Service load flow prediction method, device, equipment, medium and program product

Abstract

The invention provides a service load flow prediction method and device, equipment, a medium and a program product. The method comprises the following steps: acquiring historical insurance premium income data in a target time period; a trend component and a seasonal component in the historical insurance premium income data are extracted through a variational mode decomposition method; determining a predicted insurance premium income based on the trend component and the seasonal component by using an autoregression integral moving average model; and predicting the service load flow of the APP based on the predicted insurance premium income. By adopting the method provided by the embodiment of the invention, the accurate prediction of the APP service load flow can be realized, and the accuracy of the prediction result is improved, so that the prediction precision is improved, and the resource configuration is optimized.

Description

Technical Field

[0001] The present disclosure relates to the field of big data technology, and in particular to a method, device, equipment, medium and program product for predicting business load flow. Background Art

[0002] In the related technologies, with the popularization of mobile Internet and smart phones, APP (Application) has become an indispensable part of people's daily life. In order to ensure the stable operation and efficient response of APP, it is very important to reasonably predict the business load of APP. At present, the prediction of APP business load traffic usually relies on the historical data of APP access traffic or system monitoring data for prediction. Although this can realize the prediction of APP business load traffic, there is a deviation in the prediction result, and the accuracy of the prediction result is poor. Summary of the invention

[0003] The present disclosure provides a method, device, equipment, medium and program product for predicting business load flow. The technical solution of the present disclosure is as follows:

[0004] In a first aspect, the present disclosure provides a service load flow prediction method, comprising:

[0005] Obtain historical premium income data within the target period;

[0006] extracting trend components and seasonal components from the historical premium income data by using a variational mode decomposition method;

[0007] determining predicted premium income based on the trend component and the seasonal component using an autoregressive integrated moving average model;

[0008] The business load flow of the APP is predicted based on the predicted premium income.

[0009] In a possible implementation manner, before extracting the trend component and seasonal component from the historical premium income data by using the variational mode decomposition method, the method further includes:

[0010] Performing data preprocessing on the historical premium income data to obtain the historical premium income data after data preprocessing; wherein the data preprocessing includes at least one of sorting the historical premium income data in chronological order, performing logarithmic transformation on the historical premium income data, performing smoothing processing on the historical premium income data, and performing normalization processing on the smoothed historical premium income data;

[0011] The method of extracting trend components and seasonal components from the historical premium income data by using a variational mode decomposition method includes:

[0012] The trend component and seasonal component in the historical premium income data after the data preprocessing are extracted by variational mode decomposition method.

[0013] In a possible implementation, the extracting trend components and seasonal components from the historical premium income data by using a variational mode decomposition method includes:

[0014] Decomposing the historical premium income data into multiple modal components by using a variational modal decomposition method;

[0015] Among the multiple modal components, a trend component and a seasonal component are extracted; wherein the trend component is used to indicate the long-term trend of the premium income data, and the seasonal component is used to indicate the periodic changes of the premium income data.

[0016] In a possible implementation, the method of using the autoregressive integrated moving average model to determine the predicted premium income based on the trend component and the seasonal component includes:

[0017] Using the first autoregressive integrated moving average model to make a prediction based on the trend component to obtain a trend change value;

[0018] The second autoregressive integrated moving average model performs prediction based on the seasonal component to obtain a seasonal fluctuation value;

[0019] Based on the trend change value and the seasonal fluctuation value, the predicted premium income is determined.

[0020] In a possible implementation, the using the first autoregressive integrated moving average model to perform prediction based on the trend component to obtain the trend change value includes:

[0021] Performing a stationarity check on the trend component;

[0022] When the stability check of the trend component passes, prediction is performed based on the trend component to obtain a trend change value;

[0023] In the case that the stability check of the trend component fails, performing differential processing on the trend component to obtain a trend component after differential processing;

[0024] Prediction is performed based on the trend component after the differential processing to obtain a trend change value.

[0025] In a possible implementation manner, the predicting the service load flow of the APP based on the predicted premium income includes:

[0026] Get APP access traffic data;

[0027] Calculating the correlation coefficient between the APP access traffic data and the predicted premium income;

[0028] A linear regression model is used to predict the business load flow of the APP based on the correlation coefficient between the APP access flow data and the predicted premium income.

[0029] In a second aspect, the present disclosure provides a service load flow prediction device, comprising:

[0030] A data acquisition module is used to obtain historical premium income data within a target period;

[0031] A data extraction module, used for extracting trend components and seasonal components in the historical premium income data by using a variational mode decomposition method;

[0032] A premium forecasting module, used for determining the forecast premium income based on the trend component and the seasonal component using an autoregressive integrated moving average model;

[0033] The traffic prediction module is used to predict the business load traffic of the APP based on the predicted premium income.

[0034] In a third aspect, the present disclosure provides an electronic device, including:

[0035] processor;

[0036] a memory for storing instructions executable by the processor;

[0037] The processor is configured to execute the instructions to implement the method described in the first aspect.

[0038] In a fourth aspect, the present disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method described in the first aspect.

[0039] In a fifth aspect, the present disclosure provides a computer program product, comprising a computer program / instructions, characterized in that when the computer program / instructions are executed by a processor, the method described in the first aspect is implemented.

[0040] The technical solution disclosed in this disclosure brings at least the following beneficial effects:

[0041] In the embodiment of the present disclosure, the historical premium income data within the target period is obtained; the trend component and seasonal component in the historical premium income data are extracted by the variational mode decomposition method; the predicted premium income is determined based on the trend component and the seasonal component using the autoregressive integral moving average model; the business load flow of the APP is predicted based on the predicted premium income. In this way, the trend change and seasonal change of premium income, as well as the relationship with the APP access flow, combined with the VMD decomposition and ARIMA prediction model, can be used to achieve accurate prediction of the APP business load flow, improve the accuracy of the prediction results, thereby improving the prediction accuracy and optimizing resource allocation.

[0042] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute improper limitations on the present disclosure.

[0044] Figure 1 A flow chart of a method for predicting service load flow provided by an embodiment of the present disclosure;

[0045] Figure 2 It is a flow chart of a method for predicting business load flow provided by an embodiment of the present disclosure;

[0046] Figure 3 It is a structural diagram of a service load flow prediction device provided by an embodiment of the present disclosure;

[0047] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0048] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings.

[0049] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0050] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.) and signals involved in this disclosure are all authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions.

[0051] The acquisition, storage, use, and processing of data in the technical solution disclosed in this disclosure are in compliance with the relevant provisions of national laws and regulations.

