Cell performance index prediction method and device, electronic equipment and storage medium

Abstract

The invention provides a cell performance index prediction method and device, electronic equipment and a storage medium, and relates to the technical field of communication. The method comprises the following steps: acquiring performance index data of a target cell in a first time period; performing multivariate decomposition on the performance index data in the first time period to obtain a plurality of index component data corresponding to the time domain features; inputting the multiple pieces of index component data into respective corresponding pre-trained component data prediction models, and outputting predicted values corresponding to the multiple pieces of index component data; and obtaining a target predicted value according to the predicted values corresponding to the multiple pieces of index component data. According to the invention, the accuracy and reliability of cell performance index prediction can be improved.

Description

Background Art

[0002] In mobile communication technology, cell performance indicators are key parameters for measuring network quality. With the rapid development of communication services, operators need to estimate cell performance indicators in order to optimize network planning and improve user experience.

[0003] In the related art, cell performance indicators are usually predicted based on traditional linear models. Since cell performance indicators are time series data and will change significantly over time, it is difficult to accurately estimate cell performance indicators based on traditional linear models.

[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0005] The present disclosure provides a method, device, electronic device and storage medium for predicting a cell performance index, which at least to a certain extent overcome the problem in the related art that it is difficult to accurately predict the target cell performance index.

[0006] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0007] According to one aspect of the present disclosure, a method for predicting cell performance indicators is provided, the method comprising: obtaining performance indicator data of a target cell within a first time period; performing multivariate decomposition on the performance indicator data within the first time period to obtain a plurality of indicator component data corresponding to time domain features; inputting the plurality of indicator component data into respective corresponding pre-trained component data prediction models, and outputting prediction values ​​corresponding to the plurality of indicator component data; and obtaining a target prediction value according to the prediction values ​​corresponding to the plurality of indicator component data.

[0008] In some embodiments, the performance indicator data within the first time period is subjected to multivariate decomposition to obtain multiple indicator component data corresponding to time domain characteristics, including at least one of the following: performing moving average processing on the performance indicator data within the first time period to obtain mean component data; performing trend fitting processing on the performance indicator data within the first time period to obtain trend component data; grouping the performance indicator data within the first time period according to a preset time period, and calculating the average value of each group, and mapping the average value of each group to the corresponding time point within the first time period to obtain seasonal component data; based on holiday component coefficients corresponding to different holidays, determining holiday component data in the performance indicator data within the first time period; wherein the holiday component coefficients corresponding to different holidays are obtained by a holiday component coefficient prediction model, and the holiday component coefficient prediction model is a model pre-trained through machine learning for predicting holiday component coefficients corresponding to different holidays; removing the mean component data, the trend variable, the seasonal component data and the holiday component data from the performance indicator data within the first time period to obtain residual component data.

[0009] In some embodiments, before determining the holiday component data in the performance indicator data within the first time period based on the holiday component coefficient corresponding to different holidays, the method further includes: acquiring the performance indicator data of the target cell within the second time period, and marking the holiday information corresponding to each data point, to obtain a data set for model training; extracting the cell characteristics of the target cell and the holiday characteristics corresponding to each holiday from the data set used for model training; wherein the cell characteristics include location characteristics, and the holiday characteristics include date characteristics, type characteristics, and duration characteristics; inputting the data set used for model training, the cell characteristics, and the holiday characteristics into a machine learning model for training, and when the training is completed, obtaining a holiday component coefficient prediction model.

[0010] In some embodiments, the holiday component data is determined in the performance indicator data within the first time period based on the holiday component coefficients corresponding to different holidays, including: inputting the performance indicator data within the first time period into the holiday component coefficient prediction model, and outputting the holiday component coefficients corresponding to different holidays; and decomposing the holiday component data from the performance indicator data within the first time period according to the holiday component coefficients corresponding to different holidays and the performance indicator data within the first time period.

[0011] In some embodiments, the inputting of the multiple indicator component data into their respective corresponding pre-trained component data prediction models and outputting the predicted values ​​corresponding to the multiple indicator component data include at least one of the following: inputting the mean component data into a pre-trained mean model and outputting the predicted value of the mean component data; inputting the trend component data into a pre-trained trend model and outputting the predicted value of the trend component data; inputting the season component data into a pre-trained season model and outputting the predicted value of the season component data; inputting the holiday component data into a pre-trained holiday model and outputting the predicted value of the holiday component data; inputting the residual component data into a pre-trained residual estimation model and outputting the predicted value of the residual component data.

[0012] In some embodiments, after obtaining the performance indicator data of the target cell within a first time period, it also includes: preprocessing the performance indicator data within the first time period to obtain the preprocessed performance indicator data within the first time period, the preprocessing including one or more of sorting according to time information, supplementing missing values, removing outliers and smoothing; performing multivariate decomposition on the performance indicator data within the first time period to obtain multiple indicator component data corresponding to time domain characteristics, including: performing multivariate decomposition on the preprocessed performance indicator data within the first time period to obtain multiple indicator component data corresponding to the preprocessed time domain characteristics.

[0013] In some embodiments, obtaining a target prediction value based on the prediction values ​​corresponding to the multiple indicator component data includes: obtaining weight coefficients corresponding to each of the multiple indicator component data; and performing weighted summation of the prediction values ​​corresponding to the multiple indicator component data based on the weight coefficients corresponding to each of the multiple indicator component data to obtain a target prediction value.

[0014] According to another aspect of the present disclosure, a cell performance indicator prediction device is also provided, including: a cell performance indicator data acquisition module, used to obtain the performance indicator data of the target cell within a first time period; a decomposition module, used to perform multivariate decomposition on the performance indicator data within the first time period to obtain a plurality of indicator component data corresponding to time domain features; a prediction module, used to input the plurality of indicator component data into their respective corresponding pre-trained component data prediction models, and output prediction values ​​corresponding to the plurality of indicator component data; a target prediction value acquisition module, used to obtain a target prediction value based on the prediction values ​​corresponding to the plurality of indicator component data.

[0015] According to another aspect of the present disclosure, an electronic device is also provided, which includes: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute any one of the cell performance indicator prediction methods described above by executing the executable instructions.

[0016] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the cell performance indicator prediction method described in any one of the above is implemented.

[0017] According to another aspect of the present disclosure, a computer program product is provided, including: a computer program or instructions, wherein when the computer program or instructions are executed by a processor, any one of the above-mentioned methods for predicting cell performance indicators is implemented.

