as consumption behavior and network quality on user satisfaction.

[0056] Figure 3 FIG. 1 is a schematic diagram showing the structure of a device for predicting the main factors affecting user satisfaction according to an embodiment of the present invention. Figure 3As shown, the prediction device for the main factors affecting user satisfaction includes: a data acquisition unit 301, a satisfaction prediction unit 302, a distance calculation unit 303 and a feature output unit 304. Among them:

[0057] The data acquisition unit 301 is used to acquire user consumption data and user network data, and extract features related to satisfaction; the satisfaction prediction unit 302 is used to predict satisfaction based on the features and output the user's satisfaction; the distance calculation unit 303 is used to calculate the correlation coefficient between each feature and the satisfaction and the distance from the correlation coefficient to the individual central mean for dissatisfied users; the feature output unit 304 is used to output features that affect user satisfaction based on the distance.

[0058] In an optional manner, the data acquisition unit 301 is configured to apply a chi-square distribution algorithm to extract a first preset number of features related to satisfaction.

[0059] In an optional manner, the satisfaction prediction unit 302 is used to: obtain past satisfaction survey data samples and perform preprocessing; select the optimal model for satisfaction prediction based on the past satisfaction survey data samples; input the features into the optimal model for prediction, and output the user's satisfaction.

[0060] In an optional manner, the satisfaction prediction unit 302 is used to: obtain past satisfaction survey data samples; for the past satisfaction survey data samples, fill the empty values ​​with the mean, eliminate the abnormal values, perform binary classification according to the satisfaction score, mark the satisfaction score less than 7 points as 1 as a positive sample, and mark the rest as 0 as negative samples; perform sample optimization on the past satisfaction survey data samples, filter out frequently erroneous samples to obtain the optimal sample with a balance between positive and negative samples.

[0061] In an optional manner, the satisfaction prediction unit 302 is used to: train multiple models based on the previous satisfaction survey data samples to obtain corresponding recall rates and F1 values, wherein the models include logistic regression, classification tree, Gaussian process, support vector machine, and gradient boosting decision tree; and select the gradient boosting decision tree as the optimal model based on the recall rate and the F1 value.

[0062] In an optional manner, the distance calculation unit 303 is used to: standardize and null-value all the features of the dissatisfied user, and after de-dimensioning, obtain the standardized features of the dissatisfied user; calculate the Spearman correlation coefficient between each of the features and the satisfaction level to obtain the correlation coefficient between each of the features and the satisfaction level; calculate the distance between the correlation coefficient and the individual central mean, where the individual central mean is the average value of the correlation coefficient of the individual user.

[0063] In an optional manner, the feature output unit 304 is configured to: perform reverse order arrangement based on the distance, where the feature with the largest distance has the greatest impact on user satisfaction; and extract a second preset number of features that affect user satisfaction based on the arrangement result.

[0064] The embodiment of the present invention obtains user consumption data and user network data and extracts features related to satisfaction; performs satisfaction prediction based on the features and outputs the user's satisfaction; calculates, for dissatisfied users, the correlation coefficient between each feature and the satisfaction and the distance between the correlation coefficient and the individual central mean; outputs features that affect user satisfaction based on the distance, and can scientifically evaluate the impact of factors such as consumption behavior and network quality on user satisfaction.

[0065] An embodiment of the present invention provides a non-volatile computer storage medium storing at least one executable instruction. The computer executable instruction can execute the method for predicting the main factors affecting user satisfaction in any of the above method embodiments.

[0066] The executable instructions can be used to cause the processor to perform the following operations:

[0067] Obtain user consumption data and user network data, and extract features related to satisfaction;

[0068] Perform satisfaction prediction based on the features and output the user's satisfaction;

[0069] For dissatisfied users, calculate the correlation coefficient between each feature and the satisfaction level and the distance between the correlation coefficient and the individual central mean;

[0070] Features that affect user satisfaction are output according to the distance.

[0071] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0072] A chi-square distribution algorithm is applied to extract a first preset number of features related to satisfaction.

[0073] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0074] Obtain samples of previous satisfaction survey data and perform preprocessing;

[0075] Selecting an optimal model for satisfaction prediction based on the previous satisfaction survey data samples;

[0076] The features are input into the optimal model for prediction, and the user's satisfaction is output.

[0077] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0078] Obtain samples of past satisfaction survey data;

[0079] For the previous satisfaction survey data samples, fill in the null values ​​with the mean, remove the abnormal values, and perform binary classification according to the satisfaction scores. The satisfaction scores less than 7 points are marked as 1 as positive samples, and the rest are marked as 0 as negative samples.

