Traffic usage amount prediction method and device, electronic equipment and storage medium

Abstract

The invention relates to a traffic usage amount prediction method and apparatus, an electronic device and a storage medium. The method comprises the steps of obtaining historical traffic data of a user and multi-dimensional feature data corresponding to the historical traffic data; constructing an original traffic sequence based on a time sequence according to the historical traffic data, and constructing a feature matrix according to the multi-dimensional feature data; constructing an autoregressive integral moving average model containing an exogenous variable model according to the original traffic sequence and the feature matrix; carrying out model order determination on the autoregressive integral moving average model containing the exogenous variable model to obtain a target model; and based on the target model and the feature data of the target moment, predicting the traffic usage amount of the target moment. According to the method, historical traffic data and multi-dimensional feature data are integrated to construct an extended autoregression integral moving average model containing an exogenous variable model, so that the traffic usage amount at a target moment can be accurately predicted in combination with the feature data.

Description

Technical Field

[0001] This application relates to the technical field of traffic prediction, and specifically relates to a traffic usage prediction method, device, electronic device, and storage medium. Background Art

[0002] With the popularization of the mobile Internet, the traffic usage behaviors of vehicle Internet users show highly personalized and complex characteristics. Traditional traffic prediction methods (such as simple moving average and exponential smoothing method) only rely on historical traffic data and cannot accurately predict traffic usage, with insufficient prediction accuracy. Summary of the Invention

[0003] To solve the above technical problems or at least partially solve the above technical problems, this application provides a traffic usage prediction method, device, electronic device, and storage medium.

[0004] In a first aspect, this application provides a traffic usage prediction method, and the method includes:

[0005] Obtain the historical traffic data of a user and the multi-dimensional feature data corresponding to the historical traffic data;

[0006] Construct an original traffic sequence based on time series according to the historical traffic data, and construct a feature matrix according to the multi-dimensional feature data;

[0007] Construct an autoregressive integrated moving average model with exogenous variables according to the original traffic sequence and the feature matrix; wherein, the feature matrix serves as the exogenous variable of the autoregressive integrated moving average model with exogenous variables;

[0008] Perform model order determination on the autoregressive integrated moving average model with exogenous variables to obtain a target model;

[0009] Predict the traffic usage at the target time based on the target model and the feature data at the target time.

[0010] Optionally, constructing an original traffic sequence based on time series according to the historical traffic data includes:

[0011] Construct a time traffic sequence based on time series according to the historical traffic data;

[0012] Perform outlier detection on the time traffic sequence to obtain the outliers in the time traffic sequence;

[0013] Remove the outliers in the time traffic sequence and perform stationary processing on the time traffic sequence to obtain the original traffic sequence.

[0014] Optionally, the multi-dimensional feature data at least includes the package price, the remaining amount of the complimentary data package, the network mode, the geographical attribute, and the weight of the traffic consumption of the application program;

[0015] Construct a feature matrix according to the multi-dimensional feature data, including:

[0016] Construct a feature matrix according to the package price, the remaining amount of the complimentary data package, the network mode, the geographical attribute, and the weight of the traffic consumption of the application program;

[0017] Perform differencing processing on the features in the feature matrix so that the differencing order of the feature matrix is the same as that of the original traffic sequence.

[0018] Optionally, construct an autoregressive integrated moving average model with exogenous variables according to the original traffic sequence and the feature matrix, including:

[0019]

[0020] where, Δdy t represents the original traffic sequence and serves as the dependent variable of the autoregressive integrated moving average model with exogenous variables, p is the autoregressive order, q is the moving average order, m represents the number of exogenous variables, X t-k represents a matrix containing m exogenous variables, represents the autoregressive coefficient, θ j represents the moving average coefficient, β k represents the exogenous variable coefficient, ε t represents white noise.

[0021] Optionally, perform model order determination on the autoregressive integrated moving average model with exogenous variables to obtain a target model, including:

[0022] Perform model order determination through graphical identification according to the autocorrelation function and the partial autocorrelation function;

[0023] If the partial autocorrelation function ends after the p-th order and the autocorrelation function has a tail, it is determined as an autoregressive model;

[0024] If the autocorrelation function ends after the q-th order and the partial autocorrelation function has a tail, it is determined as a moving average model;

[0025] Estimate the autoregressive coefficient, the moving average coefficient, and the exogenous variable coefficient in the autoregressive integrated moving average model with exogenous variables through maximum likelihood estimation to obtain the target model.