[0018] The cell performance index prediction method, device, electronic device and storage medium provided in the embodiments of the present disclosure obtain the performance index data of the target cell in the first time period, perform multivariate decomposition on the performance index data in the first time period, and obtain multiple index component data corresponding to the time domain features; input the multiple index component data into the corresponding pre-trained component data prediction models, and output the prediction values ​​corresponding to the multiple index component data; obtain the target prediction value according to the prediction values ​​corresponding to the multiple index component data. The embodiments of the present disclosure improve the accuracy and reliability of the prediction of the cell performance index by introducing multivariate decomposition and component data prediction models corresponding to different component data.

[0019] 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

[0020] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0021] Figure 1 A specific flow chart of realizing cell performance indicator prediction in a prior art provided by an embodiment of the present disclosure is shown;

[0022] Figure 2 A schematic diagram of a cell performance indicator prediction system architecture in an embodiment of the present disclosure is shown;

[0023] Figure 3 A flow chart of a method for predicting cell performance indicators in an embodiment of the present disclosure is shown;

[0024] Figure 4 A flow chart of a method for predicting cell performance indicators in an embodiment of the present disclosure is shown;

[0025] Figure 5 A flow chart of a holiday component coefficient prediction model training method in an embodiment of the present disclosure is shown;

[0026] Figure 6 A flow chart of a method for acquiring holiday component data in an embodiment of the present disclosure is shown;

[0027] Figure 7 A flow chart of a method for obtaining predicted values ​​of multiple indicator component data in an embodiment of the present disclosure is shown;

[0028] Figure 8 A flow chart of a method for obtaining a target prediction value in an embodiment of the present disclosure is shown;

[0029] Fig. 9 A flow chart of a method for predicting cell performance indicators in an embodiment of the present disclosure is shown;

[0030] Fig.10 A specific flow chart for realizing cell performance indicator prediction provided by an embodiment of the present disclosure;

[0031] Fig.11 A schematic diagram of the structure of a cell performance indicator prediction device in an embodiment of the present disclosure is shown;

[0032] Fig.12 A schematic diagram of the structure of an electronic device for predicting cell performance indicators in an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0033] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the disclosure will be more comprehensive and complete and to fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0034] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and thus their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0035] Figure 1 A specific flow chart of implementing cell performance indicator prediction in the prior art provided by an embodiment of the present disclosure is shown. Figure 1 As shown, the cell performance indicator prediction method provided in the prior art of the embodiment of the present disclosure may include:

[0036] In the prediction stage, the historical data of the cell performance index is input into the trained linear regression model, and the trained linear regression model will predict and output the predicted value of the cell performance index based on the input historical data of the cell performance index.

[0037] In the training phase, the historical data of the performance index to be predicted and the real performance index data to be predicted are obtained, and the historical data of the performance index to be predicted are input into the linear regression model to obtain the predicted value. The error is calculated by comparing the predicted value output by the linear regression model with the real performance index data. The parameters of the linear regression model are adjusted according to the error to obtain the linear regression model for predicting the cell performance index.

[0038] The linear regression model makes predictions based on the linear relationship between variables. However, cell performance indicators change significantly over time and have obvious nonlinear characteristics. Therefore, the linear regression model cannot accurately capture these changes, resulting in inaccurate prediction results.

[0039] For ease of understanding, before introducing the embodiments of the present disclosure, several terms involved in the embodiments of the present disclosure are first explained as follows:

[0040] Cell: In mobile communication networks, refers to the geographical area covered by one or more base stations. Each base station communicates with user devices (such as mobile phones, tablets, etc.) within its coverage area through wireless signals. These base stations and their respective cells together constitute the entire mobile communication network, allowing users to stay connected when moving between different locations.

[0041] ARIMA: The full name of ARIMA is Autoregressive Integrated Moving Average, which is a statistical model widely used in time series forecasting. It combines three different elements: Autoregressive (AR), Integrated (I), and Moving Average (MA) to capture trends, seasonality, and random fluctuations in time series data.

[0042] The specific implementation of the embodiment of the present disclosure is described in detail below with reference to the accompanying drawings.

[0043] Figure 2FIG. 2 shows an exemplary system architecture diagram to which the cell performance indicator prediction method in the embodiment of the present disclosure can be applied. Figure 2 As shown, the system architecture includes a terminal device 101 , a network 102 and a network side device 103 .

[0044] The network 102 is a medium for providing a communication link between the terminal device 101 and the network-side device 103 , and may be a wired network or a wireless network.

[0045] Optionally, the wireless network or wired network described above uses standard communication technology and / or protocol. The network is usually the Internet, but it can also be any network, including but not limited to a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a mobile, wired or wireless network, a dedicated network or any combination of a virtual private network). In some embodiments, the data exchanged through the network is represented by technologies and / or formats including Hyper Text Mark-up Language (HTML), Extensible Markup Language (XML), etc. In addition, conventional encryption technologies such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), Internet Protocol Security (IPSec) can also be used to encrypt all or some links. In other embodiments, customized and / or dedicated data communication technologies can also be used to replace or supplement the above data communication technologies.

[0046] Optionally, the terminal device in the embodiment of the present disclosure may also be referred to as UE (User Equipment). In a specific implementation, the terminal device may be a mobile phone, a tablet computer (Tablet Personal Computer), a laptop computer (Laptop Computer), a personal digital assistant (Personal Digital Assistant, PDA), a mobile Internet device (Mobile Internet Device, MID), a wearable device (Wearable Device) or a vehicle-mounted device, etc. It should be noted that the specific type of the terminal device is not limited in the embodiment of the present invention.

[0047] The network side device may be a base station, a relay or an access point, etc. The base station may be a base station of 5G or later versions (e.g., 5G NR NB), or a base station in other communication systems (e.g., an eNB base station). It should be noted that the specific type of the network side device is not limited in the embodiments of the present disclosure.

[0048] Those skilled in the art will know that Figure 2 The number of terminal devices, networks, and network-side devices in the embodiment is only illustrative, and any number of terminals, networks, and network-side devices may be provided according to actual needs. The embodiments of the present disclosure are not limited to this.

[0049] Under the above system architecture, a method for predicting cell performance indicators is provided in an embodiment of the present disclosure. The method can be executed by any electronic device with computing and processing capabilities.

[0050] In some embodiments, the cell performance indicator prediction method provided in the embodiments of the present disclosure can be executed by the network side device in the above-mentioned system architecture; in other embodiments, the cell performance indicator prediction method provided in the embodiments of the present disclosure can be implemented by the terminal device and the network side device in the above-mentioned system architecture through interaction.

