Flow-based Data Processing Method, Apparatus, Device, Medium and Program Product

Abstract

An embodiment of the present application provides a traffic-based data processing method, apparatus, device, medium, and program product, which can be used for traffic prediction in a vehicle network (such as an in-vehicle scenario). The traffic-based data processing method includes: obtaining a traffic feature sequence of a target object; the traffic feature sequence includes traffic feature data of n + 1 cycles before the (T + 1)-th cycle; obtaining a total traffic sequence of the target object; the total traffic sequence includes total traffic data of n + 1 cycles before the (T + 1)-th cycle; performing sequence decomposition processing on the total traffic sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence; performing traffic prediction based on the traffic feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence to obtain the total traffic data of the (T + 1)-th cycle. By adopting the embodiment of the present application, the accuracy of the traffic prediction result can be improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and particularly to a data processing method, device, equipment, medium and program product based on traffic. Background Art

[0002] Traffic prediction can be understood as predicting the possible traffic consumption in a future period based on the traffic consumed in a historical period; for example, predicting the possible traffic consumption in January - December 2022 based on the traffic consumed in January - December 2020 and January - December 2021; or predicting the possible traffic consumption in the 4th month based on the traffic consumed in the first 3 months. Accurately predicting the possible traffic consumption in a future period is beneficial to the reasonable allocation of future traffic and the reduction of traffic waste. Therefore, how to improve the accuracy of traffic prediction results has become a current research hotspot. Summary of the Invention

[0003] Embodiments of this application provide a data processing method, device, equipment, medium and program product based on traffic, which can improve the accuracy of traffic prediction results.

[0004] On the one hand, embodiments of this application provide a data processing method based on traffic, and the data processing method based on traffic includes:

[0005] Obtain a traffic feature sequence of a target object; the traffic feature sequence includes traffic feature data of n + 1 cycles before the (T + 1)th cycle, and the traffic feature data is used to reflect the traffic usage characteristics of the target object in the target device in the corresponding cycle, where n is a positive integer; obtain a total traffic sequence of the target object; the total traffic sequence includes total traffic data of n + 1 cycles before the (T + 1)th cycle, and the total traffic data is used to indicate the total traffic consumed by the target object in the target device in the corresponding cycle; perform sequence decomposition processing on the total traffic sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence; perform traffic prediction based on the traffic feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence to obtain the total traffic data of the (T + 1)th cycle.

[0006] Correspondingly, embodiments of this application provide a data processing device based on traffic, and the data processing device based on traffic includes:

[0007] An acquisition unit for acquiring a traffic feature sequence of a target object; the traffic feature sequence includes traffic feature data for n + 1 cycles before the (T + 1)-th cycle, and the traffic feature data is used to reflect the traffic usage characteristics of the target object in the target device in the corresponding cycle, where n is a positive integer; and, acquiring a total traffic sequence of the target object; the total traffic sequence includes total traffic data for n + 1 cycles before the (T + 1)-th cycle, and the total traffic data is used to indicate the total traffic consumed by the target object in the target device in the corresponding cycle;

[0008] A processing unit for performing sequence decomposition processing on the total traffic sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence; and, performing traffic prediction based on the traffic feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence to obtain the total traffic data for the (T + 1)-th cycle.

[0009] In one implementation, the traffic feature data for any cycle includes at least one of the following: traffic data consumed by the target object in the application running on the target device, remaining traffic data of the target object in the target device, traffic purchase data of the target object in the target device, traffic billing data for the target object, traffic data consumed by the target object using the target device on different holidays, and traffic data consumed by the target object using the target device in different seasons.

[0010] In one implementation, when the processing unit performs traffic prediction based on the traffic feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence to obtain the total traffic data for the (T + 1)-th cycle, it is specifically used to perform the following steps:

[0011] Performing trend traffic prediction on the (T + 1)-th cycle according to the traffic feature sequence and the trend item sequence to obtain the trend traffic data for the (T + 1)-th cycle; performing seasonal traffic prediction on the (T + 1)-th cycle according to the seasonal item sequence to obtain the seasonal traffic data for the (T + 1)-th cycle; performing holiday traffic prediction on the (T + 1)-th cycle according to the holiday item sequence to obtain the holiday traffic data for the (T + 1)-th cycle; determining the total traffic data for the (T + 1)-th cycle according to the trend traffic data for the (T + 1)-th cycle, the seasonal traffic data for the (T + 1)-th cycle, and the holiday traffic data for the (T + 1)-th cycle.

[0012] In one implementation, the trend item sequence includes the trend traffic data for each of the n + 1 cycles; when the processing unit performs trend traffic prediction on the (T + 1)-th cycle according to the traffic feature sequence and the trend item sequence to obtain the trend traffic data for the (T + 1)-th cycle, it is specifically used to perform the following steps:

[0013] Determine a feature prediction sequence in the traffic feature sequence. The feature prediction sequence includes traffic feature data for a total of n cycles starting from the (T - n + 1)-th cycle to the end of the T-th cycle in the traffic feature sequence; call the target trend traffic prediction model to perform trend traffic prediction on the (T + 1)-th cycle based on the feature prediction sequence and the trend item sequence, and obtain the trend traffic data for the (T + 1)-th cycle.

[0014] In one implementation, the processing unit is further configured to perform the following steps:

[0015] Obtain an initial trend traffic prediction model; perform connection construction on the traffic feature sequence according to the object identifier of the target object to obtain a trend training sequence. The trend training sequence includes trend training data for each cycle in n + 1 cycles; determine a training sample sequence and a test sample sequence in the trend training sequence; train the initial traffic prediction model according to the trend item sequence, the training sample sequence, and the test sample sequence to obtain the target trend traffic prediction model.

[0016] In one implementation, the seasonal item sequence includes seasonal traffic data for each cycle in n + 1 cycles. When the processing unit is used to perform seasonal traffic prediction on the (T + 1)-th cycle according to the seasonal item sequence and obtain the seasonal traffic data for the (T + 1)-th cycle, it is specifically configured to perform the following steps:

[0017] Determine a seasonal prediction sequence in the seasonal item sequence. The seasonal prediction sequence includes seasonal traffic data for a total of n cycles starting from the (T - n + 1)-th cycle to the end of the T-th cycle in the seasonal item sequence; call the target seasonal traffic prediction model to perform seasonal traffic prediction on the (T + 1)-th cycle according to the seasonal prediction sequence, and obtain the seasonal traffic data for the (T + 1)-th cycle.

[0018] In one implementation, the processing unit is further configured to perform the following steps:

[0019] Obtain an initial seasonal traffic prediction model; determine a seasonal training sequence in the seasonal item sequence. The seasonal training sequence includes seasonal traffic data for a total of n cycles starting from the (T - n)-th cycle to the end of the (T - 1)-th cycle in the seasonal item sequence; determine a training sample sequence and a test sample sequence from the seasonal training sequence; train the initial seasonal traffic model according to the training sample sequence and the test sample sequence to obtain the target seasonal traffic prediction model.

[0020] In one implementation, the holiday item sequence includes holiday traffic data for each cycle in n + 1 cycles. When the processing unit is used to perform holiday traffic prediction on the (T + 1)-th cycle according to the holiday item sequence and obtain the holiday traffic data for the (T + 1)-th cycle, it is specifically configured to perform the following steps:

[0021] Call the holiday traffic prediction model to obtain the holiday data sequences corresponding to the holiday traffic data in each of the n+1 cycles. The holiday data sequence includes the traffic data of M holidays in the corresponding cycle, where M is a positive integer; determine the predicted traffic data for each of the M holidays. The predicted traffic data for the i-th holiday among the M holidays is the average of the traffic data of the i-th holiday in each of the n+1 cycles, where i is a positive integer less than or equal to M; determine the holiday traffic data for the (T+1)-th cycle according to the predicted traffic data for each of the M holidays.

[0022] In one implementation, the processing unit, when used to perform sequence decomposition processing on the total traffic sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence, specifically performs the following steps:

[0023] Perform a moving average process on the total traffic sequence to obtain a trend item sequence, where the trend item sequence includes the trend traffic data for each of the n+1 cycles; determine the seasonal item sequence according to the total traffic sequence and the trend item sequence; determine the holiday item sequence according to the total traffic sequence, the trend item sequence, and the seasonal item sequence.

[0024] In one implementation, the processing unit, when used to determine the seasonal item sequence according to the total traffic sequence and the trend item sequence, specifically performs the following steps:

[0025] Determine a detrended item sequence according to the total traffic sequence and the trend item sequence; the detrended item sequence includes the detrended traffic data for each of the n+1 cycles, and the detrended traffic data for any one cycle is determined according to the total traffic data for the corresponding cycle and the trend traffic data for the corresponding cycle; perform an average process on the detrended item sequence according to seasons to obtain a seasonal item sequence, where the seasonal item sequence includes the seasonal traffic data for each of the n+1 cycles.

[0026] In one implementation, the processing unit, when used to determine the holiday item sequence according to the total traffic sequence, the trend item sequence, and the seasonal item sequence, specifically performs the following steps:

[0027] Determine a detrended and deseasonalized item sequence according to the total traffic sequence, the trend item sequence, and the seasonal item sequence; the detrended and deseasonalized item sequence includes the detrended and deseasonalized traffic data for each of the n+1 cycles, and the detrended and deseasonalized traffic data for any one cycle is determined according to the total traffic data for the corresponding cycle, the trend traffic data for the corresponding cycle, and the seasonal traffic data for the corresponding cycle; perform an average process on the detrended and deseasonalized item sequence according to holidays to obtain a holiday item sequence, where the holiday item sequence includes the holiday traffic data for each of the n+1 cycles.

[0028] Accordingly, an embodiment of the present application provides a computer device, which includes a processor and a computer-readable storage medium. The processor is adapted to implement a computer program. The computer-readable storage medium stores a computer program, and the computer program is adapted to be loaded and executed by the processor to perform the above data processing method.