[0026] Optionally, based on the target model and the feature data at the target time, predict the traffic usage at the target time, including:

[0027] Obtain the feature data at the target moment;

[0028] Input the feature data at the target moment into the target model to obtain a recursively calculated predicted traffic sequence;

[0029] Inverse difference restore the predicted traffic sequence to obtain the predicted traffic usage at the target moment.

[0030] Optionally, after predicting the traffic usage at the target moment, the method further includes:

[0031] Perform visualization processing on the prediction result to obtain a visualization result;

[0032] Output the visualization result.

[0033] In a second aspect, the present application provides a traffic usage prediction device, and the device includes:

[0034] An acquisition module, configured to acquire the historical traffic data of the user and the multi-dimensional feature data corresponding to the historical traffic data;

[0035] A first construction module, configured to construct an original traffic sequence based on a time series according to the historical traffic data, and construct a feature matrix according to the multi-dimensional feature data;

[0036] A second construction module, configured to construct an autoregressive integrated moving average model with exogenous variables according to the original traffic sequence and the feature matrix; wherein, the feature matrix serves as an exogenous variable of the autoregressive integrated moving average model with exogenous variables;

[0037] A model optimization module, configured to perform model order determination on the autoregressive integrated moving average model with exogenous variables to obtain a target model;

[0038] A prediction module, configured to predict the traffic usage at the target moment based on the target model and the feature data at the target moment.

[0039] In a third aspect, the present application provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete mutual communication through the communication bus;

[0040] The memory is used to store a computer program;

[0041] The processor, when executing the program stored in the memory, implements the steps of the traffic usage prediction method according to any one of the embodiments in the first aspect.

[0042] Fourthly, the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the traffic usage prediction method described in any embodiment of the first aspect are implemented.

[0043] Advantages of the present application:

[0044] For the method provided in the embodiment of the present application, historical traffic data of a user and multi-dimensional feature data corresponding to the historical traffic data are obtained; an original traffic sequence based on a time series is constructed according to the historical traffic data, and a feature matrix is constructed according to the multi-dimensional feature data; an autoregressive integrated moving average model with exogenous variables is constructed according to the original traffic sequence and the feature matrix; wherein the feature matrix serves as an exogenous variable of the autoregressive integrated moving average model with exogenous variables; the autoregressive integrated moving average model with exogenous variables is model-ordered to obtain a target model; and based on the target model and the feature data at a target moment, the traffic usage at the target moment is predicted. By integrating historical traffic data and multi-dimensional feature data, the method constructs an extended autoregressive integrated moving average model with exogenous variables, so that the traffic usage at the target moment can be accurately predicted in combination with the feature data. Description of the drawings

[0045] The drawings here are incorporated into the specification and form a part of the specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.

[0047] Figure 1 It is a system architecture diagram of a traffic usage prediction method provided by an embodiment of the present application;

[0048] Figure 2 It is a flowchart of a traffic usage prediction method provided by an embodiment of the present application;

[0049] Figure 3 It is a flowchart of a traffic usage prediction method provided by another embodiment of the present application;

[0050] Figure 4 It is a structural diagram of a traffic usage prediction device provided by an embodiment of the present application;

[0051] Figure 5Schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0052] The following will illustrate the implementation manners of the present application with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in this specification. The present application can also be implemented or applied through other different specific implementation manners. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for illustrating the present application and not for limiting the protection scope of the present application.

[0053] With the popularization of the mobile Internet, the traffic usage behaviors of vehicle Internet users show highly personalized and complex characteristics. Traditional traffic prediction methods (such as simple moving average and exponential smoothing method) only rely on historical traffic data and cannot accurately predict the traffic usage. In some related technologies, the AutoRegressive Integrated Moving Average Model (ARIMA model for short) is used to predict the traffic usage of users. As a classic time series prediction tool, the ARIMA model has advantages in dealing with non-stationary data. However, the traditional ARIMA model usually only inputs a single traffic time series, ignoring the influence of user attributes (such as vehicle type, package price) and behavior characteristics (such as software usage habits, in-vehicle software installation situation, software traffic consumption situation distribution) on traffic consumption, and also cannot quantify the influence of the above characteristics on traffic changes, resulting in insufficient prediction accuracy.