[0051] Figure 3 A flow chart of a method for predicting cell performance indicators in an embodiment of the present disclosure is shown as follows: Figure 3 As shown, the cell performance indicator prediction method provided in the embodiment of the present disclosure includes the following steps:

[0052] S302: Obtain performance indicator data of a target cell in a first time period.

[0053] In this embodiment, the target cell refers to the cell whose performance index is to be predicted. The first time period can be set according to actual needs, such as one week, one month, one year or a specific time period. Performance index data refers to various data indicators that measure the network quality in the target cell. Common performance index data may include network throughput, number of user connections, latency and packet loss rate, etc.

[0054] When collecting the performance indicator data of the target cell in the first time period, different time granularities may be set according to actual needs, such as hour granularity or minute granularity, which is not limited in the embodiments of the present disclosure.

[0055] S304, performing multivariate decomposition on the performance indicator data in the first time period to obtain a plurality of indicator component data corresponding to the time domain characteristics.

[0056] In this embodiment, the time domain feature refers to the characteristics of the performance indicator data in the time domain, which can be used to describe the law and characteristics of the performance indicator data changing over time. The multiple indicator component data corresponding to the time domain feature may include one or more of mean component data, trend component data, seasonal component data, holiday component data, and residual component data.

[0057] Among them, the mean component data represents the average state of the performance indicator data; the trend component data represents the long-term change direction of the performance indicator data, such as increase or decrease; the seasonal component data represents the periodic fluctuation of the performance indicator data due to periodic changes over a period of time; the holiday component data represents the impact of specific holidays on the performance indicator data; the residual component data is the remaining part after removing the above component data, which usually contains random errors and other unconsidered factors.

[0058] Specifically, the performance indicator data within the first time period is subjected to multivariate decomposition, that is, according to the characteristics of multiple indicator components, the performance indicator data within the first time period is decomposed into time series data corresponding to mean component data, trend component data, seasonal component data, holiday component data and residual component data.

[0059] S306, inputting the plurality of indicator component data into the corresponding pre-trained component data prediction models respectively, and outputting the prediction values ​​corresponding to the plurality of indicator component data.

[0060] In this embodiment, the component data prediction model refers to a prediction model established for each index component data. These component data prediction models use information in historical cell performance index data to learn the change rules of each component data and are used to predict future component data values.

[0061] Specifically, multiple indicator component data (such as mean component data, trend component data, seasonal component data, holiday component data, etc.) are respectively input into their corresponding pre-trained component data prediction models. Each model is optimized and trained for its corresponding component data to accurately capture the changing rules of the component data.

[0062] In this embodiment, the performance indicator data is decomposed into multiple component data (such as mean component data, trend component data, seasonal component data, holiday component data and residual component data), and each component data is predicted separately to improve the accuracy and reliability of the overall prediction.

[0063] S308, obtaining a target prediction value according to the prediction values ​​corresponding to the plurality of indicator component data.

[0064] In this embodiment, in order to obtain the target prediction value, the prediction results of each component data need to be combined. The specific combination method depends on the relationship between the component data and how they work together on the performance indicator data. The following are several possible combination strategies provided by this embodiment:

[0065] Additive model: Assuming that the data of each component are linearly added, it is suitable for the situation that the changes of each component data are relatively independent and the amplitude is similar. The formula is as follows:

[0066] Target prediction value = prediction value of mean component data + prediction value of trend component data + prediction value of seasonal component data + prediction value of holiday component data + prediction value of residual component data

[0067] Multiplication model: Assuming that the component data are in a multiplication relationship, it is suitable for situations where the component data vary greatly and affect each other, such as when the seasonal effect increases over time. The formula is as follows:

[0068] Target prediction value = prediction value of mean component data × (1 + prediction value of trend component data) × (1 + prediction value of season component data) × (1 + prediction value of residual component data) × (1 + prediction value of holiday component data)

[0069] Mixed model: Combining the advantages of additive and multiplicative models, the relationship between component data can be flexibly adjusted according to actual conditions. For example, the additive form is used for some component data, while the multiplicative form is used for other component data.

[0070] In summary, in the embodiments of the present disclosure, multivariate decomposition is used to obtain multiple indicator component data. For different indicator component data, corresponding prediction models are selected according to their characteristics, which can effectively improve the prediction accuracy.

[0071] In one embodiment, there are multiple possible ways to obtain the indicator component data. Figure 4 A flowchart of a method for predicting cell performance indicators provided by an embodiment of the present disclosure, combined with Figure 4 As shown, multivariate decomposition is performed on the performance indicator data in the first time period based on the time domain features to obtain multiple indicator component data corresponding to the time domain features, including at least one of the following:

[0072] S402, performing moving average processing on the performance indicator data in the first time period to obtain mean component data.

[0073] In this embodiment, moving average processing is a time series smoothing technology that reduces random fluctuations in data by calculating the average value of data points within a certain range around a certain data point, thereby obtaining mean component data of performance indicator data.

[0074] S404, performing trend fitting processing on the performance indicator data in the first time period to obtain trend component data.

[0075] In this embodiment, the trend fitting process is to describe the long-term change trend of the performance indicator data in the first time period by fitting a straight line or a curve. The performance indicator data may be trend fitted using linear regression, polynomial regression, exponential regression, and other methods.

[0076] S406, grouping the performance indicator data in the first time period according to a preset time period, calculating the average value of each group, and mapping the average value of each group to a corresponding time point in the first time period to obtain seasonal component data.

[0077] In this embodiment, the time series data is grouped according to a preset time period (such as a week, month, quarter, etc.), and the average value or median of each group of data is calculated to obtain the average level of the performance indicator within the time period. These average values ​​are mapped back to the original time axis according to the time period to which they belong to obtain seasonal component data. Seasonal component data reflects the periodic changes of data in different time periods. For example, suppose there is a year of network traffic data, grouped by month, and the average traffic per month is calculated, where the average traffic in January is A, February is B, March is C... December is L. The average values ​​of each group are placed on the original time axis, with January corresponding to A, February corresponding to B, March corresponding to C... December corresponding to L. In this way, seasonal component data is obtained. Seasonal component data reflects the average level of network traffic in each month, thereby showing the periodic change characteristics of network traffic in different months.

[0078] S408: Determine holiday component data from the performance indicator data in the first time period based on holiday component coefficients corresponding to different holidays.