[0029] Accordingly, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is read and executed by the processor of the computer device, the computer device is caused to perform the above data processing method.

[0030] Accordingly, an embodiment of the present application provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, causing the computer device to perform the above data processing method.

[0031] In an embodiment of the present application, when it is necessary to perform traffic prediction on the traffic usage of a target object in the (T + 1)-th period, the traffic feature sequence of the target object and the overall traffic sequence of the target object can be obtained. Then, the total traffic sequence can be subjected to sequence decomposition processing to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence. Then, traffic prediction can be performed according to the traffic feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence to obtain the total traffic data in the (T + 1)-th period. Among them, the traffic feature sequence may include traffic feature data in n + 1 periods before the (T + 1)-th period, and the traffic feature data can be used to reflect the traffic usage characteristics of the target object in the target device in the corresponding period. The total traffic sequence may include total traffic data in n + 1 periods before the (T + 1)-th period, and the total traffic data can be used to indicate the total traffic consumed by the target object in the target device in the corresponding period. It can be seen from the above that using the traffic feature sequence to participate in traffic prediction can make the traffic prediction result more in line with the traffic usage characteristics of the target object in the target device and improve the accuracy of the traffic prediction result. By performing sequence decomposition processing on the total traffic sequence, traffic prediction can be performed from different perspectives, further improving the accuracy of the traffic prediction result. Description of the Drawings

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0033] Figure 1 It is a schematic flowchart of a data processing method based on traffic provided by an embodiment of the present application;

[0034] Figure 2 It is a schematic architecture diagram of a data processing system provided by an embodiment of the present application;

[0035] Figure 3 It is a schematic scenario diagram of a vehicle-mounted scenario provided by an embodiment of the present application;

[0036] Figure 4 It is a schematic flowchart of another data processing method based on traffic provided by an embodiment of the present application;

[0037] Figure 5 It is a schematic flowchart of another data processing method based on traffic provided by an embodiment of the present application;

[0038] Figure 6 It is a schematic structural diagram of a data processing device based on traffic provided by an embodiment of the present application;

[0039] Figure 7 It is a schematic structural diagram of a computer device provided by an embodiment of the present application. Detailed implementation manners

[0040] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0041] To better understand the technical solutions provided by the embodiments of the present application, the key terms related to the embodiments of the present application are introduced herein:

[0042] (1) Embodiments of this application relate to traffic. Traffic refers to the data traffic consumed by a target object (i.e., a traffic user) when accessing the Internet through a target device. The prerequisite for the target object to access the Internet through the target device is that traffic-related hardware can be installed in the target device. Such traffic-related hardware can be, for example, a vehicle networking card, an Internet of Things card, a mobile traffic card, a mobile phone card, etc. Traffic generally has seasonal effects and holiday effects; among them, the seasonal effect of traffic refers to the phenomenon that traffic changes regularly with seasons (the seasons here can be understood as each month of the year). For example, the traffic consumed during the winter vacation (including January and February) and the summer vacation (including July and August) each year is higher than that consumed in other months of the year; the holiday effect of traffic refers to the phenomenon that traffic changes regularly with holidays. For example, the traffic consumed during the National Day and Spring Festival each year is higher than that consumed in other periods of the year, and the traffic consumed on rest days is higher than that consumed on working days.

[0043] (2) Embodiments of this application relate to traffic prediction. Traffic prediction can be understood as predicting the traffic that a target object may consume when accessing the Internet through a target device in a future period based on the traffic that the target object has consumed when accessing the Internet through the target device in a historical period. For example, based on the traffic that the target object has consumed when accessing the Internet through the target device from January to December 2020 and from January to December 2021, predict the traffic that the target object may consume when accessing the Internet through the target device from January to December 2022; another example is to predict the traffic that the target object may consume when accessing the Internet through the target device in the 4th month based on the traffic that the target object has consumed when accessing the Internet through the target device from January to March.

[0044] To improve the accuracy of traffic prediction results, embodiments of this application propose a data processing method based on traffic. When performing traffic prediction, this data processing method based on traffic takes into account, on the one hand, the traffic usage characteristics of the target object in a historical period, and on the other hand, the change trend of traffic over time, the seasonal effect of traffic, and the holiday effect of traffic in a historical period. That is to say, embodiments of this application comprehensively consider the traffic usage characteristics of traffic users, the change trend of traffic over time, the seasonal effect of traffic, and the holiday effect of traffic in a historical period to predict the traffic that may be consumed in a future period, so that the traffic prediction results can better conform to the traffic usage characteristics of traffic users and the traffic prediction results are more accurate.

[0045] Herein, in combination with Figure 1The overall process of the traffic-based data processing method provided in the embodiment of the present application is sorted out as follows: ① When it is necessary to predict the traffic flow for the T+1th period, the traffic feature sequence and the total traffic flow sequence of the target object can be obtained; the traffic feature sequence can include the traffic feature data of the n+1 periods before the T+1th period, and the n+1 period can be expressed as {Tn, T-n+1, ..., T}. The traffic feature data of any period is used to reflect the traffic usage characteristics of the target object in the target device during that period; the total traffic sequence can include the total traffic data of the n+1 periods before the T+1th period, and the total traffic data of any period is used to indicate the total traffic consumed by the target object in the target device during that period. ② The total traffic sequence can be decomposed to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence; wherein the trend item sequence can include the trend traffic data of the n+1 periods, which can be used to reflect the trend of traffic changes over time; the seasonal item sequence can include the seasonal traffic data of the n+1 periods, which can be used to reflect the seasonal effect of traffic; and the holiday item sequence can include the holiday traffic data of the n+1 periods, which can be used to reflect the holiday effect of traffic. ③ The target trend flow forecast model can be called to perform trend flow forecast for the T+1th period based on the flow feature sequence and trend item sequence, obtaining trend flow data for the T+1th period. ④ The target seasonal flow forecast model can be called to perform seasonal flow forecast for the T+1th period based on the seasonal item sequence, obtaining seasonal flow data for the T+1th period. ⑤ The holiday flow forecast model can be called to perform holiday flow forecast for the T+1th period based on the holiday item sequence, obtaining holiday flow data for the T+1th period. ⑥ The trend flow data for the T+1th period, the seasonal flow data for the T+1th period, and the holiday flow data for the T+1th period can be added together to obtain total flow data for the T+1th period.

[0046] Among them, the period refers to the time interval for statistical analysis of traffic data. For example, a period can be one year, one month, one week, etc. Taking one month as an example of the period, the overall process of the traffic data-based processing method is demonstrated as follows: When T + 1 = 12 and n + 1 = 11, that is, when traffic prediction for December is required, the traffic feature sequence and the total traffic sequence can be obtained. The traffic feature sequence can include traffic feature data for the 11 months before December (i.e., January to November), and the total traffic sequence can include total traffic data for the 11 months before December (i.e., January to November). Then, the total traffic sequence can be decomposed by time series to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence. The trend item sequence can include trend traffic data for January to November, the seasonal item sequence can include seasonal traffic data for January to November, and the holiday item sequence can include holiday traffic data for January to November. Then, the trend traffic data for December can be predicted based on the traffic feature sequence and the trend item sequence, the seasonal traffic data for December can be predicted based on the seasonal item sequence, and the holiday item traffic data for December can be predicted based on the holiday item sequence. Finally, the trend traffic data for December, the seasonal traffic data for December, and the holiday traffic data for December can be added together to obtain the total traffic data for December.

[0047] The following combines Figure 2 to introduce a data processing system suitable for implementing the traffic data-based processing method provided in the embodiments of the present application. As Figure 2 shown, the data processing system can include a target device 201 and a server 202. The target device 201 and the server 202 can be directly connected through a wired communication method or indirectly connected through a wireless communication method. Among them, the target device 201 can include, but is not limited to, a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart voice interaction device, a smart home appliance, a smart watch, a vehicle-mounted terminal, etc., but is not limited thereto; the server 202 can be a server cluster or a distributed system composed of multiple physical servers, and can also be a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data.

[0048] In Figure 2In the data processing system shown, the model training process can be executed by the server. For example, the server can train the initial trend traffic prediction model to obtain the above-mentioned target trend traffic prediction model for trend traffic prediction. Another example is that the server can train the initial seasonal traffic prediction model to obtain the above-mentioned target seasonal traffic prediction model for seasonal traffic prediction. The traffic prediction process can be executed by the target device or by the server. The following introduces these two cases separately:

[0049] When the traffic prediction process is executed by the target device, the target trend traffic prediction model, the target seasonal traffic prediction model, and the holiday traffic prediction model can be deployed in the target device. After the target device obtains the traffic feature sequence and the total traffic sequence of the target object, the total traffic sequence can be decomposed into a trend item sequence, a seasonal item sequence, and a holiday item sequence. Then, the target trend traffic prediction model can be called to perform trend traffic prediction based on the traffic feature sequence and the trend item sequence to obtain the trend traffic data for the (T + 1)-th period, the target seasonal traffic model can be called to perform seasonal traffic prediction based on the seasonal item sequence to obtain the seasonal traffic data for the (T + 1)-th period, and the holiday traffic prediction model can be called to perform holiday traffic prediction based on the holiday item sequence to obtain the holiday traffic data for the (T + 1)-th period. Then, the trend traffic data for the (T + 1)-th period, the seasonal traffic data for the (T + 1)-th period, and the holiday traffic data for the (T + 1)-th period can be added together to obtain the total traffic data for the (T + 1)-th period.