[0052] It should be noted that in the embodiments of the present disclosure, there may be existing solutions in the industry such as certain software, components, models, etc., which should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present disclosure, but it does not mean that the applicant has or will necessarily use the solution.

[0053] In related technologies, with the popularization of mobile Internet and smart phones, APP applications have become an indispensable part of people's daily lives. However, with the rapid increase in the number of APP users, APP access traffic and system load are facing unprecedented challenges. In order to ensure the stable operation and efficient response of APP, it is crucial to reasonably predict the business load of APP. At present, the methods for predicting APP access traffic are mainly divided into the following categories, but these methods have certain limitations and cannot fully meet actual needs:

[0054] (1) Prediction methods that rely on a single data source. Many existing technologies rely on historical data of APP access traffic or system monitoring data for prediction. For example, the model is trained using APP server logs, network request data, user behavior data, etc. However, these methods only focus on system-level data and ignore the impact of external business data (such as premium income, sales data, etc.) on traffic. Especially when it comes to businesses with obvious seasonal changes, prediction models that rely solely on system data often fail to reflect the actual impact of seasonal fluctuations on APP traffic, resulting in deviations in the prediction results.

[0055] (2) Lack of accurate modeling of seasonal fluctuations. Existing traffic forecasting methods mostly use simple time series analysis (such as ARIMA (AutoRegressive Integrated Moving Average Model), exponential smoothing, etc.) for forecasting. These methods generally assume that the trend and seasonality of the data are stable and cannot cope well with situations with large seasonal fluctuations and strong periodicity. For businesses with complex seasonal changes (such as insurance, retail, etc.), these traditional models may lead to low forecasting accuracy due to ignoring external factors (such as fluctuations in premium income, holiday effects, etc.).

[0056] (3) Ignoring the correlation of external business data. Many traffic prediction models only consider the historical traffic data of the APP itself, but do not consider the impact of external business data (such as industry data, market trends, macroeconomic changes, etc.) on traffic. In some business scenarios, there may be a strong correlation between the fluctuation of external business data and APP traffic. For example, in the insurance industry, seasonal changes in premium income will directly affect the access traffic of the APP, but existing prediction methods usually ignore this correlation. Therefore, the prediction results of traditional models often fail to fully consider the multidimensional factors that affect traffic, resulting in large prediction errors.

[0057] Based on this, the embodiments of the present disclosure provide a method, device, equipment, medium and program product for predicting business load flow, which can combine the relationship between seasonal changes in premium income and APP access flow to achieve accurate and efficient business load prediction. Specifically, it can be used to solve the following technical problems:

[0058] (1) How to effectively utilize the impact of external business data (such as premium income) on APP access traffic. In related technologies, traffic prediction mostly relies on system data and lacks effective integration of external business data. External factors such as premium income and sales data have an important impact on APP access traffic, especially in business scenarios with large seasonal fluctuations. The impact of these external data cannot be ignored. Therefore, the embodiment of the present disclosure proposes a traffic prediction method that introduces external business data such as premium income, combines VMD (Variational Mode Decomposition) decomposition and ARIMA prediction model, and more accurately reflects the impact of external factors on APP traffic.

[0059] (2) How to accurately capture the seasonal fluctuations in premium income and correlate them with changes in APP access traffic to model. Many existing methods have limited modeling capabilities for seasonal fluctuations, especially in businesses with large fluctuations and complex cyclical changes. The disclosed embodiment of the present invention separates the trend component and the seasonal component in the premium income data through VMD decomposition, thereby better capturing the laws of seasonal fluctuations, combining it with the ARIMA model to predict it, and introducing the prediction results as exogenous variables into the APP traffic prediction model, thereby improving the modeling accuracy of seasonal fluctuations.

[0060] (3) How to combine historical data of premium income to predict future APP access traffic and provide a basis for load management. The method proposed in the implementation of this disclosure can combine historical data of premium income with APP access traffic data, and determine the relationship between the two through a regression model or an ARIMAX model, thereby using the predicted data of premium income to predict future APP access traffic. This prediction method that combines business data and system data can more comprehensively consider various influencing factors in traffic prediction, and provide a more scientific and accurate basis for APP load management and resource allocation.

[0061] (4) How to eliminate the prediction bias of traditional methods and improve the accuracy of prediction. Traditional prediction methods usually rely on a single data source, ignoring the fluctuations and influences of external factors, which easily lead to deviations in prediction results. The disclosed embodiment combines VMD decomposition, ARIMA model and regression analysis, fully considering the relationship between external factors and system internal factors, thereby improving prediction accuracy and solving the problem that traditional methods are difficult to cope with complex fluctuations and seasonal changes.

[0062] In summary, the embodiments of the present disclosure can achieve accurate prediction of APP business load by introducing the relationship between seasonal changes in premium income and APP access traffic, combining VMD decomposition and ARIMA prediction models, thereby improving prediction accuracy and optimizing resource allocation.

[0063] In the disclosed embodiment, premium income refers to the total income obtained by an insurance institution through the sale of insurance products. It is one of the key indicators of the operation of an insurance institution and reflects its business volume and market share. As business data, premium income data can affect the user traffic and usage pattern of the APP, and thus affect the load demand of the system. In the disclosed embodiment, the APP business load is predicted by analyzing the seasonal changes in premium income data. APP traffic generally refers to the total number of user requests or visits initiated in an application (APP) within a specific time period. This generally includes the number of user logins, data queries, content browsing and other behaviors. In the disclosed embodiment, APP traffic is the core goal of predicting the load, which refers to the user access volume and data traffic that the APP needs to process within a certain time period. Accurate prediction of APP traffic can help optimize server resource scheduling and system load. Time series data is data collected in chronological order and has a chronological dependency. In data analysis, time series data generally represents the observed values ​​of a variable at different time points. In the disclosed embodiment, premium income and APP traffic are both time series data because they are data collected at different time points. The analysis and prediction of time series data is crucial to this patented solution and helps predict future business loads.

[0064] VMD is a signal processing method that extracts different frequency components of the signal by decomposing a complex signal into multiple IMFs (Intrinsic Mode Functions). This method has better signal decomposition capabilities than traditional signal decomposition methods (such as EEMD (Ensemble Empirical Mode Decomposition). In the disclosed embodiment, VMD is used to decompose the time series data of premium income into multiple components, mainly divided into trend components and seasonal components, so as to predict them separately. Through this decomposition, trends and seasonal fluctuations in the data can be more clearly identified and captured, and the prediction accuracy can be improved. ARIMA is a statistical model commonly used in time series prediction, which consists of autoregression (AR) terms, difference (I) terms, and moving average (MA) terms. The model is particularly suitable for non-stationary time series data and can capture the autocorrelation of data. In the disclosed embodiment, ARIMA is used to independently predict the trend component and seasonal component after VMD decomposition. The ARIMA model can more accurately predict future premium income and APP traffic changes by considering the autocorrelation and trend of historical data.