[0080] The sample of the previous satisfaction survey data is optimized, and the samples with frequent errors are filtered out to obtain the optimal sample with a balance of positive and negative samples.

[0081] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0082] Training multiple models based on the previous satisfaction survey data samples to obtain corresponding recall rates and F1 values, wherein the models include logistic regression, classification tree, Gaussian process, support vector machine, and gradient boosting decision tree;

[0083] The gradient boosting decision tree is selected as the optimal model according to the recall rate and the F1 value.

[0084] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0085] All features of the dissatisfied users are standardized, null-valued, and dimensionless to obtain standardized features of the dissatisfied users;

[0086] Calculating the Spearman correlation coefficient between each of the characteristics and the satisfaction level to obtain the correlation coefficient between each of the characteristics and the satisfaction level;

[0087] The distance between the correlation coefficient and the individual center mean is calculated, where the individual center mean is the average value of the correlation coefficients of the individual users.

[0088] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0089] Arrange in reverse order according to the distance, the feature with the largest distance has the greatest impact on user satisfaction;

[0090] A second preset number of features that affect user satisfaction are extracted according to the arrangement result.

[0091] The embodiment of the present invention obtains user consumption data and user network data and extracts features related to satisfaction; performs satisfaction prediction based on the features and outputs the user's satisfaction; calculates, for dissatisfied users, the correlation coefficient between each feature and the satisfaction and the distance between the correlation coefficient and the individual central mean; outputs features that affect user satisfaction based on the distance, and can scientifically evaluate the impact of factors such as consumption behavior and network quality on user satisfaction.

[0092] An embodiment of the present invention provides a computer program product, which includes a computer program stored on a computer storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer is caused to execute the method for predicting the main factors affecting user satisfaction in any of the above-mentioned method embodiments.

[0093] The executable instructions can be used to cause the processor to perform the following operations:

[0094] Obtain user consumption data and user network data, and extract features related to satisfaction;

[0095] Perform satisfaction prediction based on the features and output the user's satisfaction;

[0096] For dissatisfied users, calculate the correlation coefficient between each feature and the satisfaction level and the distance between the correlation coefficient and the individual central mean;

[0097] Features that affect user satisfaction are output according to the distance.

[0098] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0099] A chi-square distribution algorithm is applied to extract a first preset number of features related to satisfaction.

[0100] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0101] Obtain samples of previous satisfaction survey data and perform preprocessing;

[0102] Selecting an optimal model for satisfaction prediction based on the previous satisfaction survey data samples;

[0103] The features are input into the optimal model for prediction, and the user's satisfaction is output.

[0104] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0105] Obtain samples of past satisfaction survey data;

[0106] For the previous satisfaction survey data samples, fill in the null values ​​with the mean, remove the abnormal values, and perform binary classification according to the satisfaction scores. The satisfaction scores less than 7 points are marked as 1 as positive samples, and the rest are marked as 0 as negative samples.

[0107] The sample of the previous satisfaction survey data is optimized, and the samples with frequent errors are filtered out to obtain the optimal sample with a balance of positive and negative samples.

[0108] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0109] Training multiple models based on the previous satisfaction survey data samples to obtain corresponding recall rates and F1 values, wherein the models include logistic regression, classification tree, Gaussian process, support vector machine, and gradient boosting decision tree;

[0110] The gradient boosting decision tree is selected as the optimal model according to the recall rate and the F1 value.

[0111] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0112] All features of the dissatisfied users are standardized, null-valued, and dimensionless to obtain standardized features of the dissatisfied users;

[0113] Calculating the Spearman correlation coefficient between each of the characteristics and the satisfaction level to obtain the correlation coefficient between each of the characteristics and the satisfaction level;

[0114] The distance between the correlation coefficient and the individual center mean is calculated, where the individual center mean is the average value of the correlation coefficients of the individual users.

[0115] In an optional manner, the executable instruction causes the processor to perform the following operations:

[0116] Arrange in reverse order according to the distance, the feature with the largest distance has the greatest impact on user satisfaction;

[0117] A second preset number of features that affect user satisfaction are extracted according to the arrangement result.

[0118] The embodiment of the present invention obtains user consumption data and user network data and extracts features related to satisfaction; performs satisfaction prediction based on the features and outputs the user's satisfaction; calculates, for dissatisfied users, the correlation coefficient between each feature and the satisfaction and the distance between the correlation coefficient and the individual central mean; outputs features that affect user satisfaction based on the distance, and can scientifically evaluate the impact of factors such as consumption behavior and network quality on user satisfaction.

[0119] Figure 4 The schematic diagram of the structure of the computing device provided by the embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the device.