[0050] When the traffic prediction process is executed by the server, the target trend traffic prediction model, the target seasonal traffic prediction model, and the holiday traffic prediction model can be deployed in the server. The server can receive the traffic feature sequence and the total traffic sequence sent by the target device. The server can decompose the total traffic sequence into a trend item sequence, a seasonal item sequence, and a holiday item sequence. Then, the target trend traffic prediction model can be called to perform trend traffic prediction based on the traffic feature sequence and the trend item sequence to obtain the trend traffic data for the (T + 1)-th period, the target seasonal traffic model can be called to perform seasonal traffic prediction based on the seasonal item sequence to obtain the seasonal traffic data for the (T + 1)-th period, and the holiday traffic prediction model can be called to perform holiday traffic prediction based on the holiday item sequence to obtain the holiday traffic data for the (T + 1)-th period. Then, the trend traffic data for the (T + 1)-th period, the seasonal traffic data for the (T + 1)-th period, and the holiday traffic data for the (T + 1)-th period can be added together to obtain the total traffic data for the (T + 1)-th period. Then, the server can send the total traffic data for the (T + 1)-th period to the target device.

[0051] In Figure 2Under the system architecture shown, by using the traffic feature sequence reflecting the traffic usage characteristics of the target object to participate in the traffic prediction process, the total traffic data for the (T + 1)-th period obtained by prediction can be made to better conform to the traffic usage characteristics of the target object, improving the accuracy of the traffic prediction result; by decomposing the total traffic sequence and performing traffic prediction from different perspectives, the accuracy of the traffic prediction result can be further improved. It can be understood that the data processing system described in the embodiments of this application is for more clearly explaining the technical solutions of the embodiments of this application and does not constitute a limitation on the technical solutions provided in the embodiments of this application. Those of ordinary skill in the art know that with the evolution of the system architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of this application are equally applicable to similar technical problems.

[0052] Under Figure 2 the system architecture shown, the following combines Figure 3 to introduce the application of the data processing solution in the vehicle network (such as the in-vehicle scenario), as Figure 3 shown, an in-vehicle terminal can be installed in the vehicle. The in-vehicle terminal can obtain the traffic feature sequence and the total traffic sequence of the target object for traffic prediction to obtain the total traffic data for the (T + 1)-th period; the in-vehicle terminal can also output the total traffic data for the (T + 1)-th period, and can also output the remaining traffic data for the (T + 1)-th period calculated based on the total traffic data for the (T + 1)-th period, so that after the target object views the total traffic data for the (T + 1)-th period and the remaining traffic data for the (T + 1)-th period through the in-vehicle terminal, it can reasonably allocate traffic for the (T + 1)-th period and reduce traffic waste. For example, if the total traffic data for the (T + 1)-th period obtained by prediction is large, resulting in very little remaining traffic data for the (T + 1)-th period, the target object can reduce some traffic-consuming operations in the in-vehicle device that consume a large amount of traffic and consume traffic quickly, such as reducing playing online songs and reducing playing online videos, etc.

[0053] Based on the above description of the data processing system and the description of the application of the traffic-based data processing method in the in-vehicle scenario, the traffic-based data processing method proposed in the embodiments of this application will be introduced in detail below with reference to the accompanying drawings.

[0054] The embodiments of this application propose a traffic-based data processing method. This traffic-based data processing method mainly introduces the traffic feature sequence and the sequence decomposition process of the total traffic sequence, etc.; this traffic-based data processing method can be executed by the target device 201 or the server 202 in the data processing system mentioned above. The embodiments of this application do not limit this. Please refer to Figure 4 , this traffic-based data processing method may include the following steps S401 to step S404:

[0055] S401. Obtain the traffic feature sequence of the target object.

[0056] The traffic feature sequence of the target object can be obtained by preprocessing the log data of the target object. The log data of the target object in n + 1 cycles before the (T + 1)-th cycle can be obtained, and then the log data in n + 1 cycles can be preprocessed to obtain the traffic feature sequence of the target object. The traffic feature sequence can include the traffic feature data in n + 1 cycles before the (T + 1)-th cycle, and the traffic feature data in any cycle can be used to reflect the traffic usage characteristics of the target object in the target device in that cycle; the traffic feature sequence can be expressed as {X T-n , X T-n+1 , …, X T}, where X T represents the traffic feature data in the T-th cycle.

[0057] The preprocessing process of the log data can be understood as a statistical process. Since the data recorded in the log data is scattered, for example, the log data records the traffic consumed by the target object in the target device per hour, the traffic consumed by the target object in each application program in the target device per hour, the traffic consumed by the target object in the target device per minute, the traffic consumed by the target object in each application program in the target device per minute, etc., the scattered data is not conducive to traffic prediction. Therefore, it is necessary to statistically process the log data and count the scattered data into the data in each cycle.

[0058] The traffic characteristic data of any period may include at least one of the following: ① The traffic data consumed by the target object in the application running on the target device. Here, the traffic data consumed in the application can have two understandings: The first understanding is the total traffic data consumed in the application, such as the traffic data consumed in the first application running on the target device, the traffic data consumed in the second application running on the target device, and so on; The second understanding is the traffic data consumed in each function of the application, such as the social session function, audio and video playback function, etc. in the first application running on the target device; The number of applications can be one or more, and the number of functions can be one or more. ② The remaining traffic data of the target object in the target device. ③ The traffic purchase data of the target object in the target device. ④ The traffic billing data for the target object. ⑤ The traffic data consumed by the target object using the target device on different holidays; Here, the traffic data consumed using the target device on holidays can have two understandings: The first understanding is the total traffic data consumed using the target device on holidays, and the second understanding is the traffic peak data consumed using the target device on holidays (for example, taking the traffic data of the day with the most consumed traffic during the holidays as the traffic peak data for the holidays). ⑥ The traffic data consumed by the target object using the target device in different seasons; Here, the traffic data consumed using the target device in each season can have two understandings: The first understanding is the total traffic data consumed using the target device in each season, and the second understanding is the traffic peak data consumed using the target device in each season (for example, taking the traffic data of the day with the most consumed traffic during the season as the traffic peak data for the season).

[0059] S402. Obtain the total traffic sequence of the target object.

[0060] The total traffic sequence of the target object may include the total traffic data of n + 1 periods before the (T + 1)-th period, and the total traffic sequence can be expressed as {Y T-n , Y T-n+1 , …, Y T}, where Y T represents the total traffic data of the T-th period. The total traffic data of any period is used to indicate the total traffic consumed by the target object in the target device in that period. Taking one period as one month as an example, an exemplary total traffic sequence can be seen in Table 1 below. The total traffic sequence shown in Table 1 includes the total traffic data of 36 months (i.e., 36 periods):

[0061] Table 1

[0062]

[0063] S403. Perform sequence decomposition processing on the total traffic sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence.

[0064] After obtaining the total traffic volume sequence of the target object, the total traffic volume sequence can be decomposed to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence. The following introduces the specific process of sequence decomposition:

[0065] (1) For the trend item sequence, the total traffic volume sequence can be processed by moving average to obtain the trend item sequence; the trend item sequence can include the trend traffic volume data of each cycle in n + 1 cycles, and the trend item sequence can be expressed as {G T-n , G T-n+1 , …, G T}, where G T represents the trend traffic volume data of the T-th cycle. The moving average processing can specifically be performed using the MA (Moving Average Model) model, and m = 12 (i.e., 12 cycles) can be taken as the moving cycle. Taking Table 1 above as an example, the traffic volume data for January 2019 can be obtained by averaging the traffic volume data for January 2019 and the traffic volume data for February - November 2018 (i.e., b1, a2, a3, a4, a5, a6, a7, a8, a9, a10, a11, a12), and the traffic volume data for January 2020 can be obtained by averaging the traffic volume data for January 2020 and the traffic volume data for February - December 2019 (i.e., c1, b2, b3, b4, b5, b6, b7, b8, b9, b10, b11, b12).

[0066] (2) For the seasonal item sequence, the seasonal item sequence can be determined based on the total traffic volume sequence and the trend item sequence. Specifically, the detrended item sequence can be determined based on the total traffic volume sequence and the trend item sequence; the detrended item sequence can include the detrended traffic volume data of each cycle in n + 1 cycles, and the detrended traffic volume data of any cycle can be determined based on the total traffic volume data of the corresponding cycle and the trend traffic volume data of the corresponding cycle, and the detrended traffic volume data of any cycle can be equal to the total traffic volume data of the corresponding cycle minus the trend traffic volume data of the corresponding cycle; the detrended item sequence can be expressed as {Y T-n - G T-n , Y T-n+1 - G T-n+1 , …, Y T - G T}, where Y T - G TDenote the detrended traffic data for the \(T\) -th period. The detrended traffic data for the \(T\) -th period can be equal to the total traffic data for the \(T\) -th period minus the trend traffic data for the \(T\) -th period. Then, a period can be regarded as a season, and the detrended term sequence can be averaged according to seasons to obtain a seasonal term sequence. For example, for monthly data (i.e., one period is one month and one month is regarded as a season), the seasonal traffic data for March can be the average of all the detrended traffic data for March in the detrended term sequence. Arranging and combining the data of these respective periods together can obtain the seasonal term sequence. Taking Table 1 above as an example, the traffic data for March 2019 can be the average of the detrended traffic data for March 2018 and the detrended traffic data for March 2019 in the detrended term sequence, and the detrended traffic data for March 2020 can be the average of the detrended traffic data for March 2019 and the detrended traffic data for March 2020 in the detrended term sequence. The seasonal term sequence can include the seasonal traffic data for each of the \(n + 1\) periods, and the seasonal term sequence can be expressed as \(\{S T-n ,S T-n+1 ,…,S T \}\), where \(S T \) represents the seasonal traffic data for the \(T\) -th period.