[0065] The technical solutions provided by various embodiments of the present disclosure are described in detail below in conjunction with the accompanying drawings.

[0066] Figure 1The flowchart of a service load flow prediction method provided by the embodiment of the present disclosure is as follows. The method can be applied to a server, for example, a single server or a server cluster. Figure 1 As shown, the service load flow prediction method may include the following steps:

[0067] S101, obtaining historical premium income data within a target period.

[0068] In the embodiment of the present disclosure, when predicting the business load flow of the APP, data acquisition can be performed first. For example, historical premium income data within a certain time period (i.e., the target period) can be acquired. It can be understood that the historical premium income data can be time series data R, and the specific time period of the target period can be set according to actual needs.

[0069] S102, extracting trend components and seasonal components from historical premium income data through variational mode decomposition method.

[0070] In an embodiment of the present disclosure, after obtaining the historical premium income data within the target period, the trend component and seasonal component in the historical premium income data can be obtained. Exemplarily, the trend component and seasonal component in the historical premium income data can be extracted by VMD. The VMD method attempts to minimize the bandwidth of each mode through variational derivation to obtain the intrinsic mode of the signal. Each mode contains a different frequency component, so VMD can effectively separate the different frequency components in the signal, which is particularly suitable for analyzing non-stationary time series data. Among them, the trend component, corresponding to the low-frequency part, can reflect the long-term trend changes of the data; the seasonal component, corresponding to the high-frequency part, can reflect the periodic fluctuations of the data. In this way, the long-term trend and periodic changes in the data can be separated to provide clear input for subsequent ARIMA forecasts. Through the adaptive characteristics of VMD, valuable information can be efficiently extracted from non-stationary premium income data to improve the accuracy and stability of the prediction.

[0071] S103, using an autoregressive integrated moving average model to determine the forecasted premium income based on the trend component and the seasonal component.

[0072] In an embodiment of the present disclosure, after extracting the trend component and seasonal component from the historical premium income data, the premium income is predicted. For example, the ARIMA model (regression integral moving average model) can be used to predict the future premium income based on the trend component and seasonal component in the historical premium income data to obtain the predicted premium income. The ARIMA model is a classic time series prediction method, which is suitable for processing non-stationary data with a certain time dependency. By using the ARIMA model, the long-term trend changes and cyclical fluctuations of the premium income data can be accurately predicted, thereby providing the necessary input for the subsequent business load prediction. It can be understood that before the premium income prediction is performed, ARIMA modeling can be performed first, and the fitted ARIMA model can be used to predict the premium income of the future trend component and seasonal component to obtain the predicted premium income. In this way, the ARIMA model can be used to model and predict the long-term trend and cyclical fluctuations, thereby providing accurate premium income data for the APP business load prediction. In this way, the future trend of premium income can be predicted more accurately, and the stability and accuracy of the prediction can be improved.

[0073] S104, predicting the business load flow of the APP based on the predicted premium income.

[0074] In an embodiment of the present disclosure, after determining the predicted premium income, the business load flow of the premium income prediction APP can be predicted. Exemplarily, the correlation between the premium income data and the APP access flow data can be calculated to find the intrinsic relationship between them, and the business load flow of the APP can be predicted based on the correlation between the premium income data and the APP access flow data. In this way, the business load flow of the APP can be predicted by combining the trend changes and seasonal changes of the premium income data, as well as the correlation between the premium income data and the APP flow, thereby improving the accuracy of predicting the business load flow of the APP.

[0075] In the embodiment of the present disclosure, the historical premium income data within the target period is obtained; the trend component and seasonal component in the historical premium income data are extracted by the variational mode decomposition method; the predicted premium income is determined based on the trend component and the seasonal component using the autoregressive integral moving average model; the business load flow of the APP is predicted based on the predicted premium income. In this way, the trend change and seasonal change of premium income, as well as the relationship with the APP access flow, combined with the VMD decomposition and ARIMA prediction model, can be used to achieve accurate prediction of the APP business load flow, improve the accuracy of the prediction results, thereby improving the prediction accuracy and optimizing resource allocation.

[0076] In some possible implementations, before extracting the trend component and seasonal component from the historical premium income data by using the variational mode decomposition method, the method further includes:

[0077] Performing data preprocessing on the historical premium income data to obtain the historical premium income data after data preprocessing; wherein the data preprocessing includes at least one of sorting the historical premium income data in chronological order, performing logarithmic transformation on the historical premium income data, performing smoothing processing on the historical premium income data, and performing normalization processing on the smoothed historical premium income data;

[0078] The variational mode decomposition method is used to extract the trend and seasonal components in the historical premium income data, including:

[0079] The trend component and seasonal component of the historical premium income data after data preprocessing are extracted by variational mode decomposition method.

[0080] In an embodiment of the present disclosure, the historical premium income data may be preprocessed before extracting the trend component and seasonal component in the historical premium income data by the variational mode decomposition method, so as to make the historical premium income data more accurate and easier to process later. Exemplarily, the historical premium income data may be sorted in chronological order. For example, the historical premium income data within the target time period may be obtained and sorted in chronological order to ensure that the data is arranged continuously in time. The input data of the process is the original premium income time series data, and the output data is the premium income data sorted in chronological order. Then, the historical premium income data sorted in chronological order may be logarithmically transformed to eliminate the influence of skewed distribution on the analysis results. The logarithmically transformed historical premium income data can reduce the excessive influence of high income values ​​on the model and make the historical premium income data more consistent with the normal distribution. The logarithmic transformation formula may be as shown in formula (1). The input data of the process is the historical premium income data sorted in chronological order, and the output data is the logarithmically transformed historical premium income data.

[0081] R log =log(R) (1)

[0082] Among them, R is the original premium income data, R log are log-transformed data.

[0083] After that, the historical premium income data can also be smoothed. For example, the SG (Savitzky-Golay) filter can be used to smooth the historical premium income data after logarithmic transformation to remove noise and retain the long-term trend of the data. The SG filter can smooth the time series data by fitting the local data with a polynomial, thereby removing high-frequency noise and retaining the trend characteristics of the original data. The filtering formula can be entered into formula (2). The input data of this process is the historical premium income data after logarithmic transformation, and the output data is the smoothed historical premium income data.