[0054] The first embodiment of the present application provides a traffic usage prediction method, which can be applied to a Figure 1 system architecture as shown. The system architecture at least includes a data acquisition module 101 and a data processing module 102, and a communication connection is established between the data acquisition module 101 and the data processing module 102. Specifically, the system architecture can be a vehicle or a server that can implement vehicle Internet. Among them, the type of vehicle is not limited. For example, it can be a fuel vehicle, a pure electric vehicle, a hybrid vehicle or a fuel cell vehicle, etc. The server can be a cloud server or a server cluster.

[0055] Next, based on this system architecture, the traffic usage prediction method will be described in detail. As Figure 2 , the traffic usage prediction method includes:

[0056] Step 201, obtain the historical traffic data of the user and the multi-dimensional feature data corresponding to the historical traffic data.

[0057] The historical traffic data of the user, representing the traffic consumption data on a monthly or quarterly basis. The multi-dimensional feature data may include features in different dimensions such as user basic attributes, behavior pattern features, environmental features, etc. Specifically, the user basic attributes may include the package price Pt, the remaining amount of the gifted traffic package Ft, the terminal type (such as 4G / 5G mobile phone, the model of the intelligent electric vehicle car machine system), etc. The behavior pattern features may include the traffic consumption weight St of the application APP (such as the proportion of video APPs), the peak traffic consumption period (such as the high-frequency usage period from 20:00 to 22:00 in the evening), etc. The environmental features may include the network mode Ct (5G base station coverage rate), the geographical attribute It (urban / rural user label), etc., without limitation.

[0058] Step 202, construct an original traffic sequence based on time series according to the historical traffic data, and construct a feature matrix according to the multi-dimensional feature data.

[0059] In one embodiment, constructing an original traffic sequence based on time series according to the historical traffic data includes: constructing a time traffic sequence based on time series according to the historical traffic data; performing outlier detection on the time traffic sequence to obtain the outliers in the time traffic sequence; removing the outliers in the time traffic sequence and performing a stationary processing on the time traffic sequence to obtain the original traffic sequence.

[0060] In this embodiment, the historical traffic data can be preprocessed first. For example, data cleaning can be performed. The 3σ principle can be used for outlier detection and repair based on the probability characteristics of the normal distribution to obtain the time traffic sequence (which can be represented by yt). After the data cleaning is completed, the time traffic sequence can be subjected to stationary processing. The stationary processing can, for example, perform a unit root test on the time traffic sequence. If the sequence is non-stationary, calculate the difference order d to obtain the stationary original traffic sequence Δdy t 。

[0061] In one embodiment, constructing a feature matrix according to the multi-dimensional feature data includes: constructing a feature matrix according to the package price, the remaining amount of the gifted traffic package, the network mode, the geographical attribute, and the traffic consumption weight of the application; performing a difference processing on the features in the feature matrix so that the difference order of the feature matrix is the same as that of the original traffic sequence.

[0062] In this embodiment, data cleaning, normalization, and differencing processing can be first performed on the multi-dimensional feature data. For example, one-hot encoding is performed on categorical variables (such as vehicle models) to generate binary feature vectors, and continuous variables are normalized to the interval [0, 1] using Min-Max normalization. Then, a feature matrix can be constructed based on the multi-dimensional feature data. For example, the feature matrix Xt = [Ft, Ct, It, St, Pt], and the feature matrix Xt is aligned with the time flow sequence yt. The differencing processing can be feature fusion differencing, that is, the feature matrix Xt is synchronously differenced (such as taking the first-order difference Δf t = f t - f t-1 , and the differencing order is kept consistent with that of the time flow sequence.

[0063] Step 203, construct an autoregressive integrated moving average model with exogenous variables based on the original flow sequence and the feature matrix; wherein, the feature matrix serves as the exogenous variable of the autoregressive integrated moving average model with exogenous variables.

[0064] The autoregressive integrated moving average model with exogenous variables is the ARIMAX model (AutoRegressive Integrated Moving Average with eXogenous variables Model). It introduces exogenous variables on the basis of the autoregressive integrated moving average model (AutoRegressive Integrated Moving Average Model, that is, the ARIMA model). In this embodiment, the feature matrix is used as the exogenous variable of the ARIMAX model.