[0067] (3) For the holiday term sequence, it can be determined based on the total traffic sequence, the trend term sequence, and the seasonal term sequence. Specifically, the detrended and deseasonalized term sequence can be determined according to the total traffic sequence, the trend term sequence, and the seasonal term sequence. The detrended and deseasonalized term sequence can include the detrended and deseasonalized traffic data for each of the \(n + 1\) periods. The detrended and deseasonalized traffic data for any period can be determined based on the total traffic data for the corresponding period, the trend traffic data for the corresponding period, and the seasonal traffic data for the corresponding period. The detrended and deseasonalized traffic data for any period can be equal to the total traffic data for the corresponding period minus the trend traffic data for the corresponding period and then minus the seasonal traffic data for the corresponding period. The detrended and deseasonalized term sequence can be expressed as \(\{Y T-n - G T-n - S T-n ,Y T-n+1 - G T-n+1 - S T-n+1 ,…,Y T - G T - S T \}\), where \(Y T - G T - S TDenote the detrended and deseasonalized flow data for the T-th period. The detrended and deseasonalized flow data for the T-th period can be equal to the total flow data for the T-th period minus the trend flow data for the T-th period and then minus the seasonal flow data for the T-th period. Then, the detrended and deseasonalized term sequence can be averaged according to holidays to obtain the holiday term sequence; the holiday term sequence can include the holiday flow data for each period in n + 1 periods, and the holiday term sequence can be expressed as {H T-n ,H T-n+1 ,…,H T}, where H T represents the holiday flow data for the T-th period.

[0068] The process of averaging the detrended and deseasonalized term sequence according to holidays to obtain the holiday term sequence can include: Each period can contain M holidays. Taking the i-th holiday among the M holidays as an example, the flow data of the i-th holiday is the average of the flow data of all the i-th holidays in the detrended and deseasonalized term sequence. Combining the data of these periods together can obtain the sequence of the i-th holiday in each period of n + 1 periods (i = 1, 2, …, M, each period contains M holidays, M is a positive integer, and i is a positive integer less than or equal to M), and then, the holiday flow data for each period in n + 1 periods can be obtained according to the following formula 1, that is, the holiday term sequence is obtained.

[0069]

[0070] Taking the holiday flow data for the T-th period in the holiday term sequence as an example for the above formula 1, the other periods except the T-th period in the holiday term sequence can refer to the T-th period. The parameters in the above formula 1 are explained here: H T represents the holiday flow data for the T-th period; i represents the i-th holiday; M represents that each period contains M holidays; represents the flow data of the i-th holiday in the T-th period; D i represents a time interval before and after the i-th holiday, represents that if T belongs to the time interval indicated by D i it takes the value of 1, and if T does not belong to the time interval indicated by D i it takes the value of 0.

[0071] In summary, for the content of decomposing the total flow sequence into the trend term sequence, the seasonal term sequence, and the holiday term sequence, taking the T-th period as an example, the sequence decomposition process can be expressed as the following formula 2 and formula 3:

[0072] Y T =G T+S T +H T +ε T Formula 2

[0073] ε T ~N(0,1) Formula 3

[0074] The parameters in the above Formulas 2 and 3 are explained below: Y T represents the total flow data of the T-th period; G T represents the trend flow data of the T-th period; S T represents the seasonal flow data of the T-th period; H T represents the holiday flow data of the T-th period; ε T represents the fitting residual sequence, and N(0,1) represents the standard normal distribution, that is, the fitting residual sequence conforms to the standard normal distribution.

[0075] S404. Perform flow prediction based on the flow feature sequence, trend item sequence, seasonal item sequence, and holiday item sequence to obtain the total flow data of the (T + 1)-th period.

[0076] After performing sequence decomposition processing on the total flow sequence to obtain the trend item sequence, seasonal item sequence, and holiday item sequence, then flow prediction can be performed based on the flow feature sequence, trend item sequence, seasonal item sequence, and holiday item sequence to obtain the total flow data of the (T + 1)-th period. Among them, performing flow prediction based on the flow feature sequence, trend item sequence, seasonal item sequence, and holiday item sequence to obtain the total flow data of the (T + 1)-th period can include: performing trend flow prediction on the (T + 1)-th period based on the flow feature sequence and trend item sequence to obtain the trend flow data of the (T + 1)-th period; performing seasonal flow prediction on the (T + 1)-th period based on the seasonal item sequence to obtain the seasonal flow data of the (T + 1)-th period; performing holiday flow prediction on the (T + 1)-th period based on the holiday item sequence to obtain the holiday flow data of the (T + 1)-th period; determining the total flow data of the (T + 1)-th period based on the trend flow data of the (T + 1)-th period, the seasonal flow data of the (T + 1)-th period, and the holiday flow data of the (T + 1)-th period.

[0077] In the embodiments of the present application, when it is necessary to perform traffic prediction on the traffic usage of a target object in the (T + 1)-th period, the traffic feature sequence of the target object and the overall traffic sequence of the target object can be obtained. Then, the total traffic sequence can be subjected to sequence decomposition processing to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence. Then, traffic prediction can be performed based on the traffic feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence to obtain the total traffic data for the (T + 1)-th period. Among them, the traffic feature sequence can include traffic feature data for n + 1 periods before the (T + 1)-th period, and the traffic feature data can be used to reflect the traffic usage characteristics of the target object in the target device in the corresponding period; the total traffic sequence can include the total traffic data for n + 1 periods before the (T + 1)-th period, and the total traffic data can be used to indicate the total traffic consumed by the target object in the target device in the corresponding period. It is not difficult to see from the above content that using the traffic feature sequence to participate in traffic prediction can make the traffic prediction result more in line with the traffic usage characteristics of the target object in the target device and improve the accuracy of the traffic prediction result; by performing sequence decomposition processing on the total traffic sequence, traffic prediction can be performed from different perspectives, further improving the accuracy of the traffic prediction result. By statistically obtaining the traffic feature sequence from the log data, the data used for traffic prediction can be made more regular, which is conducive to improving the traffic prediction efficiency.

[0078] The embodiments of the present application propose a traffic-based data processing method. This traffic-based data processing method mainly introduces the training process of the target trend traffic prediction model, the trend traffic prediction process of the target trend traffic prediction model, the training process of the target seasonal traffic prediction model, the seasonal traffic prediction process of the target seasonal traffic prediction model, the traffic feature sequence, and the holiday traffic prediction process of the holiday traffic prediction model, etc.; this traffic-based data processing method can be executed by the target device 201 or the server 202 in the above-mentioned data processing system, and the embodiments of the present application do not limit this. Please refer to Figure 5 , this traffic-based data processing method may include the following steps S501 to S507:

[0079] S501, obtain the traffic feature sequence of the target object.

[0080] In the embodiments of the present application, the execution process of step S501 is the same as the execution process of step S401 in the embodiment shown above Figure 4 , and the specific description can be referred to the description of step S401 in the embodiment shown above Figure 4 , and will not be repeated here.

[0081] S502, obtain the total traffic sequence of the target object.

[0082] The execution process of step S502 in the embodiment of the present application is the same as that of step S402 in the above Figure 4 shown embodiment, and for details, reference may be made to the description of step S402 in the above Figure 4 shown embodiment, which will not be elaborated herein.

[0083] S503. Perform sequence decomposition processing on the total flow sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence.

[0084] The execution process of step S503 in the embodiment of the present application is the same as that of step S403 in the above Figure 4 shown embodiment, and for details, reference may be made to the description of step S403 in the above Figure 4 shown embodiment, which will not be elaborated herein.

[0085] S504. Perform trend flow prediction on the (T + 1)-th cycle according to the flow feature sequence and the trend item sequence to obtain the trend flow data of the (T + 1)-th cycle.

[0086] The trend flow data of the (T + 1)-th cycle may be obtained by invoking a target trend flow prediction model for trend flow prediction. The target trend flow prediction model may be obtained by training an initial trend flow prediction model. The initial trend flow prediction model and the target trend flow prediction model mentioned in the embodiment of the present application may specifically refer to an RNN (Recurrent Neural Network) model. RNN is a type of recursive neural network with sequence data as input, performing recursion in the evolution direction of the sequence and having all nodes (recurrent units) connected in a chain, aiming to process time series data. The training process of the initial trend flow prediction model may include: First, the initial trend flow prediction model may be obtained. Second, according to the object identifier of the target object, link construction may be performed on the flow feature sequence {X T-n , X T-n+1 , …, X T} to obtain a trend training sequence; the trend training sequence may include trend training data for each cycle in n + 1 cycles, and the trend training sequence may be expressed as {Sa T-n , Sa T-n+1 , …, Sa T}, where Sa TDenote the trend training data of the T-th period; here, link construction can be understood as follows: when any traffic feature data in the traffic feature sequence includes multiple items, for example, any traffic feature data includes the traffic data consumed by the target object in the first application running on the target device and the traffic data consumed by the target object in the second application running on the target device, and both the first application and the second application are logged in and registered using the object identifier of the target object, then these two items can be associated through the object identifier of the target object and used together as the trend training data in the trend training sequence. Then, a training sample sequence and a test sample sequence can be determined in the trend training sequence. The training sample sequence and the test sample sequence can be extracted from the trend training sequence according to a preset ratio. For example, a preset ratio (such as 80%) of the trend training data can be randomly extracted from the trend training sequence as the training sample sequence, and the other trend training data in the trend training sequence except for the trend training data included in the training sample sequence can be used as the training test sample; among them, the training sample can be used to train the initial trend traffic prediction model to obtain the target trend traffic prediction model, and the test sample sequence can be used to test the performance of the target trend traffic prediction model. The initial trend traffic prediction model can be expressed as the following formulas 4 and 5:

[0087] G T = RNN(X T , G T-1 ) Formula 4

[0088]

[0089] The above formulas are illustrated by taking the T-th period and the (T - 1)-th period as examples. Here, the parameters in the above formulas 4 and 5 are explained: G T denotes the trend traffic data of the T-th period, G T-1 denotes the trend traffic data of the (T - 1)-th period; X T denotes the traffic feature data of the T-th period, X T-1 denotes the traffic feature data of the (T - 1)-th period; K T and K T-1 denote the hidden layer vectors; W k , W g and U k denote the parameter matrices; b k and b g denote the parameter vectors; σ k and σ g denote the activation functions. For example, σ k can be the tanh (hyperbolic tangent) function or the ReLU (Linear Rectification Function), and σ gIt can be a sigmoid function; the hidden layer K0 can adopt the trend traffic data of the (T - n)th period (i.e., G0 = G T-n ).