[0079] In this embodiment, the holiday component coefficient is used to measure the degree of difference between the performance indicator data on holidays and non-holidays. The holiday component coefficient reflects whether the performance indicator increases or decreases during a specific holiday compared to a normal day (non-holiday), and how much the increase or decrease is, so the holiday component coefficient can also be represented by the increase or decrease coefficients of different holidays.

[0080] The holiday component coefficients corresponding to different holidays can be obtained through a holiday component coefficient prediction model. The holiday component coefficient prediction model is a model obtained in advance through machine learning training and used to predict the holiday component coefficients corresponding to different holidays.

[0081] Specifically, based on historical cell performance index data and holiday types, a holiday component coefficient prediction model is trained through machine learning. The trained holiday component coefficient prediction model is used to predict holiday component coefficients of different holidays. According to the predicted holiday component coefficient and performance index data, the holiday component data is decomposed from the performance index data in the first time period.

[0082] In an optional embodiment, since the holiday component data has strong uncertainty and different cells are affected differently by holidays, in order to increase the accuracy of the decomposition of the holiday component data, the holiday component coefficients of different holidays can be predicted by a machine learning algorithm. It should be noted that the holiday component coefficient is modeled as a proportional increase or decrease in the value of the non-holiday time index. Using a machine learning algorithm, the increase or decrease coefficients of different holidays can be learned based on cell characteristics and historical data.

[0083] S410, removing mean component data, trend variables, seasonal component data and holiday component data from the performance indicator data in the first time period to obtain residual component data.

[0084] In this embodiment, the residual component data is the remaining part after removing the mean component data, trend component data, seasonal component data and holiday component data from the original time series data. It reflects the part of the data that cannot be explained by the above component data, usually including random errors, measurement errors, etc.

[0085] Specifically, the mean component data, the trend component data, the seasonal component data and the holiday component data are subtracted from the performance indicator data to obtain the residual component data.

[0086] For example, assuming that the performance indicator data is [100, 120, 110, 130, 140, 150, 160], the mean component data, trend component data, seasonal component data, and holiday component data obtained after the above steps are [110, 120, 120, 130, 140, 153.33, 160] respectively (assuming that the trend component data is 0, the seasonal component data and the holiday component data are also 0 to simplify the example), then the residual component data is calculated as follows:

[0087] Residual component data = performance index data - mean component data = [100-110, 120-120, 110-120, 130-130, 140-140, 150-153.33, 160-160] = [-10, 0, -10, 0, 0, -3.33, 0].

[0088] In this embodiment, by separating multiple indicator component data corresponding to the time domain features, future behaviors can be modeled and predicted more accurately.

[0089] In one embodiment, Figure 5 As shown, before determining the holiday component data in the performance indicator data within the first time period based on the holiday component coefficients corresponding to different holidays, the cell performance indicator prediction method provided in the embodiment of the present disclosure can also train the holiday component coefficient prediction model through the following steps:

[0090] S502, obtaining performance indicator data of the target cell in a second time period, and marking the holiday information corresponding to each data point to obtain a data set for model training.

[0091] In this embodiment, the second time period is a time period including multiple holidays and non-holidays; the performance indicator data in the second time period is the performance indicator data in a certain period including holidays and non-holidays. By acquiring the performance indicator data (such as user traffic, average delay time, etc.) in the second time period of the target cell, marking each data point, and recording its corresponding holiday information (such as whether it is a holiday, the specific name of the holiday, etc.), a data set for model training is generated.

[0092] S504: extracting the cell features of the target cell and the holiday features corresponding to each holiday from the data set used for model training.

[0093] In the data set used for model training, the cell features of the target cell are extracted. The cell features include but are not limited to:

[0094] Location characteristics: refers to the specific geographical location of the target cell, such as the city center, suburbs, etc.; user characteristics of the target cell can also be extracted, such as the number of user connections in the target cell, PDCP (Packet Data Convergence Protocol) traffic, etc. Among them, PDCP traffic is related to the user's data usage behavior (such as video streaming, downloading files, etc.), which can reflect the activeness of data transmission in the target cell.

[0095] In the data set used for model training, the cell features of the target cell and the holiday features corresponding to each holiday are extracted; wherein the cell features include location features, and the holiday features include but are not limited to:

[0096] Date features: specific date, day of the week, etc.; Type features: holiday type (such as statutory holidays, traditional holidays, local holidays, etc.); Duration features: the length of holidays (such as single-day, multi-day consecutive holidays, etc.). By introducing multiple features, the model's ability to understand different types of holidays is enhanced, which improves the accuracy of prediction.

[0097] S506, inputting the data set, community characteristics and holiday characteristics used for model training into the machine learning model for training, and when the training is completed, obtaining a holiday component coefficient prediction model.

[0098] In this embodiment, a machine learning method is used to predict the increase and decrease coefficients of different holidays (i.e., holiday component coefficients) based on the data set, cell characteristics, and holiday characteristics used for model training. When the model converges and the training is completed, a holiday component coefficient prediction model is obtained. It should be noted that the holiday component coefficient prediction model training process supports the selection of multiple machine learning algorithms (such as random forests, gradient boosting trees, neural networks, etc.), and a suitable model can be selected according to actual needs, and this embodiment does not limit this.

[0099] In this embodiment, the influence of different holidays on performance indicators is automatically learned through machine learning algorithms, reducing the need for manual setting of rules. The model can also be adaptively adjusted according to the specific characteristics of the community, improving the flexibility and accuracy of the prediction.

[0100] In one embodiment, Figure 6 A flow chart of a method for obtaining holiday component data provided by an embodiment of the present disclosure, combined with Figure 6 As shown, based on the holiday component coefficients corresponding to different holidays, determining the holiday component data in the performance indicator data in the first time period may include:

[0101] S602: Input the performance indicator data in the first time period into the holiday component coefficient prediction model, and output the holiday component coefficients corresponding to different holidays.

[0102] The performance indicator data in the first time period is input into the pre-trained holiday component coefficient prediction model. The model outputs the holiday component coefficients corresponding to different holidays based on the input performance indicator data and the extracted cell characteristics and holiday characteristics.

[0103] S604: Decompose the holiday component data from the performance indicator data in the first time period according to the holiday component coefficients corresponding to different holidays and the performance indicator data in the first time period.

[0104] According to the holiday component coefficients corresponding to different holidays and the performance indicator data in the first time period, the specific value of each holiday component data in the time period is calculated.

[0105] The holiday component data can be calculated using the following formula:

[0106] Holiday component data = performance index data × holiday component coefficient;

[0107] Or, holiday component data = performance index data × (holiday component coefficient - 1).