[0120] like Figure 4 As shown, the computing device may include: a processor 402 , a communications interface 404 , a memory 406 , and a communication bus 408 .

[0121] Processor 402, communication interface 404, and memory 406 communicate with each other via communication bus 408. Communication interface 404 is used to communicate with other devices, such as clients or other server network elements. Processor 402 is used to execute program 410, which may specifically perform the steps described in the embodiment of the method for predicting the main factors affecting user satisfaction.

[0122] Specifically, the program 410 may include program codes, which include computer operation instructions.

[0123] Processor 402 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement an embodiment of the present invention. The one or more processors included in the device may be processors of the same type, such as one or more CPUs, or may be processors of different types, such as one or more CPUs and one or more ASICs.

[0124] The memory 406 is used to store the program 410. The memory 406 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0125] The program 410 may be specifically configured to cause the processor 402 to perform the following operations:

[0126] Obtain user consumption data and user network data, and extract features related to satisfaction;

[0127] Perform satisfaction prediction based on the features and output the user's satisfaction;

[0128] For dissatisfied users, calculate the correlation coefficient between each feature and the satisfaction level and the distance between the correlation coefficient and the individual central mean;

[0129] Features that affect user satisfaction are output according to the distance.

[0130] In an optional manner, the program 410 enables the processor to perform the following operations:

[0131] A chi-square distribution algorithm is applied to extract a first preset number of features related to satisfaction.

[0132] In an optional manner, the program 410 enables the processor to perform the following operations:

[0133] Obtain samples of previous satisfaction survey data and perform preprocessing;

[0134] Selecting an optimal model for satisfaction prediction based on the previous satisfaction survey data samples;

[0135] The features are input into the optimal model for prediction, and the user's satisfaction is output.

[0136] In an optional manner, the program 410 enables the processor to perform the following operations:

[0137] Obtain samples of past satisfaction survey data;

[0138] For the previous satisfaction survey data samples, fill in the null values ​​with the mean, remove the abnormal values, and perform binary classification according to the satisfaction scores. The satisfaction scores less than 7 points are marked as 1 as positive samples, and the rest are marked as 0 as negative samples.

[0139] The sample of the previous satisfaction survey data is optimized, and the samples with frequent errors are filtered out to obtain the optimal sample with a balance of positive and negative samples.

[0140] In an optional manner, the program 410 enables the processor to perform the following operations:

[0141] Training multiple models based on the previous satisfaction survey data samples to obtain corresponding recall rates and F1 values, wherein the models include logistic regression, classification tree, Gaussian process, support vector machine, and gradient boosting decision tree;

[0142] The gradient boosting decision tree is selected as the optimal model according to the recall rate and the F1 value.

[0143] In an optional manner, the program 410 enables the processor to perform the following operations:

[0144] All features of the dissatisfied users are standardized, null-valued, and dimensionless to obtain standardized features of the dissatisfied users;

[0145] Calculating the Spearman correlation coefficient between each of the characteristics and the satisfaction level to obtain the correlation coefficient between each of the characteristics and the satisfaction level;

[0146] The distance between the correlation coefficient and the individual center mean is calculated, where the individual center mean is the average value of the correlation coefficients of the individual users.

[0147] In an optional manner, the program 410 enables the processor to perform the following operations:

[0148] Arrange in reverse order according to the distance, the feature with the largest distance has the greatest impact on user satisfaction;

[0149] A second preset number of features that affect user satisfaction are extracted according to the arrangement result.

[0150] The embodiment of the present invention obtains user consumption data and user network data and extracts features related to satisfaction; performs satisfaction prediction based on the features and outputs the user's satisfaction; calculates, for dissatisfied users, the correlation coefficient between each feature and the satisfaction and the distance between the correlation coefficient and the individual central mean; outputs features that affect user satisfaction based on the distance, and can scientifically evaluate the impact of factors such as consumption behavior and network quality on user satisfaction.

[0151] The algorithm or demonstration provided herein are not inherently relevant to any particular computer, virtual system or other equipment. Various general-purpose systems may also be used together with the teachings based on this. According to the above description, it is apparent that the structure required for constructing this type of system. In addition, the embodiment of the present invention is not directed to any specific programming language yet. It should be understood that various programming languages ​​can be utilized to realize the content of the present invention described herein, and the above description of specific languages ​​is for the purpose of disclosing the best mode of the present invention.

[0152] In the description provided herein, numerous specific details are described. However, it is understood that embodiments of the present invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques are not shown in detail so as not to obscure the understanding of this description.

[0153] Similarly, it should be understood that in order to streamline the present invention and facilitate understanding of one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, various features of the embodiments of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, this method of disclosure should not be interpreted as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim.