[0090] After determining the training sample sequence and the test sample sequence in the trend training sequence, the initial trend traffic prediction model can be trained according to the trend item sequence, the training sample sequence, and the test sample sequence to obtain the parameter matrix and the parameter matrix Thus, the target trend traffic prediction model can be obtained.

[0091] After training the target trend traffic prediction model, a feature prediction sequence can be determined in the traffic feature sequence {X T-n , X T-n+1 , …, X T}. The feature prediction sequence can include the traffic feature data of a total of n periods from the (T - n + 1)th period to the Tth period in the traffic feature sequence, and the feature prediction sequence can be expressed as {X T-n+1 , X T-n+2 , …, X T}; then, the target trend traffic prediction model can be called to predict the trend traffic of the (T + 1)th period according to the feature prediction sequence and the trend item sequence, and the trend traffic data of the (T + 1)th period can be obtained Among them, in the process of calling the target trend traffic prediction model for trend traffic prediction, the trend traffic data of the (T - n + 1)th period included in the trend item sequence can be used as the hidden layer K0, and the feature prediction sequence {X T-n+1 , X T-n+2 , …, X T} can be used as the input data of the target trend traffic prediction model. After the prediction of the target trend traffic prediction model, the output data of the target trend traffic prediction model is the trend traffic data of the (T + 1)th period

[0092] S505. Perform seasonal traffic prediction on the (T + 1)th period according to the seasonal item sequence to obtain the seasonal traffic data of the (T + 1)th period.

[0093] The seasonal traffic data of the (T + 1)th period can be obtained by calling the target seasonal traffic prediction model for seasonal traffic prediction. The target seasonal traffic prediction model can be obtained by training the initial seasonal traffic prediction model. The training process of the initial seasonal traffic prediction model can include: First, the initial seasonal traffic prediction model can be obtained. Second, in the seasonal item sequence {S T-n , S T-n+1 , …, S T} Determine the seasonal training sequence. The seasonal training sequence may include the seasonal flow data of a total of n cycles starting from the (T - n)-th cycle to the end of the (T - 1)-th cycle in the seasonal item sequence. The seasonal training data can be expressed as {S T-n , S T-n+1 , …, S T-1}. Then, the training sample sequence and the test sample sequence can be determined from the seasonal training sequence. The training sample sequence and the test sample sequence can be extracted from the seasonal training sequence according to a preset ratio. For example, a preset ratio (such as 80%) of the seasonal flow data can be randomly extracted from the seasonal training sequence as the training sample sequence, and the other seasonal flow data in the seasonal training sequence except the seasonal flow data included in the training sample sequence can be used as the training test sample. Among them, the training sample sequence can be used to train the initial seasonal flow prediction model to obtain the target seasonal flow prediction model, and the test sample sequence can be used to test the performance of the target seasonal flow prediction model. Then, the initial seasonal flow model can be trained according to the training sample sequence and the test sample sequence to obtain the target seasonal flow prediction model. The initial seasonal flow prediction model can be expressed by the following formula 6:

[0094]

[0095] Taking the holiday flow data of the T-th cycle in the seasonal item sequence as an example in the above formula 6, other cycles in the seasonal item sequence except the T-th cycle can refer to the T-th cycle. Here, the parameters in the above formula 6 are explained: S T can represent the seasonal flow data of the T-th cycle; N represents the number of seasons is N; t n can represent the n-th time cycle, where n is an integer greater than or equal to 1 and less than or equal to N; both a and b are model parameters of the initial seasonal flow prediction model.

[0096] Based on the above formula 6, the process of training the initial seasonal flow model according to the training sample sequence and the test sample sequence may include: substituting the seasonal flow data of each cycle in the training sample sequence into the above formula 6 for linear regression processing to obtain the parameter values of the model parameters of the initial seasonal flow prediction model (including the parameter value of the model parameter a and the parameter value of the model parameter b), then substituting the parameter values of the model parameters into the above formula 6 to obtain the intermediate seasonal flow prediction model, and then performing performance testing on the target seasonal flow model according to the test sample sequence, and continuously optimizing the model parameters of the intermediate seasonal flow prediction model when the performance test result does not meet the performance condition until the performance test result meets the performance condition, and the intermediate seasonal flow prediction model with the performance test result meeting the performance condition can be determined as the target seasonal flow prediction model.

[0097] After training the target seasonal traffic prediction model, a seasonal prediction sequence can be determined in the seasonal item sequence. The seasonal prediction sequence can include the seasonal traffic data of a total of n cycles starting from the (T - n + 1)-th cycle to the T-th cycle in the seasonal item sequence, and the seasonal prediction sequence can be expressed as {S T-n+1 , S T-n+2 , …, S T}. Then, the target seasonal traffic prediction model can be called to perform seasonal traffic prediction on the (T + 1)-th cycle according to the seasonal prediction sequence, and the seasonal traffic data of the (T + 1)-th cycle can be obtained. The seasonal traffic data of the (T + 1)-th cycle can be expressed as

[0098] S506, perform holiday traffic prediction on the (T + 1)-th cycle according to the holiday item sequence, and obtain the holiday traffic data of the (T + 1)-th cycle.

[0099] As can be seen from the foregoing, the holiday item sequence {H T-n , H T-n+1 , …, H T} can include the holiday traffic data of each cycle in n + 1 cycles. The process of performing holiday traffic prediction on the (T + 1)-th cycle according to the holiday item sequence to obtain the holiday traffic data of the (T + 1)-th cycle can include:

[0100] First, the holiday traffic prediction model can be called to obtain the holiday data sequence corresponding to the holiday traffic data of each cycle in n + 1 cycles. Any holiday data sequence can include the traffic data of M holidays in the corresponding cycle, where M is a positive integer. Taking the T-th cycle as an example, the holiday data sequence corresponding to the T-th cycle can be expressed as The holiday data sequences corresponding to the holiday traffic data of all cycles in n + 1 cycles can be generally expressed as

[0101] Second, the predicted traffic data of each of the M holidays can be determined. Taking the i-th holiday among the M holidays as an example, the predicted traffic data of the i-th holiday among the M holidays can be the average value of the traffic data of the i-th holiday in each cycle of n + 1 cycles, where i is a positive integer less than or equal to M. The determination process of the predicted traffic data of the i-th holiday among the M holidays can be referred to the following formula 7:

[0102]

[0103] Here, the parameters in the above formula 7 are explained: represents the predicted traffic data of the i-th holiday; represents the traffic data of the i-th holiday in the (T - j)-th cycle; Represents the sum of the traffic data for the i-th holiday in n + 1 cycles.

[0104] Then, based on the predicted traffic data for each holiday among the M holidays, the holiday traffic data for the (T + 1)-th cycle can be determined. This process can be specifically referred to the following formula 8:

[0105]

[0106] Here, the parameters in the above formula 8 are explained: Represents the holiday traffic data for the (T + 1)-th cycle; Represents the predicted traffic data for the i-th holiday; D i Represents a time interval before and after the i-th holiday, Represents that if T + 1 belongs to the time interval indicated by D i The value is 1 if T + 1 belongs to the time interval indicated by D, and the value is 0 if T + 1 does not belong to the time interval indicated by D i The value is 0 if T + 1 does not belong to the time interval indicated by D.

[0107] S507. Determine the total traffic data for the (T + 1)-th cycle according to the trend traffic data for the (T + 1)-th cycle, the seasonal traffic data for the (T + 1)-th cycle, and the holiday traffic data for the (T + 1)-th cycle.

[0108] Determining the total traffic data for the (T + 1)-th cycle according to the trend traffic data for the (T + 1)-th cycle, the seasonal traffic data for the (T + 1)-th cycle, and the holiday traffic data for the (T + 1)-th cycle may include: adding the trend traffic data for the (T + 1)-th cycle, the seasonal traffic data for the (T + 1)-th cycle, and the holiday traffic data for the (T + 1)-th cycle to obtain the total traffic data for the (T + 1)-th cycle. The process of determining the total traffic data for the (T + 1)-th cycle according to the trend traffic data for the (T + 1)-th cycle, the seasonal traffic data for the (T + 1)-th cycle, and the holiday traffic data for the (T + 1)-th cycle can be referred to the following formula 9:

[0109]

[0110] The following explains the parameters in the above formula 9: Represents the total traffic data for the (T + 1)-th cycle; Represents the trend traffic data for the (T + 1)-th cycle; Represents the seasonal traffic data for the (T + 1)-th cycle; Represents the holiday traffic data for the (T + 1)-th cycle.

[0111] In an embodiment of the present application, when it is necessary to perform traffic prediction on the traffic usage of a target object in the (T + 1)-th period, the traffic feature sequence of the target object and the overall traffic sequence of the target object can be obtained. Then, the total traffic sequence can be subjected to sequence decomposition processing to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence. Then, traffic prediction can be performed based on the traffic feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence to obtain the total traffic data for the (T + 1)-th period. Among them, the traffic feature sequence may include traffic feature data for n + 1 periods before the (T + 1)-th period, and the traffic feature data can be used to reflect the traffic usage characteristics of the target object in the target device in the corresponding period; the total traffic sequence may include total traffic data for n + 1 periods before the (T + 1)-th period, and the total traffic data can be used to indicate the total traffic consumed by the target object in the target device in the corresponding period. It is not difficult to see from the above that using the traffic feature sequence to participate in traffic prediction can make the traffic prediction result more in line with the traffic usage characteristics of the target object in the target device and improve the accuracy of the traffic prediction result; by performing sequence decomposition processing on the total traffic sequence, traffic prediction can be performed from different perspectives, further improving the accuracy of the traffic prediction result.

[0112] The method of the embodiment of the present application is described in detail above. To facilitate better implementation of the above solution of the embodiment of the present application, correspondingly, a device of the embodiment of the present application is provided below.