[0108] For multiplicative seasonal changes, the holiday component coefficient represents the increase or decrease ratio relative to the normal mean; for additive seasonal changes, a fixed increment is directly added. In this way, the holiday component data can be separated from the performance indicator data to obtain the holiday component data.

[0109] In this embodiment, by introducing the holiday component coefficient, the accuracy of extracting the holiday component data can be improved.

[0110] In one embodiment, Figure 7 A flowchart for obtaining predicted values ​​of multiple index component data provided by the embodiment of the present disclosure, combined with Figure 7 As shown, multiple indicator component data are respectively input into the corresponding pre-trained component data prediction models, and the prediction values ​​corresponding to the multiple indicator component data are output, including at least one of the following:

[0111] S702, input the mean component data into a pre-trained mean model, and output the predicted value of the mean component data.

[0112] In time series analysis, since the mean component data represents the average value of the performance indicator data, the predicted value of the mean component data can be predicted by calculating the average value of the mean component data. The mean model is a prediction model built based on the mean algorithm, which aims to use the average mean in historical data to predict future values.

[0113] In this embodiment, the mean model sums the mean component data and then divides it by the number of time points to obtain an overall average value as the predicted value of the mean component data.

[0114] S704, input the trend component data into a pre-trained trend model, and output the predicted value of the trend component data.

[0115] In time series analysis, trend component data reflects the change pattern of data gradually increasing or decreasing over time. Since trend component data fluctuates relatively slowly and has simple characteristics, the ARIMA algorithm can be used to build a trend model to predict the predicted value of trend component data.

[0116] In this embodiment, the pre-trained trend model can be implemented in the following way:

[0117] Separate the historical trend component data from the historical cell performance index data, use the extracted historical trend component data as training data, and learn the parameters of the ARIMA model. Specifically, perform a stationarity test on the historical trend component data, perform d-order differences on the non-stationary historical trend component data, and convert them into stationary historical trend component data. Use the autocorrelation function and partial autocorrelation function to calculate the stationary historical trend component data to obtain p and q, where p represents the order of the autoregressive term, d represents the order of the difference, and q represents the order of the moving average. From the above p, d, and q values, the ARIMA model is obtained, and the ARIMA model is tested to ensure that the model converges, that is, reaches a stable state, and no longer significantly improves the prediction error, and a trend model is obtained.

[0118] The trend component data decomposed from the performance indicator data in the first time period is input into a pre-trained trend model, and the trend model outputs a predicted value of the trend component data based on the input trend component data.

[0119] In this embodiment, considering that the trend component data fluctuates relatively slowly and has simple characteristics, the ARIMA algorithm is used to predict the trend component data, thereby achieving effective prediction of the trend component data and improving the reliability of the prediction of the cell performance index.

[0120] S706, input the seasonal component data into a pre-trained seasonal model, and output the predicted value of the seasonal component data.

[0121] In time series analysis, seasonal component data reflects the periodic characteristics of the data. Since seasonal component data is periodic data, the HW (Holt-Winters) algorithm is a method suitable for predicting periodic data.

[0122] In this embodiment, according to the periodic characteristics of seasonal component data, HW algorithm variants, such as additive or multiplicative, are selected to construct the HW algorithm model. The additive HW algorithm is suitable for situations where seasonal changes are relatively stable, and the multiplicative HW algorithm is suitable for situations where seasonal changes increase with the growth of mean component data.

[0123] Considering that different cells may have different characteristics (such as user behavior patterns, geographic locations, etc.), additional features can be introduced as input to further optimize the model performance. Use historical seasonal component data as training data to learn the parameters of the HW algorithm model. During the model training process, ensure that the model converges, that is, reaches a stable state, and no longer significantly improves the prediction error, and obtains a seasonal model.

[0124] The extracted seasonal component data is input into the pre-trained seasonal model. The seasonal model outputs the predicted value of the seasonal component data based on the input seasonal component data.

[0125] S708, inputting the holiday component data into a pre-trained holiday model, and outputting a predicted value of the holiday component data.

[0126] In the time series prediction algorithm, the Prophet algorithm can be used to predict the holiday component. In this embodiment, a holiday model is obtained based on the Prophet algorithm training. The holiday model captures the special impact of holidays on data by introducing holiday variables into the model.

[0127] In this embodiment, in combination with S602-504, a machine learning algorithm is used to obtain holiday component data, a holiday model is obtained by training based on the Prophet algorithm, and the holiday component data is predicted using the trained holiday model to obtain a predicted value of the holiday component data.

[0128] In this embodiment, a machine learning algorithm and a Prophet algorithm are used to predict the impact of holidays on cell performance indicators, and the predicted values ​​of holiday component data of different types of cells are intelligently adjusted. Compared with traditional prediction methods, the prediction results provided by this embodiment are more accurate.

[0129] S710, inputting the residual component data into a pre-trained residual estimation model, and outputting a predicted value of the residual component data.

[0130] In time series analysis, since the residual component data characterizes the long-term evolution trend of cell characteristics, the ARIMA algorithm can be used to extract information from the residual component data to obtain the predicted value of the residual component data.

[0131] The historical residual component data is used as training data in advance to train the ARIMA model. The ARIMA model parameters corresponding to the residual estimation model are determined. During the model training process, the model is ensured to converge, that is, to reach a stable state, and no longer significantly improve the prediction error, and the residual estimation model is obtained.

[0132] In this embodiment, the extracted residual component data is input into a pre-trained residual estimation model, and the residual estimation model outputs the predicted value of the residual component data based on the input residual component data. The ARIMA model is used to capture the trend in the data that was originally regarded as "noise" from the residual component data, thereby improving the accuracy of the prediction.

[0133] In an alternative embodiment, Figure 8 A flow chart of a method for obtaining a target prediction value provided by an embodiment of the present disclosure, combined with Figure 8 As shown in the figure, the target prediction value is obtained according to the prediction values ​​corresponding to the multiple indicator component data, including:

[0134] S802, obtaining weight coefficients corresponding to each of the plurality of indicator component data.

[0135] In this embodiment, the multiple indicator component data may be one or more of mean component data, trend component data, seasonal component data, holiday component data and residual component data, and a weight coefficient is assigned to each of the mean component data, trend component data, seasonal component data, holiday component data and residual component data, and the weight coefficients corresponding to different indicator component data may be obtained based on the performance of different indicator component historical data, expert experience or through a machine learning algorithm. The weight coefficient reflects the relative importance of each indicator component data in the final prediction, ensuring that the model can better capture the impact of different indicator component data.