[0084] R smooth =SG(R log ) (2)

[0085] Among them, R log is the logarithmic transformed data, R smooth The data is smoothed by SG filtering.

[0086] Finally, the smoothed data can also be normalized to ensure that the data is in the same dimension and to avoid the unbalanced impact of data of different magnitudes on the subsequent model. The commonly used normalization method is minimum-maximum normalization. The normalization formula is as shown in formula (3). The input data of this process is the smoothed historical premium income data, and the output data is the normalized premium income data. Accordingly, when extracting the trend component and seasonal component in the historical premium income data through the variational mode decomposition method, the trend component and seasonal component in the historical premium income data after data preprocessing can also be extracted through the variational mode decomposition method.

[0087]

[0088] Among them, R smooth is the smoothed data, R min and R max are the minimum and maximum values ​​in the data, R normalized is the normalized data.

[0089] In some possible implementations, the trend component and seasonal component in the historical premium income data are extracted by using a variational mode decomposition method, including:

[0090] The historical premium income data is decomposed into multiple modal components by using the variational mode decomposition method;

[0091] Among the multiple modal components, trend components and seasonal components are extracted; wherein the trend component is used to indicate the long-term trend of the premium income data, and the seasonal component is used to indicate the cyclical changes of the premium income data.

[0092] In the embodiments of the present disclosure, the VMD method can be derived through variational calculus to try to minimize the bandwidth of each mode to obtain the intrinsic modal components of the signal. Each modal component contains a different frequency component. Therefore, VMD can effectively separate the different frequency components in the signal and is particularly suitable for analyzing non-stationary time series data. Assume that the input signal x(t) can be decomposed into multiple modal signals u k (t), k = 1, 2, ..., K, where K is the number of modes, and each modal signal u k (t) has a specific center frequency u k (t) and bandwidth constraints. The goal of VMD is to decompose the signal through an optimization process and minimize the bandwidth of each mode. The VMD process can be implemented through an iterative optimization algorithm, which is optimized based on the following objective function, as shown in formula (4). After VMD decomposition, the original premium income data will be decomposed into multiple intrinsic mode components, which mainly include the following two categories: one is the trend component, which corresponds to the low-frequency part and reflects the long-term trend changes of the data. The other is the seasonal component, which corresponds to the high-frequency part and reflects the cyclical fluctuations of the data. The trend component and the seasonal component will be used for subsequent predictive modeling. For example, if the decomposed modes are u1(t), u2(t),…,u K (t), it can be classified into trend component and seasonal component according to the characteristics of frequency or time domain. There is usually only one trend component, but there can be multiple seasonal components.

[0093]

[0094] Among them, u k (t) is the decomposed mode, ω k is the center frequency of each mode, α is the penalty term coefficient used to control the frequency smoothness, and ω0 is the reference frequency.

[0095] In this way, the historical premium income data is decomposed by VMD, and the pre-processed premium income data is decomposed into multiple modal components using the VMD method, from which trend components and seasonal components are extracted. In this way, the long-term trend and cyclical changes in the data can be separated, providing clear input for subsequent ARIMA forecasts. Through the adaptive characteristics of VMD, valuable information can be efficiently extracted from non-stationary premium income data, improving the accuracy and stability of the forecast.

[0096] In some possible implementations, the predicted premium income is determined based on the trend component and the seasonal component using an autoregressive integrated moving average model, including:

[0097] Use the first autoregressive integrated moving average model to make predictions based on the trend component and obtain the trend change value;

[0098] The second autoregressive integrated moving average model is used to predict seasonal components and obtain seasonal fluctuation values;

[0099] Determine the forecast premium income based on trend change values ​​and seasonal fluctuation values.

[0100] In an embodiment of the present disclosure, the ARIMA (autoregressive integrated moving average) model can be used to predict the trend component and the seasonal component respectively. The ARIMA model is a classic time series prediction method, which is suitable for processing non-stationary data with a certain time dependency. By using the ARIMA model, the long-term trend changes and cyclical fluctuations of historical premium income data can be accurately predicted, thereby providing the necessary input for subsequent business load prediction. Exemplarily, the trend component can be ARIMA modeled to obtain an autoregressive integrated moving average model (also known as the first autoregressive integrated moving average model), and the trend component is predicted by the first autoregressive integrated moving average model, as follows: and the prediction is as follows:

[0101] First, the trend component can be extracted. The trend component is the long-term change part in the premium income data, which usually contains lower frequency components. Before using the ARIMA model for prediction, ARIMA modeling can be performed to perform ARIMA modeling on the trend component data and select appropriate p, d and parameters q. The BIC (Bayesian Information Criterion) method is used to select parameters and optimize the fitting effect of the model. The prediction formula of the ARIMA model is as follows (5). Finally, predict the trend component. Use the fitted ARIMA model to predict the future trend component and obtain the trend change value in the future. The input data of this process is the trend component data, and the output data is the future predicted value of the trend component.

[0102]

[0103] Among them, Y t is the value at the current moment, μ is a constant term, φ i is the coefficient of the autoregressive term, Y t-i is the past observation, θ i is the coefficient of the sliding average term, ε t-i is the past error, ε t is the error term at the current moment (usually white noise).

[0104] You can also perform ARIMA modeling on the seasonal component to obtain an autoregressive integrated moving average model (also known as the second autoregressive integrated moving average model), and use the second autoregressive integrated moving average model to predict the seasonal component as follows:

[0105] First, extract the seasonal component. The seasonal component reflects the cyclical fluctuations in the historical premium income data, and usually contains higher frequency changes. When performing ARIMA modeling, it is necessary to identify the cyclical characteristics of the seasonal component and make corresponding adjustments. Perform ARIMA modeling on the seasonal component. The modeling method is the same as modeling the trend component, which will not be repeated here. After that, the seasonal component can be predicted, and the future seasonal component can be predicted using the fitted seasonal ARIMA model to obtain the seasonal fluctuation value in the future. The input data of this process is the seasonal component data, and the output data is the future predicted value of the seasonal component. It can be understood that the ARIMA modeling process can be pre-fitted.

[0106] Finally, the future forecast value of the trend component and the future forecast value of the seasonal component can be combined to obtain the ARIMA forecast result, that is, the forecast premium income. For example, the final forecast result can be a synthesis of the trend component and the seasonal component, expressed as the forecast value of future premium income, that is, the forecast premium income, as shown in the following formula (6).

[0107]

[0108] in, is the forecast value of future premium income, is the predicted value of the trend component, is the predicted value of the seasonal component.