[0113] Please refer to Figure 6 , Figure 6 FIG. is a schematic structural diagram of a traffic-based data processing device provided by an embodiment of the present application. The traffic-based data processing device can be set in the computer device provided by the embodiment of the present application. The traffic-based data processing device can be a computer program (including program code) running in the computer device. The computer device can be the target device or server mentioned above. The traffic-based data processing device can be used to execute Figure 4 or Figure 5 The corresponding steps in the method embodiment shown. Please refer to Figure 6 , the traffic-based data processing device may include the following units:

[0114] An acquisition unit 601, configured to acquire the traffic feature sequence of the target object; the traffic feature sequence includes traffic feature data for n + 1 periods before the (T + 1)-th period, and the traffic feature data is used to reflect the traffic usage characteristics of the target object in the target device in the corresponding period, where n is a positive integer; and, acquire the total traffic sequence of the target object; the total traffic sequence includes total traffic data for n + 1 periods before the (T + 1)-th period, and the total traffic data is used to indicate the total traffic consumed by the target object in the target device in the corresponding period;

[0115] A processing unit 602 is configured to perform sequence decomposition processing on the total traffic sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence; and perform traffic prediction based on the traffic feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence to obtain the total traffic data for the (T + 1)-th period.

[0116] In one implementation, the traffic feature data for any period includes at least one of the following: the traffic data consumed by the target object in the application running on the target device, the remaining traffic data of the target object in the target device, the traffic purchase data of the target object in the target device, the traffic billing data for the target object, the traffic data consumed by the target object using the target device on different holidays, and the traffic data consumed by the target object using the target device in different seasons.

[0117] In one implementation, when the processing unit 602 is configured to perform traffic prediction based on the traffic feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence to obtain the total traffic data for the (T + 1)-th period, it is specifically configured to perform the following steps:

[0118] Perform trend traffic prediction on the (T + 1)-th period based on the traffic feature sequence and the trend item sequence to obtain the trend traffic data for the (T + 1)-th period; perform seasonal traffic prediction on the (T + 1)-th period based on the seasonal item sequence to obtain the seasonal traffic data for the (T + 1)-th period; perform holiday traffic prediction on the (T + 1)-th period based on the holiday item sequence to obtain the holiday traffic data for the (T + 1)-th period; and determine the total traffic data for the (T + 1)-th period based on the trend traffic data for the (T + 1)-th period, the seasonal traffic data for the (T + 1)-th period, and the holiday traffic data for the (T + 1)-th period.

[0119] In one implementation, the trend item sequence includes the trend traffic data for each of the n + 1 periods. When the processing unit 602 is configured to perform trend traffic prediction on the (T + 1)-th period based on the traffic feature sequence and the trend item sequence to obtain the trend traffic data for the (T + 1)-th period, it is specifically configured to perform the following steps:

[0120] Determine a feature prediction sequence in the traffic feature sequence, where the feature prediction sequence includes the traffic feature data for a total of n periods starting from the (T - n + 1)-th period to the T-th period in the traffic feature sequence; and call a target trend traffic prediction model to perform trend traffic prediction on the (T + 1)-th period based on the feature prediction sequence and the trend item sequence to obtain the trend traffic data for the (T + 1)-th period.

[0121] In one implementation, the processing unit 602 is further configured to perform the following steps:

[0122] Obtain an initial trend traffic prediction model; construct a connection of the traffic feature sequences according to the object identifier of the target object to obtain a trend training sequence, where the trend training sequence includes the trend training data of each of the n+1 cycles; determine a training sample sequence and a test sample sequence in the trend training sequence; train the initial traffic prediction model according to the trend item sequence, the training sample sequence, and the test sample sequence, so as to obtain a target trend traffic prediction model.

[0123] In one implementation, the seasonal item sequence includes the seasonal traffic data of each of the n+1 cycles; when the processing unit 602 is used to perform seasonal traffic prediction on the (T+1)-th cycle according to the seasonal item sequence to obtain the seasonal traffic data of the (T+1)-th cycle, it is specifically used to perform the following steps:

[0124] Determine a seasonal prediction sequence in the seasonal item sequence, where the seasonal prediction sequence includes the seasonal traffic data of a total of n cycles starting from the (T-n+1)-th cycle to the T-th cycle in the seasonal item sequence; call the target seasonal traffic prediction model to perform seasonal traffic prediction on the (T+1)-th cycle according to the seasonal prediction sequence to obtain the seasonal traffic data of the (T+1)-th cycle.

[0125] In one implementation, the processing unit 602 is further used to perform the following steps:

[0126] Obtain an initial seasonal traffic prediction model; determine a seasonal training sequence in the seasonal item sequence, where the seasonal training sequence includes the seasonal traffic data of a total of n cycles starting from the (T-n)-th cycle to the (T-1)-th cycle in the seasonal item sequence; determine a training sample sequence and a test sample sequence from the seasonal training sequence; train the initial seasonal traffic model according to the training sample sequence and the test sample sequence, so as to obtain a target seasonal traffic prediction model.

[0127] In one implementation, the holiday item sequence includes the holiday traffic data of each of the n+1 cycles. When the processing unit 602 is used to perform holiday traffic prediction on the (T+1)-th cycle according to the holiday item sequence to obtain the holiday traffic data of the (T+1)-th cycle, it is specifically used to perform the following steps:

[0128] Call the holiday traffic prediction model to obtain the holiday data sequence corresponding to the holiday traffic data of each of the n+1 cycles. The holiday data sequence includes the traffic data of M holidays in the corresponding cycle, where M is a positive integer; determine the predicted traffic data of each of the M holidays. The predicted traffic data of the i-th holiday among the M holidays is the average value of the traffic data of the i-th holiday in each of the n+1 cycles, where i is a positive integer less than or equal to M; determine the holiday traffic data of the (T+1)-th cycle according to the predicted traffic data of each of the M holidays.

[0129] In one implementation, when the processing unit 602 is used to perform sequence decomposition processing on the total traffic sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence, it is specifically used to perform the following steps:

[0130] Perform a moving average process on the total traffic sequence to obtain a trend item sequence, where the trend item sequence includes the trend traffic data for each period in n + 1 periods; determine the seasonal item sequence based on the total traffic sequence and the trend item sequence; determine the holiday item sequence based on the total traffic sequence, the trend item sequence, and the seasonal item sequence.

[0131] In one implementation, when the processing unit 602 is used to determine the seasonal item sequence based on the total traffic sequence and the trend item sequence, it is specifically used to perform the following steps:

[0132] Determine a detrended item sequence based on the total traffic sequence and the trend item sequence; the detrended item sequence includes the detrended traffic data for each period in n + 1 periods, and the detrended traffic data for any one period is determined based on the total traffic data for the corresponding period and the trend traffic data for the corresponding period; perform an averaging process on the detrended item sequence according to seasons to obtain a seasonal item sequence, where the seasonal item sequence includes the seasonal traffic data for each period in n + 1 periods.

[0133] In one implementation, when the processing unit 602 is used to determine the holiday item sequence based on the total traffic sequence, the trend item sequence, and the seasonal item sequence, it is specifically used to perform the following steps:

[0134] Determine a detrended and deseasonalized item sequence based on the total traffic sequence, the trend item sequence, and the seasonal item sequence; the detrended and deseasonalized item sequence includes the detrended and deseasonalized traffic data for each period in n + 1 periods, and the detrended and deseasonalized traffic data for any one period is determined based on the total traffic data for the corresponding period, the trend traffic data for the corresponding period, and the seasonal traffic data for the corresponding period; perform an averaging process on the detrended and deseasonalized item sequence according to holidays to obtain a holiday item sequence, where the holiday item sequence includes the holiday traffic data for each period in n + 1 periods.

[0135] According to another embodiment of the present application, Figure 6Each unit in the flow-based data processing device shown can be separately or wholly combined into one or several other units to form, or a certain one (or some) of the units can be further split into multiple smaller units with finer functions to form, which can achieve the same operations without affecting the realization of the technical effects of the embodiments of the present application. The above units are divided based on logical functions. In practical applications, the function of one unit can also be realized by multiple units, or the functions of multiple units can be realized by one unit. In other embodiments of the present application, the flow-based data processing device may also include other units. In practical applications, these functions can also be assisted by other units and can be realized through the cooperation of multiple units.

[0136] According to another embodiment of the present application, it can be achieved by running a computer program (including program code) capable of executing the respective steps involved in the corresponding method shown, such as Figure 4 or Figure 5 shown, on a general computing device such as a computer including processing elements and storage elements such as a central processing unit (CPU), a random access storage medium (RAM), and a read-only storage medium (ROM), to construct the flow-based data processing device shown in Figure 6 and to implement the flow-based data processing method of the embodiments of the present application. The computer program can be recorded on, for example, a computer-readable storage medium, loaded into the above computing device through the computer-readable storage medium, and run therein.

[0137] In the embodiments of the present application, when it is necessary to perform flow prediction on the flow usage of the target object in the (T + 1)-th period, the flow feature sequence of the target object and the overall flow sequence of the target object can be obtained. Then, the total flow sequence can be subjected to sequence decomposition processing to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence. Then, based on the flow feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence, flow prediction can be performed to obtain the total flow data in the (T + 1)-th period. Among them, the flow feature sequence may include the flow feature data in n + 1 periods before the (T + 1)-th period, and the flow feature data can be used to reflect the flow usage characteristics of the target object in the target device in the corresponding period; the total flow sequence may include the total flow data in n + 1 periods before the (T + 1)-th period, and the total flow data can be used to indicate the total flow consumed by the target object in the target device in the corresponding period. It is not difficult to see from the above content that using the flow feature sequence to participate in flow prediction can make the flow prediction result more in line with the flow usage characteristics of the target object in the target device and improve the accuracy of the flow prediction result; by performing sequence decomposition processing on the total flow sequence, flow prediction can be performed from different perspectives, further improving the accuracy of the flow prediction result.