[0065] In one embodiment, constructing an autoregressive integrated moving average model with exogenous variables based on the original flow sequence and the feature matrix includes:

[0066]

[0067] wherein, Δdy t represents the original flow sequence and serves as the dependent variable of the ARIMAX model, p is the autoregressive order, q is the moving average order, m represents the number of exogenous variables, X t-k represents a matrix containing m exogenous variables, represents the autoregressive coefficient, θ j represents the moving average coefficient, β k represents the exogenous variable coefficient, ε t represents white noise. Among them, β k can directly reflect the impact of each feature on the flow change. For example, β 终端类型>0 indicates that the traffic consumption of the intelligent terminal user is higher.

[0068] In this embodiment, since the feature matrix is an exogenous variable of the ARIMAX model, the constructed ARIMAX model can reflect the multi-dimensional feature data into the ARIMAX model. After that, the influence of each feature on the traffic change can be obtained through the ARIMAX model.

[0069] Step 204: Determine the order of the autoregressive integrated moving average model with exogenous variables to obtain the target model.

[0070] In one embodiment, determining the order of the autoregressive integrated moving average model with exogenous variables to obtain the target model includes: determining the order of the model through graph recognition according to the autocorrelation function and the partial autocorrelation function; if the partial autocorrelation function ends after the p-th order and the autocorrelation function has a tail, it is determined as an autoregressive model; if the autocorrelation function ends after the q-th order and the partial autocorrelation function has a tail, it is determined as a moving average model; estimating the autoregressive coefficient, the moving average coefficient, and the exogenous variable coefficient in the autoregressive integrated moving average model with exogenous variables through maximum likelihood estimation to obtain the target model.

[0071] In this embodiment, the order of the model can be determined through graph recognition according to the autocorrelation function (Autocorrelation Function, abbreviated as ACF) and the partial autocorrelation function (Partial Autocorrelation Function, abbreviated as PACF). For example, if the PACF truncates after the p-th order and the ACF has a tail, it is determined as an autoregressive model (AR(p) model); if the ACF truncates after the q-th order and the PACF has a tail, it is determined as a moving average model (MA(q) model). The order-determined model can be optimized. For example, the Akaike Information Criterion (AIC criterion) and the Bayesian Information Criterion (Bayesian Information Criterion, abbreviated as BIC criterion) are used to screen the optimal model. AIC = 2k - 2ln(L), BIC = kln(n) - 2ln(L), where k is the number of model parameters, and L is the likelihood function value, to ensure the balance between the fitting accuracy and the complexity of the model. Next, the autoregressive coefficient, the moving average coefficient, and the exogenous variable coefficient in the ARIMAX model can be estimated through maximum likelihood estimation to obtain the target model. For example, optimization algorithms such as BFGS can be used to estimate the parameters such as the autoregressive coefficient φi, the moving average coefficient θj, and the exogenous variable coefficient βk in the model, that is, to estimate {φi, θj, βk}, and minimize the sum of squared prediction errors. Of course, the residual test can also be performed on the target model to verify whether the residual sequence is white noise. If there is autocorrelation, the model order can be adjusted to retrain the model.

[0072] Step 205: Predict the traffic usage at the target moment based on the target model and the feature data at the target moment.

[0073] This method constructs an extended autoregressive integrated moving average model with exogenous variables (ARIMAX model) by integrating historical traffic data and multi-dimensional feature data, so as to accurately predict the traffic usage at the target moment in combination with the feature data.

[0074] In one embodiment, predicting the traffic usage at the target moment based on the target model and the feature data at the target moment includes: obtaining the feature data at the target moment; inputting the feature data at the target moment into the target model to obtain a recursively calculated predicted traffic sequence; and performing inverse differencing reduction on the predicted traffic sequence to obtain the predicted traffic usage at the target moment.

[0075] In this embodiment, the target moment refers to a future time node for which traffic usage needs to be predicted. When predicting the target moment, it is necessary to input the feature data at the target moment, such as the preset free traffic package and package price, to predict the traffic usage at the target moment under this free traffic package and package price. Of course, the feature data can include one feature or multiple features, without limitation. After inputting the feature data at the target moment into the target model, a recursively calculated predicted traffic sequence is obtained, and inverse differencing reduction is performed on the predicted traffic sequence to obtain the predicted traffic usage at the target moment.