[0109] In this way, the ARIMA model can be used to predict the trend component and seasonal component, and the future premium income can be predicted by performing ARIMA modeling on the trend component and seasonal component respectively. In this way, the ARIMA model can be used to model and predict long-term trends and cyclical fluctuations, thereby providing accurate premium income data for APP business load prediction. In this way, the future trend of premium income can be predicted more accurately, and the stability and accuracy of the prediction can be improved.

[0110] In a possible implementation, a first autoregressive integrated moving average model is used to perform prediction based on the trend component to obtain a trend change value, including:

[0111] Perform a stationarity check on the trend component;

[0112] When the stability check of the trend component is passed, prediction is performed based on the trend component to obtain the trend change value;

[0113] When the trend component stability check fails, the trend component is subjected to differential processing to obtain the trend component after differential processing;

[0114] Predictions are made based on the trend components after differential processing to obtain trend change values.

[0115] In an embodiment of the present disclosure, a prediction is made based on a trend component using the first autoregressive integrated moving average model. Before that, the trend component may be checked for stationarity, for example, by MK (Mann-Kendall) test, DF (Dickey-Fuller Test), ADF (Augmented Dickey-Fuller Test), KPSS (Kwiatkowski-Phillips-Schmidt-Shin Test), graphical method, autocorrelation graph test and other methods. If the trend component data is not stationary, differential processing may be performed to make the data stable. Then, the trend component after differential processing may be predicted to obtain a trend change value. If the trend component passes the stationarity check, a prediction may be made based on the trend component to obtain a trend change value. In this way, the prediction accuracy of the trend change value may be further improved.

[0116] In a possible implementation, predicting the service load flow of the APP based on the predicted premium income includes:

[0117] Get APP access traffic data;

[0118] Calculate the correlation coefficient between APP access traffic data and predicted premium income;

[0119] A linear regression model is used to predict the business load traffic of the APP based on the correlation coefficient between the APP access traffic data and the predicted premium income.

[0120] In an embodiment of the present disclosure, when predicting the business load flow of an APP based on the predicted premium income, the APP access flow data can be first obtained. The APP access flow data generally refers to the number of users accessing the application within a certain time range, page views or other metrics. In this step, it is necessary to ensure that the access flow data is also properly preprocessed to ensure that it is aligned with the time of the premium income data. Then, the correlation coefficient between the APP access flow data and the predicted premium income can be calculated. For example, a statistical method can be used to calculate the Pearson correlation coefficient (Pearson Correlation Coefficient), which measures the strength and direction of the linear relationship between two variables. The calculation formula of the Pearson correlation coefficient is as shown in formula (7).

[0121]

[0122] Among them, X i and Y iare the values ​​of premium income and APP traffic at the i-th time point respectively. and are the means of premium income and APP traffic, respectively. ρ(X,Y) is the correlation coefficient between premium income and APP traffic, ranging from [-1,1]. ρ(X,Y) is close to 1, indicating that there is a strong positive correlation between premium income and APP traffic; ρ(X,Y) is close to -1, indicating that there is a strong negative correlation between premium income and APP traffic. ρ(X,Y) is close to 0, indicating that there is almost no linear relationship between the two.

[0123] By calculating the correlation coefficient, the impact factor of premium income on APP traffic can be determined. If the correlation coefficient between premium income and APP traffic is very high, it means that premium income has a strong impact on APP access traffic, otherwise it means that the relationship between the two is weak. The correlation coefficient can be used as an impact factor multiplier. In calculating the correlation between premium income and APP traffic, by calculating the correlation between premium income and APP traffic, the impact factor of premium income on APP load is analyzed. This process provides crucial basic data for subsequent business load forecasting, and more accurately understands how business data drives APP traffic, thereby improving the accuracy and practicality of the forecast.

[0124] Then, the business load flow of the APP can be predicted by the prediction model. Exemplarily, a linear regression model can be used to predict the business load flow of the APP based on the correlation coefficient between the APP access flow data and the predicted premium income. For example, the above-mentioned future premium income data (i.e., predicted premium income), the correlation coefficient between the predicted premium income and the APP access flow data, and the historical APP flow data (i.e., APP access flow data) can be prepared as input data. The historical APP flow data can be used to establish an initial prediction model to help understand the trend of APP flow and adjust the APP flow according to the change in premium income. After that, the regression model can be used to predict the APP flow. In this step, a regression analysis method can be used to predict the business load flow of the APP using the correlation coefficient between the predicted premium income data and the historical APP access flow data. For example, a linear regression model can be used, and a comprehensive prediction can be made in combination with the influencing factors obtained from the correlation analysis. The linear regression model assumes that there is a linear relationship between the business load flow of the APP and the predicted premium income data. The prediction formula can be as shown in formula (8).

[0125] Y flow =α+β·X premium +ε (8)

[0126] Among them, Y flow is the predicted APP traffic, X premiumis the forecast data of premium income, α is the intercept term of the regression model, β is the regression coefficient, which indicates the impact factor of premium income on APP traffic, and ε is the error term, which indicates the forecast error of the model. By training the linear regression model, α and β can be obtained, and according to the premium income forecast data X premium To predict the corresponding APP traffic Y flow .

[0127] In this way, by using the predicted premium income data to predict the business load traffic of the APP, the changes in premium income can be used to predict future APP access traffic. This process combines the trend changes and seasonal changes of premium income data and the correlation with APP access traffic data, providing a more accurate model for APP business load traffic prediction. Ultimately, the prediction results can help decision makers better plan the resource scheduling and optimization of the APP, thereby improving the stability of the system and user experience.

[0128] To make the service load flow prediction method provided by the embodiment of the present disclosure clearer, the following is combined with Figure 2 As shown in Figure 2, the service load flow prediction method provided by the embodiment of the present disclosure includes:

[0129] Step 1: Preprocessing of premium income data, including:

[0130] (1) Data acquisition and sorting

[0131] Obtain the historical data of premium income within the target time period (i.e., historical premium income data) and sort it in chronological order to ensure that the data is arranged continuously in time.

[0132] Input: Original premium income time series data R.

[0133] Output: Premium income data sorted in chronological order.

[0134] (2) Logarithmic transformation

[0135] The premium income data is logarithmically transformed to eliminate the impact of skewed distribution on the analysis results. The logarithmically transformed data can reduce the excessive impact of high income values ​​on the model and make the data more consistent with normal distribution.

[0136] The logarithmic transformation formula is as follows:

[0137] R log =log(R)

[0138] Among them, R is the original premium income data, R log are log-transformed data.

[0139] Input: Premium income data sorted by time.

[0140] Output: Logarithmically transformed premium income data.