[0138] Based on the above method and apparatus embodiments, an embodiment of the present application provides a computer device. Please refer to Figure 7 , Figure 7 which is a schematic structural diagram of a computer device provided by an embodiment of the present application. Figure 7 The computer device shown at least includes a processor 701, an input interface 702, an output interface 703, and a computer-readable storage medium 704. Among them, the processor 701, the input interface 702, the output interface 703, and the computer-readable storage medium 704 can be connected by a bus or other means.

[0139] The input interface 702 can be used to obtain traffic feature sequences, total traffic sequences, etc.; the output interface 703 can be used to output the predicted total traffic data for the (T + 1)-th period, etc.

[0140] The computer-readable storage medium 704 can be stored in the memory of the computer device. The computer-readable storage medium 704 is used to store a computer program, and the computer program includes computer instructions. The processor 701 is used to execute the program instructions stored in the computer-readable storage medium 704. The processor 701 (or CPU (Central Processing Unit)) is the computing core and control core of the computer device, and is adapted to implement one or more computer instructions, specifically adapted to load and execute one or more computer instructions to implement the corresponding method flow or corresponding function.

[0141] An embodiment of the present application also provides a computer-readable storage medium (Memory). The computer-readable storage medium is a memory device in the computer device, and is used to store programs and data. It can be understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device, and of course can also include the extended storage medium supported by the computer device. The computer-readable storage medium provides a storage space, and the operating system of the computer device is stored in this storage space. And, one or more computer instructions suitable for being loaded and executed by the processor are also stored in this storage space. These computer instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory, or a non-volatile memory (Non-Volatile Memory), such as at least one disk memory; optionally, it can also be at least one computer-readable storage medium located far from the aforementioned processor.

[0142] The computer device can be the aforementioned target device or server, and can be loaded and executed by the processor 701 with one or more computer instructions stored in the computer-readable storage medium 704 to implement the above-mentioned related Figure 4 or Figure 5The corresponding steps of the traffic-based data processing method shown. In a specific implementation, the computer instructions in the computer-readable storage medium 704 are loaded and executed by the processor 701 to perform the following steps:

[0143] Obtain the traffic feature sequence of the target object; the traffic feature sequence includes the traffic feature data of n + 1 cycles before the (T + 1)-th cycle, and the traffic feature data is used to reflect the traffic usage characteristics of the target object in the target device in the corresponding cycle, where n is a positive integer; and, obtain the total traffic sequence of the target object; the total traffic sequence includes the total traffic data of n + 1 cycles before the (T + 1)-th cycle, and the total traffic data is used to indicate the total traffic consumed by the target object in the target device in the corresponding cycle; perform sequence decomposition processing on the total traffic sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence; and, perform traffic prediction based on the traffic feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence to obtain the total traffic data of the (T + 1)-th cycle.

[0144] In one implementation, the traffic feature data of any cycle includes at least one of the following: the traffic data consumed by the target object in the application running on the target device, the remaining traffic data of the target object in the target device, the traffic purchase data of the target object in the target device, the traffic billing data for the target object, the traffic data consumed by the target object using the target device on different holidays, and the traffic data consumed by the target object using the target device in different seasons.

[0145] In one implementation, when the computer instructions in the computer-readable storage medium 704 are loaded and executed by the processor 701 to perform traffic prediction based on the traffic feature sequence, the trend item sequence, the seasonal item sequence, and the holiday item sequence to obtain the total traffic data of the (T + 1)-th cycle, it is specifically used to perform the following steps:

[0146] Perform trend traffic prediction on the (T + 1)-th cycle according to the traffic feature sequence and the trend item sequence to obtain the trend traffic data of the (T + 1)-th cycle; perform seasonal traffic prediction on the (T + 1)-th cycle according to the seasonal item sequence to obtain the seasonal traffic data of the (T + 1)-th cycle; perform holiday traffic prediction on the (T + 1)-th cycle according to the holiday item sequence to obtain the holiday traffic data of the (T + 1)-th cycle; determine the total traffic data of the (T + 1)-th cycle according to the trend traffic data of the (T + 1)-th cycle, the seasonal traffic data of the (T + 1)-th cycle, and the holiday traffic data of the (T + 1)-th cycle.

[0147] In one implementation, the trend item sequence includes the trend traffic data of each of the n + 1 periods; when the computer instructions in the computer-readable storage medium 704 are loaded and executed by the processor 701 to perform trend traffic prediction for the (T + 1)-th period according to the traffic feature sequence and the trend item sequence to obtain the trend traffic data of the (T + 1)-th period, it is specifically used to perform the following steps:

[0148] Determine a feature prediction sequence in the traffic feature sequence, where the feature prediction sequence includes the traffic feature data of a total of n periods starting from the (T - n + 1)-th period to the end of the T-th period in the traffic feature sequence; call the target trend traffic prediction model to perform trend traffic prediction for the (T + 1)-th period according to the feature prediction sequence and the trend item sequence to obtain the trend traffic data of the (T + 1)-th period.

[0149] In one implementation, the computer instructions in the computer-readable storage medium 704 are loaded and executed by the processor 701 and are further used to perform the following steps:

[0150] Obtain an initial trend traffic prediction model; perform connection construction on the traffic feature sequence according to the object identifier of the target object to obtain a trend training sequence, where the trend training sequence includes the trend training data of each of the n + 1 periods; determine a training sample sequence and a test sample sequence in the trend training sequence; train the initial traffic prediction model according to the trend item sequence, the training sample sequence, and the test sample sequence to obtain the target trend traffic prediction model.

[0151] In one implementation, the season item sequence includes the seasonal traffic data of each of the n + 1 periods; when the computer instructions in the computer-readable storage medium 704 are loaded and executed by the processor 701 to perform seasonal traffic prediction for the (T + 1)-th period according to the season item sequence to obtain the seasonal traffic data of the (T + 1)-th period, it is specifically used to perform the following steps:

[0152] Determine a season prediction sequence in the season item sequence, where the season prediction sequence includes the seasonal traffic data of a total of n periods starting from the (T - n + 1)-th period to the end of the T-th period in the season item sequence; call the target season traffic prediction model to perform seasonal traffic prediction for the (T + 1)-th period according to the season prediction sequence to obtain the seasonal traffic data of the (T + 1)-th period.

[0153] In one implementation, the computer instructions in the computer-readable storage medium 704 are loaded and executed by the processor 701 and are further used to perform the following steps:

[0154] Obtain an initial seasonal flow prediction model; determine a seasonal training sequence in the seasonal item sequence, where the seasonal training sequence includes the seasonal flow data for a total of n cycles starting from the (T - n)-th cycle to the end of the (T - 1)-th cycle in the seasonal item sequence; determine a training sample sequence and a test sample sequence from the seasonal training sequence; train the initial seasonal flow model according to the training sample sequence and the test sample sequence to obtain a target seasonal flow prediction model.

[0155] In one implementation, the holiday item sequence includes the holiday flow data for each cycle in n + 1 cycles. When the computer instructions in the computer-readable storage medium 704 are loaded and executed by the processor 701 to perform holiday flow prediction for the (T + 1)-th cycle according to the holiday item sequence to obtain the holiday flow data for the (T + 1)-th cycle, it is specifically used to perform the following steps:

[0156] Call the holiday flow prediction model to obtain a holiday data sequence corresponding to the holiday flow data for each cycle in n + 1 cycles, where the holiday data sequence includes the flow data for M holidays in the corresponding cycle, and M is a positive integer; determine the predicted flow data for each of the M holidays, and the predicted flow data for the i-th holiday among the M holidays is the average of the flow data for the i-th holiday in each cycle in n + 1 cycles, where i is a positive integer less than or equal to M; determine the holiday flow data for the (T + 1)-th cycle according to the predicted flow data for each of the M holidays.

[0157] In one implementation, when the computer instructions in the computer-readable storage medium 704 are loaded and executed by the processor 701 to perform sequence decomposition processing on the total flow sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence, it is specifically used to perform the following steps:

[0158] Perform a moving average process on the total flow sequence to obtain a trend item sequence, where the trend item sequence includes the trend flow data for each cycle in n + 1 cycles; determine the seasonal item sequence according to the total flow sequence and the trend item sequence; determine the holiday item sequence according to the total flow sequence, the trend item sequence, and the seasonal item sequence.

[0159] In one implementation, when the computer instructions in the computer-readable storage medium 704 are loaded and executed by the processor 701 to determine the seasonal item sequence according to the total flow sequence and the trend item sequence, it is specifically used to perform the following steps:

[0160] Determine a detrended term sequence based on the total flow sequence and the trend term sequence; the detrended term sequence includes the detrended flow data for each of the n + 1 cycles, and the detrended flow data for any one cycle is determined based on the total flow data for the corresponding cycle and the trend flow data for the corresponding cycle; perform an averaging process on the detrended term sequence according to seasons to obtain a seasonal term sequence, and the seasonal term sequence includes the seasonal flow data for each of the n + 1 cycles.

[0161] In one implementation, when the computer instructions in the computer-readable storage medium 704 are loaded and executed by the processor 701 to determine the holiday term sequence according to the total flow sequence, the trend term sequence, and the seasonal term sequence, it is specifically used to execute the following steps:

[0162] Determine a detrended and deseasonalized term sequence based on the total flow sequence, the trend term sequence, and the seasonal term sequence; the detrended and deseasonalized term sequence includes the detrended and deseasonalized flow data for each of the n + 1 cycles, and the detrended and deseasonalized flow data for any one cycle is determined based on the total flow data for the corresponding cycle, the trend flow data for the corresponding cycle, and the seasonal flow data for the corresponding cycle; perform an averaging process on the detrended and deseasonalized term sequence according to holidays to obtain a holiday term sequence, and the holiday term sequence includes the holiday flow data for each of the n + 1 cycles.