[0076] In one embodiment, after predicting the traffic usage at the target moment, the method further includes: performing visualization processing on the prediction result to obtain a visualization result; and outputting the visualization result.

[0077] In this embodiment, the predicted visualization result can be output, such as obtaining visualization graphs, tables, etc. as prediction results, and outputting the visualization result to the traffic operation management system, so that the traffic usage at a future target moment can be obtained intuitively and conveniently.

[0078] In a specific embodiment, such as Figure 3 , the traffic usage prediction method includes:

[0079] Step S1: Obtain the user's historical traffic data and multi-dimensional feature data.

[0080] Traffic time series: The user's historical traffic usage (such as monthly / quarterly consumption data), forming a time series input.

[0081] User basic attributes: Package price Pt, remaining free traffic package Ft, terminal type (such as 4G / 5G mobile phone, intelligent electric vehicle car machine system model).

[0082] Behavior pattern features: The traffic consumption weight St of frequently used APPs (such as the proportion of video APPs), and the peak time period of traffic consumption (such as the high-frequency usage period from 20:00 to 22:00 in the evening).

[0083] Environmental features: Network mode (5G base station coverage rate), geographical attributes (urban / rural user labels).

[0084] Step S2, perform data cleaning, normalization, encoding, and differencing processing.

[0085] Data cleaning: Detect and repair outliers in the time traffic sequence yt (such as based on the 3σ principle); perform one-hot encoding (One-Hot Encoding) on categorical variables (such as vehicle models) to generate binary feature vectors, and perform Min-Max normalization on continuous variables to the [0,1] interval; construct the feature matrix Xt = [Ft, Ct, It, St, Pt], which is aligned with the time traffic sequence yt.

[0086] Stationarity processing: Perform a unit root test on the time traffic sequence yt. If the sequence is non-stationary, calculate the differencing order d to obtain the stationary original traffic sequence Δdyt.

[0087] Feature fusion differencing: Perform differencing processing on the feature matrix Xt synchronously (such as taking the first-order difference ΔF t = F t - F t-1 ) for the remaining amount of the given traffic package, and keep it consistent with the differencing order of the traffic sequence.

[0088] Step S3, implement ARIMAX model order determination, parameter estimation, and residual test.

[0089] Model structure: Use the preprocessed stationary original traffic sequence Δdyt as the dependent variable and the fused feature matrix Xt as the exogenous variable to construct an ARIMAX model.

[0090] Model order determination: Identify the autoregressive and moving average orders through the ACF / PACF diagram. If the PACF truncates after the p-th order and the ACF tails off, it is determined as an AR(p) model; if the ACF truncates after the q-th order and the PACF tails off, it is determined as an MA(q) model.

[0091] Model optimization: Use the AIC / BIC criterion to screen the optimal model, AIC = 2k - 2ln(L), BIC = kln(n) - 2ln(L), where k is the number of model parameters, L is the likelihood function value, to ensure a balance between the fitting accuracy and complexity of the model.

[0092] Step S4, generate the prediction results of the traffic usage volume at future time nodes.

[0093] Maximum Likelihood Estimation: Use optimization algorithms such as BFGS to estimate the model parameters {φi, θj, βk}, and minimize the sum of squared prediction errors.

[0094] Residual Test: Verify whether the residual sequence is white noise (which can be done through the Ljung-Box test). If there is autocorrelation, adjust the model order and retrain.

[0095] Multi-step Prediction: Based on the trained ARIMAX model, input the feature data of future time nodes (such as preset free data packages and package prices), and obtain the predicted traffic sequence through recursive calculation.

[0096] Inverse Differencing Restoration: Perform inverse differencing on the predicted differenced sequence to obtain the predicted value of traffic usage at the original scale.

[0097] In step S5, visualize the prediction results and output them to the traffic operation management system.

[0098] In the embodiments of the present application, through data-driven feature fusion, multi-dimensional features such as user attributes, behaviors, and environments are combined with historical traffic sequences to form an input matrix containing time dependence and external influences. It realizes interpretable prediction of statistical models. Utilizing the linear superposition property of the ARIMAX model, it captures the historical dependence of the traffic sequence through autoregressive terms, and quantifies the feature influence through exogenous variable terms, ensuring that the prediction results are traceable and verifiable. It realizes modular system support. The data collection and preprocessing module adapts to multi-source data, the model construction module integrates classic statistical tools, and the application layer outputs analysis results with both prediction accuracy and business value.