[0141] (3) Smoothing (SG filtering)

[0142] SG filtering (Savitzky-Golay filtering) is used to smooth the data, remove noise and retain the long-term trend of the data. SG filtering smoothes the time series data by fitting the local data with a polynomial, thereby removing high-frequency noise and retaining the trend characteristics of the original data. The formula is as follows:

[0143] R smooth =SG(R log )

[0144] Among them, R log is the logarithmic transformed data, R smooth The data is smoothed by SG filtering.

[0145] Input: Logarithmically transformed premium income data.

[0146] Output: Smoothed premium income data.

[0147] (4) Data normalization

[0148] Normalize the smoothed data to ensure that the data is in the same dimension and avoid the unbalanced impact of data of different magnitudes on the subsequent model. The commonly used normalization method is minimum-maximum normalization. The normalization formula is as follows:

[0149]

[0150] Among them, R smooth is the smoothed data, R min and R max are the minimum and maximum values ​​in the data, R normalized is the normalized data.

[0151] Input: Smoothed premium income data.

[0152] Output: Normalized premium income data.

[0153] Step 2 VMD decomposes premium income data, including:

[0154] In this step, the preprocessed premium income data is decomposed using variational mode decomposition (VMD). VMD is an adaptive signal decomposition method that can decompose complex time series signals into multiple intrinsic mode functions (IMFs), where each IMF represents a different frequency component in the data. Through VMD decomposition, the trend component and seasonal component in the data can be extracted, further providing valuable input for subsequent ARIMA forecasting. The specific processing process is as follows:

[0155] (1) VMD decomposition principle. The VMD method attempts to minimize the bandwidth of each mode through variational derivation to obtain the intrinsic mode of the signal. Each mode contains a different frequency component. Therefore, VMD can effectively separate the different frequency components in the signal and is particularly suitable for analyzing non-stationary time series data.

[0156] Assume that the input signal x(t) can be decomposed into multiple modal signals u k (t), k = 1, 2, ..., K, where K is the number of modes and each modal signal u k (t) has a specific center frequency ω k and bandwidth constraints. The goal of VMD is to decompose the signal and minimize the bandwidth of each mode through an optimization process.

[0157] (2) Decomposition process. The specific VMD process is implemented through an iterative optimization algorithm, which is optimized based on the following objective function:

[0158]

[0159] Among them, u k (t) is the decomposed mode, ω k is the center frequency of each mode, α is the penalty term coefficient used to control the frequency smoothness, and ω0 is the reference frequency.

[0160] (3) Extraction of decomposed components. After VMD decomposition, the original premium income data will be decomposed into multiple intrinsic modal components, which mainly include the following two categories: one is the trend component, which corresponds to the low-frequency part and reflects the long-term trend change of the data. The other is the seasonal component, which corresponds to the high-frequency part and reflects the cyclical fluctuation of the data. In this step, the trend component and seasonal component will be used for subsequent prediction modeling. For example, if the decomposed modes are u1(t), u2(t), ..., u K (t), it can be classified into trend component and seasonal component according to the characteristics of frequency or time domain. There is usually only one trend component, but there can be multiple seasonal components.

[0161] In Step 2, VMD decomposes the premium income data. The preprocessed premium income data is decomposed into multiple modal components using the VMD method, and trend components and seasonal components are extracted. In this way, the long-term trend and cyclical changes in the data can be separated, providing clear input for subsequent ARIMA forecasts. Through the adaptive characteristics of VMD, valuable information can be efficiently extracted from non-stationary premium income data, improving the accuracy and stability of the forecast.

[0162] Step 3 Use the ARIMA model to predict trend and seasonal components

[0163] In this step, the ARIMA (Autoregressive Integrated Moving Average) model will be used to predict the trend component and seasonal component respectively. The ARIMA model is a classic time series prediction method, which is suitable for processing non-stationary data with a certain time dependency. By using the ARIMA model, the long-term trend changes and cyclical fluctuations of premium income data can be accurately predicted, thereby providing the necessary input for subsequent business load prediction. Including:

[0164] (1) ARIMA modeling and forecasting of trend components

[0165] First, extract the trend component. The trend component obtained from step 2 is the long-term change in the premium income data, which usually contains lower frequency components. Before using the ARIMA model for prediction, first perform a stationarity test on the trend component. If the data is not stationary, perform a difference process.

[0166] Secondly, perform ARIMA modeling. Perform ARIMA modeling on the trend component data and select appropriate , and parameters. Use the BIC (Bayesian Information Criterion) method to select parameters and optimize the model's fitting effect. The prediction formula of the ARIMA model is as follows:

[0167]

[0168] Among them, Y t is the value at the current moment, μ is a constant term, φ i is the coefficient of the autoregressive term, Y t-i is the past observation, θ i is the coefficient of the sliding average term, ε t-i is the past error, ε t is the error term at the current moment (usually white noise).

[0169] Finally, the trend component is predicted. The fitted ARIMA model is used to predict the future trend component and obtain the trend change value in the future period.

[0170] Input: Trend component data.

[0171] Output: Future forecast values ​​of the trend component.

[0172] (2) ARIMA modeling and forecasting of seasonal components

[0173] First, extract the seasonal component. The seasonal component extracted from step 2 reflects the cyclical fluctuations in the premium income data, which usually contains higher frequency changes. When performing ARIMA modeling, it is necessary to identify the cyclical characteristics of the seasonal component and make corresponding adjustments.

[0174] Secondly, ARIMA modeling is performed on the seasonal component, and the modeling method is the same as the prediction of the trend component.

[0175] Finally, the seasonal components are predicted. The fitted seasonal ARIMA model is used to predict the future seasonal components and obtain the seasonal fluctuation values ​​in the future.

[0176] Input: Seasonal component data.

[0177] Output: Future forecasts of seasonal components.

[0178] (3) Combination of ARIMA forecast results

[0179] In this step, ARIMA modeling and forecasting are performed on the trend component and seasonal component respectively. The final forecast result will be the synthesis of the trend component and the seasonal component, expressed as the forecast value of future premium income, as follows:

[0180]

[0181] in, is the forecast value of future premium income, is the predicted value of the trend component, is the predicted value of the seasonal component.

[0182] In Step 3, using the ARIMA model to predict trend components and seasonal components, future premium income is predicted by performing ARIMA modeling on trend components and seasonal components respectively. This process uses the ARIMA model to model and predict long-term trends and cyclical fluctuations, thereby providing accurate premium income data for APP business load forecasting. In this way, the future trend of premium income can be predicted more accurately, improving the stability and accuracy of the forecast.