[0154] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and in addition may be divided into multiple submodules or subunits or subcomponents. All features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed herein may be combined in any combination, except that at least some of such features and / or processes or units are mutually exclusive. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.

[0155] It should be noted that the above embodiments illustrate rather than limit the invention, and that alternative embodiments may be devised by a person skilled in the art without departing from the scope of the appended claims. In the claims, any reference signs placed between brackets should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising several different elements and by means of appropriately programmed computers. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments should not be understood as limiting the order of execution unless otherwise specified.

Claims

1. A method for predicting the main factors affecting user satisfaction, characterized by: The method comprises: Obtain user consumption data and user network data, and extract features related to satisfaction; Perform satisfaction prediction based on the features and output the user's satisfaction; For dissatisfied users, calculating the correlation coefficient between each feature and the satisfaction level and the distance between the correlation coefficient and the individual central mean; wherein the individual central mean is the average value of the correlation coefficients of the individual users; outputting features that affect user satisfaction according to the distance; Among them, for dissatisfied users, calculating the correlation coefficient between each feature and the satisfaction level and the distance from the correlation coefficient to the individual central mean includes: standardizing and null-value processing all the features of the dissatisfied users, and removing the dimensions to obtain the standardized features of the dissatisfied users; calculating the Spearman correlation coefficient between each feature and the satisfaction level to obtain the correlation coefficient between each feature and the satisfaction level; and calculating the distance between the correlation coefficient and the individual central mean.

2. The method according to claim 1, characterized in that The extraction of features related to satisfaction includes: A chi-square distribution algorithm is applied to extract a first preset number of features related to satisfaction.

3. The method according to claim 1, characterized in that The performing satisfaction prediction based on the features and outputting the user's satisfaction includes: Obtain samples of previous satisfaction survey data and perform preprocessing; Selecting an optimal model for satisfaction prediction based on the previous satisfaction survey data samples; The features are input into the optimal model for prediction, and the user's satisfaction is output.

4. The method according to claim 3, characterized in that The obtaining of past satisfaction survey data samples and preprocessing thereof include: Obtain samples of past satisfaction survey data; For the previous satisfaction survey data samples, fill in the null values ​​with the mean, remove the abnormal values, and perform binary classification according to the satisfaction scores. The satisfaction scores less than 7 points are marked as 1 as positive samples, and the rest are marked as 0 as negative samples. The sample of the previous satisfaction survey data is optimized, and the samples with frequent errors are filtered out to obtain the optimal sample with a balance of positive and negative samples.

5. The method according to claim 3, characterized in that The optimal model for satisfaction prediction is selected based on the previous satisfaction survey data samples, including: Training multiple models based on the previous satisfaction survey data samples to obtain corresponding recall rates and F1 values, wherein the models include logistic regression, classification tree, Gaussian process, support vector machine, and gradient boosting decision tree; The gradient boosting decision tree is selected as the optimal model according to the recall rate and the F1 value.

6. The method according to claim 1, characterized in that Outputting the features that affect user satisfaction according to the distance includes: Arrange in reverse order according to the distance, the feature with the largest distance has the greatest impact on user satisfaction; A second preset number of features that affect user satisfaction are extracted according to the arrangement result.

7. A device for predicting the main factors affecting user satisfaction, characterized in that: The device comprises: A data acquisition unit, used to acquire user consumption data and user network data, and extract features related to satisfaction; a satisfaction prediction unit, configured to predict satisfaction based on the features and output the user's satisfaction; a distance calculation unit, configured to calculate, for dissatisfied users, a correlation coefficient between each feature and the satisfaction level and a distance from the correlation coefficient to an individual central mean; wherein the individual central mean is an average value of the correlation coefficients of individual users; a feature output unit, configured to output features affecting user satisfaction according to the distance; Among them, for dissatisfied users, calculating the correlation coefficient between each feature and the satisfaction level and the distance from the correlation coefficient to the individual central mean includes: standardizing and null-value processing all the features of the dissatisfied users, and removing the dimensions to obtain the standardized features of the dissatisfied users; calculating the Spearman correlation coefficient between each feature and the satisfaction level to obtain the correlation coefficient between each feature and the satisfaction level; and calculating the distance between the correlation coefficient and the individual central mean.

8. A computing device, characterized in that include: A processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute the steps of the method for predicting the main factors affecting user satisfaction according to any one of claims 1-6.

9. A computer storage medium, characterized in that The storage medium stores at least one executable instruction, and the executable instruction enables the processor to execute the steps of the method for predicting the main factors affecting user satisfaction according to any one of claims 1 to 6.

Images