[0136] S804, based on the weight coefficients corresponding to the multiple indicator component data respectively, weighted sum is performed on the prediction values ​​corresponding to the multiple indicator component data to obtain the target prediction value.

[0137] In this embodiment, the weighted sum of the predicted values ​​corresponding to the multiple indicator component data can be calculated by the following formula:

[0138] Y=∑(i×y i )

[0139] Among them, Y represents the target prediction value, i represents the weight coefficient, and y i Represents the predicted value of indicator component data.

[0140] In this embodiment, by introducing the weight coefficient, the proportion of each indicator component data in the final prediction can be adjusted according to its actual impact on the performance indicator data, so that the prediction is closer to the actual situation and the accuracy and reliability of the prediction are improved.

[0141] In one embodiment, in order to further improve the accuracy of prediction, Fig. 9 A flowchart of a cell performance indicator prediction method provided by an embodiment of the present disclosure, combined with Fig. 9 As shown, after obtaining the performance indicator data of the target cell in the first time period, the cell performance indicator prediction method provided in the embodiment of the present disclosure may further include the following steps:

[0142] S902: Preprocess the performance indicator data within the first time period to obtain preprocessed performance indicator data within the first time period.

[0143] In this embodiment, the preprocessing includes one or more of sorting according to time series, supplementing missing values, removing outliers, and smoothing.

[0144] Among them, sorting by time series is to arrange in ascending or descending order according to the timestamps corresponding to the performance indicator data of the target cell in the first time period, in order to provide ordered time series data for subsequent steps and ensure the effectiveness of time series analysis.

[0145] Complementing missing values ​​refers to filling in missing data points in the time series to ensure data continuity. In this embodiment, forward filling (using the previous non-missing value), backward filling (using the next non-missing value), interpolation (such as linear interpolation), etc. can be used to estimate missing values, thereby reducing analysis bias caused by missing data and improving the accuracy of prediction results.

[0146] Removing outliers refers to removing data that deviates from the normal range in the performance indicator data. In this embodiment, the normal value range can be defined using data statistical methods (such as box plots, Z scores) or professional knowledge in this field. For values ​​of performance indicator data that exceed the normal value range, by deleting or replacing them with upper and lower limits, or replacing the data that deviates from the normal range in the performance indicator data with the average of adjacent values, the influence of outliers on the target prediction value is avoided, ensuring the reliability of the prediction result.

[0147] Smoothing refers to smoothing the fluctuations in performance indicator data through smoothing algorithms, removing short-term random changes, and making the trend more obvious. Smoothing algorithms can include moving average smoothing, exponential moving average, quantile smoothing, and log-normal distribution smoothing. Smoothing can enhance the trend characteristics of performance indicator data and reduce the impact of short-term fluctuations.

[0148] It should be noted that in the embodiments of the present disclosure, one or a combination of any multiple methods of the above preprocessing can be selected, and the present disclosure does not limit this.

[0149] S904, performing multivariate decomposition on the pre-processed performance indicator data in the first time period to obtain a plurality of indicator component data corresponding to the time domain features.

[0150] In this embodiment, the multivariate decomposition method for the performance indicator data in the first time period can be referred to as described above, and the performance indicator data in the first time period after preprocessing can be decomposed into multiple variables, which will not be described in detail in this disclosure. In this embodiment, by preprocessing the performance indicator data in the first time period, a high-quality, accurate and representative data basis is provided for subsequent analysis and modeling. This helps to improve the accuracy and reliability of the prediction results.

[0151] Fig.10 A specific implementation flow chart of a method for predicting cell performance indicators in an embodiment of the present disclosure is shown as follows: Fig.10 As shown, the cell performance indicator prediction method provided in the embodiment of the present disclosure may include:

[0152] In the prediction stage, the historical cell performance index data is decomposed in time series, i.e., multivariate decomposition, to obtain multiple index components corresponding to the time domain characteristics, including mean component, trend component, seasonal component, and holiday component. Among them, the mean component reflects the average level of the data; the trend component reveals the long-term increase or decrease trend of the data; the seasonal component shows periodic changes, such as quarterly or monthly fluctuations; and the holiday component reflects the impact of specific holidays on the data.

[0153] The mean component, trend component, seasonal component and holiday component are input into respective pre-trained models, and the pre-trained models include a level model (also called a mean model), a trend model, a seasonal model and a holiday model. The mean model, the trend model, the seasonal model and the holiday model output respective prediction values, and a comprehensive prediction value is obtained based on the respective prediction values. Considering the existence of random noise or other factors not considered in the historical cell performance index data, the residual component is obtained by removing the above-mentioned mean component, trend component, seasonal component and holiday component from the historical cell performance index data. The residual component is input into the residual estimation model to obtain the prediction value of the residual component, and finally the target prediction value is obtained according to the comprehensive prediction value and the prediction value of the residual component.

[0154] During the training phase of this embodiment, by acquiring historical data of the performance indicator to be predicted and real performance indicator data to be predicted, the historical data of the performance indicator to be predicted is subjected to time series multivariate decomposition to obtain the mean component, trend component, seasonal component and holiday component of the historical data of the performance indicator to be predicted.

[0155] A horizontal model is obtained by training based on the mean component of the historical data of the performance indicator to be predicted and the actual performance indicator data; a trend model is obtained by training based on the trend component of the historical data of the performance indicator to be predicted and the actual performance indicator data; a seasonal model is obtained by training based on the seasonal component of the historical data of the performance indicator to be predicted and the actual performance indicator data; a holiday model is obtained by training based on the holiday component of the historical data of the performance indicator to be predicted and the actual performance indicator data, and a residual estimation model is obtained by training based on the residual component of the historical data of the performance indicator to be predicted and the actual data.

[0156] In this way, each model can capture different characteristics of the data respectively, and finally improve the overall prediction accuracy of the prediction performance indicators by combining the prediction results of these models.

[0157] Based on the same inventive concept, a cell performance indicator prediction device is also provided in the embodiment of the present disclosure, as described in the following embodiment. Since the principle of solving the problem in the embodiment of the device is similar to that in the above method embodiment, the implementation of the embodiment of the device can refer to the implementation of the above method embodiment, and the repeated parts will not be repeated.

[0158] Fig.11 A schematic diagram of a cell performance indicator prediction device in an embodiment of the present disclosure is shown. Fig.11 As shown, the device includes: a cell performance indicator data acquisition module 111, a decomposition module 112, a prediction module 113 and a target prediction value acquisition module 114.