[0163] In the embodiments of the present application, when it is necessary to perform flow prediction on the flow usage of the target object in the (T + 1)-th cycle, the flow feature sequence of the target object and the overall flow sequence of the target object can be obtained, and then the total flow sequence can be subjected to sequence decomposition processing to obtain a trend term sequence, a seasonal term sequence, and a holiday term sequence. Then, flow prediction can be performed according to the flow feature sequence, the trend term sequence, the seasonal term sequence, and the holiday term sequence to obtain the total flow data for the (T + 1)-th cycle. Among them, the flow feature sequence may include the flow feature data for n + 1 cycles before the (T + 1)-th cycle, and the flow feature data can be used to reflect the flow usage characteristics of the target object in the target device under the corresponding cycle; the total flow sequence may include the total flow data for n + 1 cycles before the (T + 1)-th cycle, and the total flow data can be used to indicate the total flow consumed by the target object in the target device under the corresponding cycle. It is not difficult to see from the above that using the flow feature sequence to participate in the flow prediction can make the flow prediction result more in line with the flow usage characteristics of the target object in the target device and improve the accuracy of the flow prediction result; by performing sequence decomposition processing on the total flow sequence, flow prediction can be performed from different perspectives, further improving the accuracy of the flow prediction result.

[0164] According to one aspect of the present application, there is provided a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the traffic-based data processing method provided in the above various alternative manners.

[0165] As described above, the above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A data processing method based on traffic, characterized in that The method includes: Obtaining a traffic feature sequence of a target object; the traffic feature sequence includes traffic feature data of n + 1 cycles before the (T + 1)-th cycle, and the traffic feature data is used to reflect the traffic usage characteristics of the target object in a target device in the corresponding cycle, where n is a positive integer; Obtaining a total traffic sequence of the target object; the total traffic sequence includes total traffic data of the n + 1 cycles before the (T + 1)-th cycle, and the total traffic data is used to indicate the total traffic consumed by the target object in the target device in the corresponding cycle; Performing sequence decomposition processing on the total traffic sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence; Performing trend traffic prediction on the (T + 1)-th cycle according to the traffic feature sequence and the trend item sequence to obtain trend traffic data of the (T + 1)-th cycle; Performing seasonal traffic prediction on the (T + 1)-th cycle according to the seasonal item sequence to obtain seasonal traffic data of the (T + 1)-th cycle; Performing holiday traffic prediction on the (T + 1)-th cycle according to the holiday item sequence to obtain holiday traffic data of the (T + 1)-th cycle; Determining the total traffic data of the (T + 1)-th cycle according to the trend traffic data of the (T + 1)-th cycle, the seasonal traffic data of the (T + 1)-th cycle, and the holiday traffic data of the (T + 1)-th cycle, and the total traffic data of the (T + 1)-th cycle conforms to the traffic usage characteristics of the target object.

2. The method according to claim 1, characterized in that, The traffic feature data of any cycle includes at least one of the following: Traffic data consumed by the target object in the application programs running on the target device, remaining traffic data of the target object in the target device, traffic purchase data of the target object in the target device, traffic billing data for the target object, traffic data consumed by the target object using the target device on different holidays, and traffic data consumed by the target object using the target device in different seasons.

3. The method according to claim 1, characterized in that The trend item sequence includes trend traffic data of each of the n + 1 cycles; the performing trend traffic prediction on the (T + 1)-th cycle according to the traffic feature sequence and the trend item sequence to obtain trend traffic data of the (T + 1)-th cycle includes: Determining a feature prediction sequence in the traffic feature sequence, where the feature prediction sequence includes traffic feature data of a total of n cycles starting from the (T - n + 1)-th cycle to the end of the T-th cycle in the traffic feature sequence; Invoking a target trend traffic prediction model to perform trend traffic prediction on the (T + 1)-th cycle according to the feature prediction sequence and the trend item sequence to obtain trend traffic data of the (T + 1)-th cycle.

4. The method according to claim 3, wherein The method further includes: Obtaining an initial trend traffic prediction model; Performing connection construction on the traffic feature sequence according to the object identifier of the target object to obtain a trend training sequence, where the trend training sequence includes trend training data of each of the n + 1 cycles; Determining a training sample sequence and a test sample sequence in the trend training sequence; Train the initial traffic prediction model according to the trend item sequence, the training sample sequence, and the test sample sequence, so as to obtain the target trend traffic prediction model.

5. The method according to claim 1, characterized in that The seasonal item sequence includes the seasonal traffic data of each of the n + 1 cycles; the predicting the seasonal traffic of the (T + 1)-th cycle according to the seasonal item sequence to obtain the seasonal traffic data of the (T + 1)-th cycle includes: Determine a seasonal prediction sequence in the seasonal item sequence, where the seasonal prediction sequence includes the seasonal traffic data of a total of n cycles starting from the (T - n + 1)-th cycle to the end of the T-th cycle in the seasonal item sequence; Call the target seasonal traffic prediction model to predict the seasonal traffic of the (T + 1)-th cycle according to the seasonal prediction sequence, so as to obtain the seasonal traffic data of the (T + 1)-th cycle.

6. The method according to claim 5, characterized in that, The method further includes: Obtain an initial seasonal traffic prediction model; Determine a seasonal training sequence in the seasonal item sequence, where the seasonal training sequence includes the seasonal traffic data of a total of n cycles starting from the (T - n)-th cycle to the end of the (T - 1)-th cycle in the seasonal item sequence; Determine a training sample sequence and a test sample sequence from the seasonal training sequence; Train the initial seasonal traffic model according to the training sample sequence and the test sample sequence, so as to obtain the target seasonal traffic prediction model.

7. The method according to claim 1, characterized in that, The holiday item sequence includes the holiday traffic data of each of the n + 1 cycles, and the predicting the holiday traffic of the (T + 1)-th cycle according to the holiday item sequence to obtain the holiday traffic data of the (T + 1)-th cycle includes: Call the holiday traffic prediction model to obtain a holiday data sequence corresponding to the holiday traffic data of each of the n + 1 cycles, where the holiday data sequence includes the traffic data of M holidays in the corresponding cycle, and M is a positive integer; Determine the predicted traffic data of each of the M holidays, where the predicted traffic data of the i-th holiday among the M holidays is the average value of the traffic data of the i-th holiday in each of the n + 1 cycles, and i is a positive integer less than or equal to M; Determine the holiday traffic data of the (T + 1)-th cycle according to the predicted traffic data of each of the M holidays.

8. The method according to claim 1, wherein The performing sequence decomposition processing on the total traffic sequence to obtain a trend item sequence, a seasonal item sequence, and a holiday item sequence includes: Perform a moving average process on the total traffic sequence to obtain the trend item sequence, where the trend item sequence includes the trend traffic data of each of the n + 1 cycles; Determine the seasonal item sequence according to the total traffic sequence and the trend item sequence; Determine the holiday item sequence according to the total traffic sequence, the trend item sequence, and the seasonal item sequence.

9. The method according to claim 8, wherein The determining the seasonal item sequence according to the total traffic sequence and the trend item sequence includes: Determine a detrended term sequence according to the total flow sequence and the trend term sequence; the detrended term sequence includes detrended flow data for each of the n + 1 cycles, and the detrended flow data for any one cycle is determined according to the total flow data for the corresponding cycle and the trend flow data for the corresponding cycle; Perform an averaging process on the detrended term sequence by season to obtain the seasonal term sequence, where the seasonal term sequence includes seasonal flow data for each of the n + 1 cycles.

10. The method according to claim 9, wherein The determining the holiday term sequence according to the total flow sequence, the trend term sequence, and the seasonal term sequence includes: Determine a detrended and deseasonalized term sequence according to the total flow sequence, the trend term sequence, and the seasonal term sequence; the detrended and deseasonalized term sequence includes detrended and deseasonalized flow data for each of the n + 1 cycles, and the detrended and deseasonalized flow data for any one cycle is determined according to the total flow data for the corresponding cycle, the trend flow data for the corresponding cycle, and the seasonal flow data for the corresponding cycle; Perform an averaging process on the detrended and deseasonalized term sequence by holiday to obtain the holiday term sequence, where the holiday term sequence includes holiday flow data for each of the n + 1 cycles.

11. A data processing device based on traffic, characterized in that, The apparatus includes: An acquisition unit, configured to acquire a flow feature sequence of a target object; the flow feature sequence includes flow feature data for n + 1 cycles before the (T + 1)-th cycle, and the flow feature data is used to reflect the flow usage feature of the target object in the target device under the corresponding cycle, where n is a positive integer; The acquisition unit is configured to acquire the total flow sequence of the target object; the total flow sequence includes total flow data for the n + 1 cycles before the (T + 1)-th cycle, and the total flow data is used to indicate the total flow consumed by the target object in the target device under the corresponding cycle; A processing unit, configured to perform sequence decomposition processing on the total flow sequence to obtain a trend term sequence, a seasonal term sequence, and a holiday term sequence; The processing unit is configured to perform trend flow prediction on the (T + 1)-th cycle according to the flow feature sequence and the trend term sequence to obtain trend flow data for the (T + 1)-th cycle; perform seasonal flow prediction on the (T + 1)-th cycle according to the seasonal term sequence to obtain seasonal flow data for the (T + 1)-th cycle; perform holiday flow prediction on the (T + 1)-th cycle according to the holiday term sequence to obtain holiday flow data for the (T + 1)-th cycle; and determine the total flow data for the (T + 1)-th cycle according to the trend flow data for the (T + 1)-th cycle, the seasonal flow data for the (T + 1)-th cycle, and the holiday flow data for the (T + 1)-th cycle.

12. A computer device, characterized in that, The computer device includes: A processor, adapted to implement a computer program; A computer-readable storage medium storing a computer program, the computer program being adapted to be loaded and executed by the processor to perform the flow-based data processing method according to any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and the computer program is adapted to be loaded and executed by a processor to perform the traffic-based data processing method according to any one of claims 1 to 10.

14. A computer program product, characterized in that, The computer program product includes computer instructions, and when the computer instructions are executed by a processor, the traffic-based data processing method according to any one of claims 1 to 10 is implemented.

Images