[0183] Step 4 Use the ARIMA model to predict trend and seasonal components

[0184] The goal of Step 4 is to find out the intrinsic relationship between premium income data and APP access traffic data by calculating the correlation between them. This step is crucial to understanding and modeling how premium income affects APP load. Through correlation analysis, it is possible to identify which changes in premium income will directly affect APP access traffic and provide key inputs for subsequent traffic forecasts, including:

[0185] (1) Acquisition of premium income and APP access traffic data

[0186] Premium income data: The premium income data is preprocessed and VMD decomposed through Step 1 and Step 2 to obtain trend components and seasonal components. These data represent the time series of premium income, and after stabilization and denoising, they can more accurately reflect the actual premium income situation.

[0187] APP access traffic data: APP access traffic usually refers to the number of users accessing the application within a certain time range, page views or other metrics. In this step, it is necessary to ensure that the access traffic data is also properly pre-processed to ensure that it is aligned with the time of premium income data.

[0188] (2) Calculate correlation

[0189] To calculate the correlation between premium income and APP traffic, the statistical method used in the present invention is the Pearson Correlation Coefficient, which measures the strength and direction of the linear relationship between two variables. The calculation formula of the Pearson Correlation Coefficient is:

[0190]

[0191] Among them, X i and Y i They are the values ​​of premium income and APP traffic at the th time point respectively. and are the means of premium income and APP traffic, respectively. ρ(X,Y) is the correlation coefficient between premium income and APP traffic, ranging from [-1,1]. If it is close to 1, it means there is a strong positive correlation between premium income and APP traffic; if it is close to -1, it means there is a strong negative correlation between premium income and APP traffic. If it is close to 0, it means there is almost no linear relationship between the two.

[0192] (3) Determine the impact factor

[0193] By calculating the correlation coefficient, we can determine the impact factor of premium income on APP traffic. If the correlation coefficient between premium income and APP traffic is very high, it means that premium income has a strong impact on APP access traffic, otherwise it means that the relationship between the two is weak. Here, the correlation coefficient can be used as an impact factor multiplier.

[0194] In Step 4, we calculated the correlation between premium income and APP traffic and analyzed the factors affecting premium income on APP load. This process provides crucial basic data for subsequent business load forecasting, helping us to more accurately understand how business data drives APP traffic, thereby improving the accuracy and practicality of forecasting.

[0195] Step 5 Use the premium income forecast data to predict APP traffic

[0196] In this step, the predicted premium income data will be used to predict future APP traffic. Specifically, given the known premium income, a suitable prediction model will be established to predict the APP access traffic within a certain period of time in the future. This process will be combined with the aforementioned correlation analysis results to further refine the traffic prediction.

[0197] (1) Input data preparation

[0198] Premium income forecast data: The future premium income data obtained using the ARIMA model in Step 3. The trend component and seasonal component of premium income have been predicted separately by ARIMA.

[0199] Correlation analysis results: Through the correlation between premium income and APP traffic calculated in Step 4, we obtained the impact factor of premium income on APP traffic. The purpose of this step is to understand the impact of changes in premium income on APP traffic so that we can introduce this impact factor into the forecasting process.

[0200] Historical APP traffic data: used to build an initial forecasting model to help understand APP traffic trends and adjust APP traffic based on changes in premium income.

[0201] (2) Using regression models to predict app traffic

[0202] In this step, regression analysis is used to predict APP traffic using the relationship between premium income forecast data and historical APP traffic data. A linear regression model is used, combined with the influencing factors obtained from the correlation analysis, to make a comprehensive prediction. As follows:

[0203] The linear regression model assumes that there is a linear relationship between APP traffic and premium income, and the prediction formula is:

[0204] Y flow =α+β·X premium +ε

[0205] Among them, Y flow is the predicted APP traffic, X premium is the forecast data of premium income, α is the intercept term of the regression model, β is the regression coefficient, which indicates the impact factor of premium income on APP traffic, and ε is the error term, which indicates the forecast error of the model. By training the linear regression model, α and β can be obtained, and according to the premium income forecast data X premium To predict the corresponding APP traffic.

[0206] By using premium income forecast data to predict APP traffic in Step 5, changes in premium income can be used to predict future APP traffic. This process combines the trend changes and seasonal changes in premium income data and the correlation with APP traffic to provide a more accurate model for APP load forecasting. Ultimately, the forecast results can help decision makers better plan APP resource scheduling and optimization, thereby improving system stability and user experience.

[0207] In summary, the business load flow prediction method provided by the embodiment of the present disclosure can use historical premium income data as external business data and combine it with the business load flow prediction of the APP. In this way, by introducing external business data (such as premium income), not only system factors are considered, but also the volatility of business operations is integrated, and it has better applicability and prediction accuracy in real scenarios. It is also possible to perform refined preprocessing on the premium income data, including logarithmization, smoothing, normalization, etc., and extract trend and seasonal components through VMD decomposition. In this way, the noise problem in the time series data can be effectively solved and the stability and accuracy of the prediction can be improved; VMD decomposition makes it possible to predict trend components and seasonal components separately, thereby reducing the interference of various uncertain factors on model prediction. In addition, by calculating the dynamic correlation between premium income and APP traffic, the business cycle fluctuation factor is introduced, so that the prediction results are closer to the actual operation situation. In this way, the intrinsic connection between business factors and system load can be better reflected. At the same time, the application of linear regression models can successfully combine premium income data with APP traffic prediction, thereby providing a simple, efficient and easy-to-implement traffic prediction method, while avoiding the overfitting problem of complex deep learning models on small data sets.

[0208] The specific implementation and technical effects of each step of this embodiment are similar to those of the above method embodiments, and will not be repeated here.

[0209] Based on the same inventive concept, the embodiment of the present disclosure also provides a service load flow prediction device. Figure 3 As shown, the service load flow prediction device 300 includes:

[0210] The data acquisition module 310 is used to acquire the historical premium income data within the target period;

[0211] A data extraction module 320, for extracting trend components and seasonal components from the historical premium income data by using a variational mode decomposition method;

[0212] A premium prediction module 330, for determining a predicted premium income based on the trend component and the seasonal component using an autoregressive integrated moving average model;

[0213] The traffic prediction module 340 is used to predict the business load traffic of the APP based on the predicted premium income.

[0214] In a possible implementation manner, the service load flow prediction device 300 further includes:

[0215] A data processing module, used for performing data preprocessing on the historical premium income data to obtain the historical premium income data after data preprocessing; wherein the data preprocessing includes at least one of sorting the historical premium income data in chronological order, performing logarithmic conversion on the historical premium income data, performing smoothing processing on the historical premium income data, and performing normalization processing on the smoothed historical premium income data;

[0216] The data extraction module 320 is used to:

[0217] The trend component and seasonal component in the historical premium income data after the data preprocessing are extracted by variational mode decomposition method.