[0159] Among them, the cell performance indicator data acquisition module 111 is used to obtain the performance indicator data of the target cell within the first time period; the decomposition module 112 is used to perform multivariate decomposition on the performance indicator data within the first time period to obtain multiple indicator component data corresponding to the time domain characteristics; the prediction module 113 is used to input the multiple indicator component data into their respective corresponding pre-trained component data prediction models, and output the prediction values ​​corresponding to the multiple indicator component data; the target prediction value acquisition module 114 is used to obtain the target prediction value according to the prediction values ​​corresponding to the multiple indicator component data.

[0160] In some embodiments of the present disclosure, the decomposition module 112 is specifically used for at least one of the following: performing moving average processing on the performance indicator data within the first time period to obtain mean component data; performing trend fitting processing on the performance indicator data within the first time period to obtain trend component data; grouping the performance indicator data within the first time period according to a preset time period, and calculating the average value of each group, and mapping the average value of each group to the corresponding time point within the first time period to obtain seasonal component data; determining holiday component data in the performance indicator data within the first time period based on holiday component coefficients corresponding to different holidays; wherein the holiday component coefficients corresponding to different holidays are obtained through a holiday component coefficient prediction model, and the holiday component coefficient prediction model is a model obtained in advance through machine learning training for predicting holiday component coefficients corresponding to different holidays; removing the mean component data, trend variables, seasonal component data and holiday component data from the performance indicator data within the first time period to obtain residual component data.

[0161] In some embodiments of the present disclosure, the decomposition module 112 is further used to: obtain performance indicator data of the target cell within the second time period, and mark the holiday information corresponding to each data point to obtain a data set for model training; a feature extraction unit is used to extract the cell characteristics of the target cell and the holiday characteristics corresponding to each holiday in the data set used for model training; wherein the cell characteristics include location characteristics, and the holiday characteristics include date characteristics, type characteristics, and duration characteristics; a training unit is used to input the data set, cell characteristics, and holiday characteristics used for model training into a machine learning model for training, and when the training is completed, a holiday component coefficient prediction model is obtained.

[0162] In some embodiments of the present disclosure, the decomposition module 112 is specifically used to: input the performance indicator data within the first time period into the holiday component coefficient prediction model, and output the holiday component coefficients corresponding to different holidays; decompose the holiday component data from the performance indicator data within the first time period according to the holiday component coefficients corresponding to different holidays and the performance indicator data within the first time period.

[0163] In some embodiments of the present disclosure, the prediction module 113 is specifically used for at least one of the following: inputting mean component data into a pre-trained mean model, and outputting a predicted value of the mean component data; inputting trend component data into a pre-trained trend model, and outputting a predicted value of the trend component data; inputting season component data into a pre-trained season model, and outputting a predicted value of the season component data; inputting holiday component data into a pre-trained holiday model, and outputting a predicted value of the holiday component data; inputting residual component data into a pre-trained residual estimation model, and outputting a predicted value of the residual component data.

[0164] In some embodiments of the present disclosure, the cell performance indicator data acquisition module 111 is also used to: preprocess the performance indicator data within the first time period to obtain the performance indicator data within the first time period after preprocessing, and the preprocessing includes one or more of sorting according to time information, supplementing missing values, removing outliers and smoothing processing; the decomposition module 112 is specifically used to: perform multivariate decomposition on the performance indicator data within the first time period after preprocessing to obtain multiple indicator component data corresponding to the preprocessed time domain characteristics.

[0165] In some embodiments of the present disclosure, the target prediction value acquisition module 114 is specifically used to: obtain the weight coefficient corresponding to each of the multiple indicator component data; based on the weight coefficient corresponding to each of the multiple indicator component data, perform weighted summation on the prediction values ​​corresponding to the multiple indicator component data to obtain the target prediction value.

[0166] It should be noted that the examples and application scenarios implemented by the modules in the above-mentioned device embodiment are the same as those of the corresponding steps in the method embodiment, but are not limited to the contents disclosed in the above-mentioned method embodiment. It should be noted that the above-mentioned modules as part of the device can be executed in a computer system such as a set of computer executable instructions.

[0167] Those skilled in the art will appreciate that various aspects of the present disclosure may be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation that combines hardware and software aspects, which may be collectively referred to herein as a "circuit," "module," or "system."

[0168] Based on the same inventive concept, an electronic device is also provided in an embodiment of the present disclosure, the electronic device comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute any one of the above-mentioned methods for predicting cell performance indicators by executing the executable instructions. Since the principle of solving the problem in the electronic device embodiment is similar to that in the above-mentioned method embodiment, the implementation of the electronic device embodiment can refer to the implementation of the above-mentioned method embodiment, and the repeated parts will not be repeated.

[0169] Refer to the following Fig.12 1200 according to this embodiment of the present disclosure is described. Fig.12 The electronic device 1200 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0170] like Fig.12 As shown, the electronic device 1200 is in the form of a general computing device. The components of the electronic device 1200 may include but are not limited to: at least one processing unit 1210, at least one storage unit 1220, and a bus 1230 connecting different system components (including the storage unit 1220 and the processing unit 1210).

[0171] Among them, the storage unit stores program code, and the program code can be executed by the processing unit 1210, so that the processing unit 1210 executes the steps described in the above "Exemplary Method" section of this specification according to various exemplary embodiments of the present disclosure. For example, the processing unit 1210 can execute the following steps of the above method embodiment: obtaining performance indicator data of the target cell in the first time period; performing multivariate decomposition on the performance indicator data in the first time period to obtain multiple indicator component data corresponding to the time domain features; inputting the multiple indicator component data into the corresponding pre-trained component data prediction model, and outputting the prediction values ​​corresponding to the multiple indicator component data; obtaining the target prediction value according to the prediction values ​​corresponding to the multiple indicator component data.

[0172] The storage unit 1220 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 12201 and / or a cache storage unit 12202 , and may further include a read-only storage unit (ROM) 12203 .

[0173] The storage unit 1220 may also include a program / utility 12204 having a set (at least one) of program modules 12205, such program modules 12205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0174] Bus 1230 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0175] The electronic device 1200 may also communicate with one or more external devices 1240 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 1200, and / or communicate with any device that enables the electronic device 1200 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 1250. Furthermore, the electronic device 1200 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 1260. As shown, the network adapter 1260 communicates with other modules of the electronic device 1200 via a bus 1230. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 1200, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0176] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0177] Based on the same inventive concept, a computer-readable storage medium is also provided in an embodiment of the present disclosure, on which a computer program is stored, and when the computer program is executed by a processor, any of the above-mentioned methods for predicting cell performance indicators is implemented. Since the principle of solving the problem in the computer-readable storage medium embodiment is similar to that in the above-mentioned method embodiment, the implementation of the computer-readable storage medium embodiment can refer to the implementation of the above-mentioned method embodiment, and the repeated parts will not be repeated.