[0099] Adopting the traffic usage prediction method of the embodiments of the present application can accurately predict future traffic usage trends when the data volume is not high. The comprehensive analysis of this method has the following characteristics:

[0100] Cross-scenario Generalizability: By replacing the feature matrix (such as replacing "terminal type" with "industrial equipment model"), it can be quickly applied to industrial Internet of Things traffic prediction without redesigning the model architecture.

[0101] Low-cost and High-efficiency Modeling: When the data volume is reduced by 50% (1000 user data), the prediction accuracy only drops by 3% (compared with a 15% drop in deep learning models), which is suitable for rapid deployment by small and medium-sized enterprises.

[0102] Policy Formulation Support: Through feature contribution analysis, operators can specifically launch "5G user exclusive data packages" or "video application targeted discounts", which can increase the conversion rate of user traffic package purchases by 20%.

[0103] Based on the same technical concept, the second embodiment of the present application provides a traffic usage prediction device, as Figure 4, the device includes:

[0104] An acquisition module 401, configured to acquire the historical traffic data of a user and the multi-dimensional feature data corresponding to the historical traffic data;

[0105] A first construction module 402, configured to construct an original traffic sequence based on a time series according to the historical traffic data, and construct a feature matrix according to the multi-dimensional feature data;

[0106] A second construction module 403, configured to construct an autoregressive integrated moving average model with exogenous variables according to the original traffic sequence and the feature matrix; wherein, the feature matrix serves as an exogenous variable of the autoregressive integrated moving average model with exogenous variables;

[0107] A model optimization module 404, configured to perform model order determination on the autoregressive integrated moving average model with exogenous variables to obtain a target model;

[0108] A prediction module 405, configured to predict the traffic usage at a target time based on the target model and the feature data at the target time.

[0109] The device constructs an extended autoregressive integrated moving average model with exogenous variables by integrating historical traffic data and multi-dimensional feature data, so as to accurately predict the traffic usage at a target time in combination with the feature data.

[0110] As Figure 5 shown, the third embodiment of the present application provides an electronic device, including a processor 111, a communication interface 112, a memory 113, and a communication bus 114. Among them, the processor 111, the communication interface 112, and the memory 113 complete communication with each other through the communication bus 114,

[0111] The memory 113 is used to store a computer program;

[0112] In one embodiment, when the processor 111 is configured to execute the program stored on the memory 113, it implements the traffic usage prediction method provided in any one of the foregoing method embodiments.

[0113] The memory and the processor in the above-mentioned electronic device communicate through the communication bus and the communication interface. The communication bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into an address bus, a data bus, a control bus, etc.

[0114] The memory may include a Random Access Memory (RAM), or may also include a non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0115] The aforementioned processor may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0116] The fourth embodiment of the present application provides a computer-readable medium having non-volatile program code executable by a processor.

[0117] Optionally, in the embodiments of the present application, the computer-readable medium is configured to store program code for the processor to execute the above method.

[0118] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments, and will not be elaborated herein.

[0119] When the embodiments of the present application are specifically implemented, reference may be made to the above respective embodiments, and corresponding technical effects are achieved.

[0120] It can be understood that the embodiments described herein can be implemented using hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions of this application, or a combination thereof.

[0121] For software implementation, the technologies herein can be implemented by units that execute the functions herein. The software code can be stored in a memory and executed by a processor. The memory can be implemented within the processor or externally to the processor.

[0122] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented using electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of this application.

[0123] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0124] In the embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of modules is only a logical function division, and there can be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of devices or units can be in electrical, mechanical, or other forms.

[0125] The unit described as a separation component may or may not be physically separated. The component displayed as a unit may or may not be a physical unit, that is, it may be located in one place or distributed across multiple network units. Some or all of these units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0126] In addition, in each embodiment of the present application, each functional unit may be integrated in a processing unit, may exist separately as individual physical units, or two or more units may be integrated in one unit.

[0127] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of the present application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, ROM, RAM, magnetic disks, or optical discs that can store program codes.

[0128] It should be noted that in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, the element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0129] The above embodiments are only preferred embodiments given to fully illustrate the present application, and the protection scope of the present application is not limited thereto. Equivalent substitutions or transformations made by those skilled in the art on the basis of the present application are within the protection scope of the present application.