[0218] In a possible implementation, the data extraction module 320 is further configured to:

[0219] Decomposing the historical premium income data into multiple modal components by using a variational modal decomposition method;

[0220] Among the multiple modal components, a trend component and a seasonal component are extracted; wherein the trend component is used to indicate the long-term trend of the premium income data, and the seasonal component is used to indicate the periodic changes of the premium income data.

[0221] In a possible implementation, the premium prediction module 330 is used to:

[0222] Using the first autoregressive integrated moving average model to make a prediction based on the trend component to obtain a trend change value;

[0223] The second autoregressive integrated moving average model performs prediction based on the seasonal component to obtain a seasonal fluctuation value;

[0224] Based on the trend change value and the seasonal fluctuation value, the predicted premium income is determined.

[0225] In a possible implementation manner, the premium prediction module 330 is further used for:

[0226] Performing a stationarity check on the trend component;

[0227] When the stability check of the trend component passes, prediction is performed based on the trend component to obtain a trend change value;

[0228] In the case that the stability check of the trend component fails, performing differential processing on the trend component to obtain a trend component after differential processing;

[0229] Prediction is performed based on the trend component after the differential processing to obtain a trend change value.

[0230] In a possible implementation manner, the traffic prediction module 340 is used to:

[0231] Get APP access traffic data;

[0232] Calculating the correlation coefficient between the APP access traffic data and the predicted premium income;

[0233] A linear regression model is used to predict the business load flow of the APP based on the correlation coefficient between the APP access flow data and the predicted premium income.

[0234] The specific implementation method and technical effect of the device provided in the embodiment of the present disclosure are similar to those of the above-mentioned method embodiment, and will not be repeated here.

[0235] According to an embodiment of the present disclosure, the present disclosure also discloses an electronic device, a computer-readable storage medium, and a computer program product.

[0236] Figure 4A schematic block diagram of an example electronic device 400 that can be used to implement an embodiment of the present disclosure is shown. The electronic device 400 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0237] like Figure 4 As shown, the electronic device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 402 or a computer program loaded from a storage unit 408 into a random access memory (RAM) 403. In the RAM 403, various programs and data required for the operation of the device 400 can also be stored. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0238] Multiple components in the electronic device 400 are connected to the I / O interface 405, including: an input unit 406, such as a keyboard, a mouse, etc.; an output unit 407, such as various types of displays, speakers, etc.; a storage unit 408, such as a disk, an optical disk, etc.; and a communication unit 409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 409 allows the electronic device 400 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0239] The computing unit 401 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 401 performs the various methods and processes described above, such as a business load flow prediction method. For example, in some embodiments, the business load flow prediction method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 400 via the ROM 402 and / or the communication unit 409. When the computer program is loaded into the RAM 403 and executed by the computing unit 401, one or more steps of the business load flow prediction method described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to perform the business load flow prediction method in any other appropriate manner (e.g., by means of firmware).

[0240] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0241] The program code of the computer program product for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0242] In the context of the present disclosure, a computer-readable storage medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may be a machine-readable signal medium or a machine-readable storage medium. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a computer-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0243] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0244] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), the Internet, and a blockchain network.

[0245] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server may also be a server of a distributed system, or a server combined with a blockchain.

[0246] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0247] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for predicting business load flow, characterized in that: include: Obtain historical premium income data within the target period; extracting trend components and seasonal components from the historical premium income data by using a variational mode decomposition method; determining predicted premium income based on the trend component and the seasonal component using an autoregressive integrated moving average model; The business load flow of the APP is predicted based on the predicted premium income.

2. The method for predicting business load flow according to claim 1, characterized in that: Before extracting the trend component and seasonal component in the historical premium income data by the variational mode decomposition method, the method further includes: Performing data preprocessing on the historical premium income data to obtain the historical premium income data after data preprocessing; wherein the data preprocessing includes at least one of sorting the historical premium income data in chronological order, performing logarithmic transformation on the historical premium income data, performing smoothing processing on the historical premium income data, and performing normalization processing on the smoothed historical premium income data; The method of extracting trend components and seasonal components from the historical premium income data by using a variational mode decomposition method includes: The trend component and seasonal component in the historical premium income data after the data preprocessing are extracted by variational mode decomposition method.

3. The method for predicting business load flow according to claim 1, characterized in that: The method of extracting trend components and seasonal components from the historical premium income data by using a variational mode decomposition method includes: Decomposing the historical premium income data into multiple modal components by using a variational modal decomposition method; Among the multiple modal components, a trend component and a seasonal component are extracted; wherein the trend component is used to indicate the long-term trend of the premium income data, and the seasonal component is used to indicate the periodic changes of the premium income data.

4. The method for predicting business load flow according to claim 3, characterized in that: The method of using the autoregressive integrated moving average model to determine the predicted premium income based on the trend component and the seasonal component includes: Using the first autoregressive integrated moving average model to make a prediction based on the trend component to obtain a trend change value; The second autoregressive integrated moving average model performs prediction based on the seasonal component to obtain a seasonal fluctuation value; Based on the trend change value and the seasonal fluctuation value, the predicted premium income is determined.

5. The method for predicting business load flow according to claim 4, characterized in that: The method of using the first autoregressive integrated moving average model to predict based on the trend component to obtain a trend change value includes: Performing a stationarity check on the trend component; When the stability check of the trend component passes, prediction is performed based on the trend component to obtain a trend change value; In the case that the stability check of the trend component fails, performing differential processing on the trend component to obtain a trend component after differential processing; Prediction is performed based on the trend component after the differential processing to obtain a trend change value.

6. The method for predicting business load flow according to claim 4, characterized in that: The business load flow of the APP based on the predicted premium income prediction includes: Get APP access traffic data; Calculating the correlation coefficient between the APP access traffic data and the predicted premium income; A linear regression model is used to predict the business load flow of the APP based on the correlation coefficient between the APP access flow data and the predicted premium income.

7. A service load flow prediction device, characterized in that: include: A data acquisition module is used to obtain historical premium income data within a target period; A data extraction module, used for extracting trend components and seasonal components in the historical premium income data by using a variational mode decomposition method; A premium forecasting module, used for determining the forecast premium income based on the trend component and the seasonal component using an autoregressive integrated moving average model; The traffic prediction module is used to predict the business load traffic of the APP based on the predicted premium income.

8. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images