[0178] More specific examples of computer-readable storage media in the present disclosure may include, but are not limited to, an electrical connection having one or more conductors, 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), 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.

[0179] In the present disclosure, a computer readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, wherein a readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A readable signal medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0180] Alternatively, the program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0181] In a specific implementation, the program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0182] Based on the same inventive concept, a computer program product is also provided in an embodiment of the present disclosure, including a computer program product, including: a computer program or an instruction, which implements any one of the cell performance indicator prediction methods in the above method embodiments when the computer program or the instruction is executed by a processor. Since the principle of solving the problem in the computer program product embodiment is similar to that in the above method embodiment, the implementation of the computer program product embodiment can refer to the implementation of the above method embodiment, and the repeated parts will not be repeated.

[0183] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0184] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0185] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0186] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the appended claims.

Claims

1. A method for predicting cell performance indicators, characterized in that: include: Obtaining performance indicator data of the target cell within a first time period; Performing multivariate decomposition on the performance indicator data in the first time period to obtain a plurality of indicator component data corresponding to the time domain characteristics; Inputting the plurality of index component data into the corresponding pre-trained component data prediction models respectively, and outputting the prediction values ​​corresponding to the plurality of index component data; A target prediction value is obtained according to the prediction values ​​corresponding to the multiple indicator component data.

2. The method for predicting cell performance indicators according to claim 1, characterized in that: The performing multivariate decomposition on the performance indicator data in the first time period to obtain a plurality of indicator component data corresponding to the time domain characteristics includes at least one of the following: Performing moving average processing on the performance indicator data within the first time period to obtain mean component data; Performing trend fitting processing on the performance indicator data within the first time period to obtain trend component data; The performance indicator data in the first time period are grouped according to a preset time period, and an average value of each group is calculated, and the average value of each group is mapped to a corresponding time point in the first time period to obtain seasonal component data; Determining holiday component data in the performance indicator data within the first time period based on holiday component coefficients corresponding to different holidays; wherein the holiday component coefficients corresponding to different holidays are obtained through a holiday component coefficient prediction model, and the holiday component coefficient prediction model is a model obtained in advance through machine learning training for predicting holiday component coefficients corresponding to different holidays; The mean component data, the trend variable, the seasonal component data and the holiday component data are removed from the performance indicator data in the first time period to obtain residual component data.

3. The method for predicting cell performance indicators according to claim 2, characterized in that: Before determining the holiday component data in the performance indicator data within the first time period based on the holiday component coefficients corresponding to different holidays, the method further includes: Obtain performance indicator data of the target cell in the second time period, and mark the holiday information corresponding to each data point to obtain a data set for model training; In the data set used for model training, cell features of the target cell and holiday features corresponding to each holiday are extracted; wherein the cell features include location features, and the holiday features include date features, type features, and duration features; The data set for model training, the cell characteristics and the holiday characteristics are input into the machine learning model for training. When the training is completed, a holiday component coefficient prediction model is obtained.

4. The method for predicting cell performance indicators according to claim 3, characterized in that: The determining of the holiday component data in the performance indicator data in the first time period based on the holiday component coefficients corresponding to different holidays includes: Inputting the performance indicator data in the first time period into the holiday component coefficient prediction model, and outputting the holiday component coefficients corresponding to different holidays; According to the holiday component coefficients corresponding to different holidays and the performance indicator data in the first time period, the holiday component data is decomposed from the performance indicator data in the first time period.

5. The method for predicting cell performance indicators according to claim 2, characterized in that: The step of inputting the plurality of indicator component data into the corresponding pre-trained component data prediction models respectively, and outputting the prediction values ​​corresponding to the plurality of indicator component data, comprises at least one of the following: Inputting the mean component data into a pre-trained mean model, and outputting a predicted value of the mean component data; Inputting the trend component data into a pre-trained trend model, and outputting a predicted value of the trend component data; Inputting the seasonal component data into a pre-trained seasonal model, and outputting a predicted value of the seasonal component data; Inputting the holiday component data into a pre-trained holiday model, and outputting a predicted value of the holiday component data; The residual component data is input into a pre-trained residual estimation model, and a predicted value of the residual component data is output.

6. The method for predicting cell performance indicators according to claim 1, characterized in that: After acquiring the performance indicator data of the target cell in the first time period, the method further includes: Preprocessing the performance indicator data in the first time period to obtain preprocessed performance indicator data in the first time period, wherein the preprocessing includes one or more of sorting according to time series, supplementing missing values, removing outliers, and smoothing; Perform multivariate decomposition on the performance indicator data in the first time period to obtain multiple indicator component data corresponding to the time domain characteristics, including: Based on the time domain features, multivariate decomposition is performed on the performance indicator data in the first time period after the preprocessing to obtain a plurality of indicator component data corresponding to the time domain features after the preprocessing.

7. The method for predicting cell performance indicators according to claim 1, characterized in that: Obtaining a target prediction value according to the prediction values ​​corresponding to the plurality of indicator component data includes: Obtaining weight coefficients corresponding to each of the plurality of indicator component data; Based on the weight coefficients corresponding to the multiple indicator component data respectively, the predicted values ​​corresponding to the multiple indicator component data are weighted and summed to obtain the target predicted value.

8. A cell performance index prediction device, characterized in that: include: A cell performance indicator data acquisition module, used to acquire performance indicator data of a target cell within a first time period; A decomposition module, used to perform multivariate decomposition on the performance indicator data within the first time period to obtain a plurality of indicator component data corresponding to the time domain characteristics; A prediction module, used to input the plurality of indicator component data into the corresponding pre-trained component data prediction models respectively, and output the prediction values ​​corresponding to the plurality of indicator component data; The target prediction value obtaining module is used to obtain the target prediction value according to the prediction values ​​corresponding to the multiple indicator component data.

9. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to execute the cell performance indicator prediction method described in any one of claims 1 to 7 by executing the executable instructions.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the cell performance indicator prediction method according to any one of claims 1 to 7 is implemented.

11. A computer program product comprising: A computer program or instruction, characterized in that when the computer program or instruction is executed by a processor, it implements the cell performance indicator prediction method described in any one of claims 1 to 7.

Images