Claims

1. A method for predicting traffic usage, characterized in that, The method includes: Obtaining the historical traffic data of the user and the multi-dimensional feature data corresponding to the historical traffic data; Constructing an original traffic sequence based on time series according to the historical traffic data, and constructing a feature matrix according to the multi-dimensional feature data; Constructing an autoregressive integrated moving average model with exogenous variables according to the original traffic sequence and the feature matrix; wherein, the feature matrix serves as an exogenous variable of the autoregressive integrated moving average model with exogenous variables; Performing model order determination on the autoregressive integrated moving average model with exogenous variables to obtain a target model; Predicting the traffic usage at the target moment based on the target model and the feature data at the target moment.

2. The method according to claim 1, wherein Constructing an original traffic sequence based on time series according to the historical traffic data, including: Constructing a time traffic sequence based on time series according to the historical traffic data; Performing outlier detection on the time traffic sequence to obtain the outliers in the time traffic sequence; Removing the outliers in the time traffic sequence and performing stationary processing on the time traffic sequence to obtain the original traffic sequence.

3. The method according to claim 1, wherein The multi-dimensional feature data at least includes the package price, the remaining amount of the given traffic package, the network mode, the geographical attribute, and the traffic consumption weight of the application program; Constructing a feature matrix according to the multi-dimensional feature data, including: Constructing a feature matrix according to the package price, the remaining amount of the given traffic package, the network mode, the geographical attribute, and the traffic consumption weight of the application program; Performing difference processing on the features in the feature matrix so that the difference order of the feature matrix is the same as that of the original traffic sequence.

4. The method according to claim 1, characterized in that, Constructing an autoregressive integrated moving average model with exogenous variables according to the original traffic sequence and the feature matrix, including: where, Δdy t represents the original flow sequence and serves as the dependent variable of the autoregressive integrated moving average model with exogenous variables. p is the autoregressive order, q is the moving average order, m represents the number of exogenous variables, and X t-k represents a matrix containing m exogenous variables, represents the autoregressive coefficient, θ j represents the moving average coefficient, β k represents the exogenous variable coefficient, ε t represents white noise.

5. The method according to claim 4, characterized in that, Performing model order determination on the autoregressive integrated moving average model with exogenous variables to obtain a target model, including: Performing model order determination through graph recognition according to the autocorrelation function and the partial autocorrelation function; If the partial autocorrelation function ends after the p-th order and the autocorrelation function has a tail, it is determined as an autoregressive model; If the autocorrelation function ends after the q-th order and the partial autocorrelation function has a tail, it is determined as a moving average model; Estimating the autoregressive coefficient, the moving average coefficient, and the exogenous variable coefficient in the autoregressive integrated moving average model with exogenous variables through maximum likelihood estimation to obtain the target model.

6. The method according to claim 1, characterized in that, Predicting the traffic usage at the target moment based on the target model and the feature data at the target moment, including: Obtaining the feature data at the target moment; Inputting the feature data at the target moment into the target model to obtain a recursively calculated predicted traffic sequence; Restoring the predicted traffic sequence through inverse differencing to obtain the predicted traffic usage at the target moment.

7. The method according to claim 1, characterized in that After predicting the traffic usage at the target moment, the method further includes: Performing visualization processing on the prediction result to obtain a visualization result; Outputting the visualization result.

8. A traffic usage prediction device, characterized in that The device includes: An obtaining module, configured to obtain the historical traffic data of the user and the multi-dimensional feature data corresponding to the historical traffic data; A first construction module for constructing an original traffic sequence based on time series according to the historical traffic data, and constructing a feature matrix according to the multi-dimensional feature data; A second construction module for constructing an autoregressive integrated moving average model with exogenous variables according to the original traffic sequence and the feature matrix; wherein the feature matrix serves as an exogenous variable of the autoregressive integrated moving average model with exogenous variables; A model optimization module for performing model order determination on the autoregressive integrated moving average model with exogenous variables to obtain a target model; A prediction module for predicting the traffic usage at the target time based on the target model and the feature data at the target time.

9. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete mutual communication through the communication bus; The memory is used for storing computer programs; The processor is used to implement the method according to any one of claims 1-7 when executing the program stored on the memory.

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

Images