Traffic volume prediction model construction method and device, and computer readable storage medium

Abstract

This application provides a method, apparatus, and computer-readable storage medium for constructing a call volume prediction model. The method includes: obtaining the total call volume before the transfer from a first call center and the total call volume before the transfer from a second call center; obtaining the transfer ratio of the first call center within a first time period, where the transfer ratio is the ratio of the target call volume to the total call volume before the transfer from the first call center; determining call transfer information from the first call center to the second call center within the first time period based on the total call volume before the transfer from the first call center and the transfer ratio; summing the total call volume before the transfer from the second call center and the call transfer information to reconstruct the first historical call volume information of the second call center, obtaining the reconstructed target historical call volume data; and constructing a target call volume prediction model based on the target historical call volume data. The embodiments of this application result in higher prediction accuracy for the reconstructed call volume prediction model.

Description

Technical Field

[0001] This application relates to the field of information processing technology, specifically to a method, apparatus, and computer-readable storage medium for constructing a traffic volume prediction model. Background Technology

[0002] With the rapid development of information technology, after-sales service exists in all walks of life, and the most common form of after-sales service is currently provided directly by telephone. Many aspects of people's lives and work can now be handled and processed online, such as money transfers, payments, and shopping. As the internet becomes more widely used, the demand for customer service is increasing, leading to a surge in customer service calls.

[0003] However, because the scope of call volume changes with operational strategies (e.g., human call volume may decrease due to the introduction of intelligent voice systems or shift in human call volume), there is an inconsistency between the distribution of historical and future call volume data, failing to meet the data consistency assumption of traditional prediction methods. Without optimization to address this issue, the prediction model will provide results that significantly deviate from actual call volume, resulting in low prediction accuracy.

[0004] Therefore, how to effectively improve the prediction accuracy of call volume prediction models is a technical problem that urgently needs to be solved in the field of information processing technology. Summary of the Invention

[0005] This application provides a method, apparatus, and computer-readable storage medium for constructing a traffic volume prediction model, aiming to solve the problem of how to effectively improve the prediction accuracy of the traffic volume prediction model.

[0006] On the one hand, this application provides a method for constructing a traffic volume prediction model, the method comprising:

[0007] Obtain the total call volume before the first call center transfer and the total call volume before the second call center transfer;

[0008] Obtain the transfer percentage value of the first call center within the first time period, wherein the transfer percentage value is the ratio of the target call volume to the total call volume of the first call center before the transfer;

[0009] Based on the total call volume before the transfer to the first call center and the transfer percentage, determine the call transfer information from the first call center to the second call center within the first time period;

[0010] The total call volume before the transfer to the second call center and the call transfer information are summed to reconstruct the first historical call volume information of the second call center, so as to obtain the reconstructed target historical call volume data.

[0011] Based on the target's historical call data, a target call volume prediction model is constructed.

[0012] In one possible implementation of this application, the first time period includes a second time period and a third time period, wherein the second time period is the length of the time range within the first time period in which call transfer information can be statistically analyzed, and the third time period is the length of the time range within the first time period in which call transfer information cannot be statistically analyzed.

[0013] The step of obtaining the transfer rate of the first call center within the first time period includes:

[0014] Obtain the first transfer percentage value of the first call center during the second time period;

[0015] Calculate the estimated transfer rate of the first call center during the third time period using the first transfer rate value;

[0016] Based on the first transfer percentage value and the estimated transfer percentage value, the transfer percentage value of the first call center within the first time period is determined.

[0017] In one possible implementation of this application, the step of constructing a target call volume prediction model based on the target historical call volume data includes:

[0018] Training sample data is determined from the target historical call data;

[0019] Based on the training sample data, a basic call volume prediction model is established.

[0020] Adjust the parameters in the primary call volume prediction model to obtain the target call volume prediction model.

[0021] In one possible implementation of this application, adjusting the parameters in the primary traffic volume prediction model to obtain the target traffic volume prediction model includes:

[0022] Obtain the second historical call traffic information within the fourth time period;

[0023] The second historical call information is preprocessed to generate the third historical call information;

[0024] Obtain pre-generated model candidate parameters;

[0025] Based on the third historical call volume information and the model candidate parameters, the parameters in the primary call volume prediction model are adjusted to obtain the target call volume prediction model.

[0026] In one possible implementation of this application, the step of adjusting the parameters in the primary traffic volume prediction model based on the third historical traffic information and the model candidate parameters to obtain the target traffic volume prediction model includes:

[0027] Based on the third historical call information and the candidate parameters of the model, the optimal parameters of the model are determined;

[0028] Based on the optimal parameters, the parameters in the primary call volume prediction model are adjusted to obtain the target call volume prediction model.

[0029] In one possible implementation of this application, determining the optimal model parameters based on the third historical call information and the model candidate parameters includes:

[0030] Based on the aforementioned third historical call traffic information, an intermediate call traffic prediction model is established;

[0031] The traffic volume in the fifth time period is predicted using the intermediate traffic volume prediction model to obtain the first traffic volume prediction data.

[0032] Based on the first call volume prediction data, calculate the rolling prediction error data within the fourth time period;

[0033] Based on the rolling prediction error data and the candidate parameters of the model, the optimal parameters of the model are determined.

[0034] On the other hand, this application provides a traffic volume prediction method, the method comprising:

[0035] Obtain historical call data for the sixth time period;

[0036] The historical call volume data is input into the target call volume prediction model to predict the call volume in the seventh time period, wherein the seventh time period is after the sixth time period, and the target call volume prediction model is the target call volume prediction model described in the above implementation.

[0037] On the other hand, this application provides a traffic volume prediction model construction apparatus, the apparatus comprising:

[0038] The first acquisition unit is used to acquire the total call volume before the first call center transfer and the total call volume before the second call center transfer.

[0039] The second acquisition unit is used to acquire the transfer ratio of the first call center within a first time period, wherein the transfer ratio is the ratio of the target call volume to the total call volume of the first call center before the transfer.

[0040] The first determining unit is used to determine the call transfer information from the first call center to the second call center within the first time period based on the total call volume before the transfer at the first call center and the transfer ratio value.

[0041] The first reconstruction unit is used to sum the total call volume before the transfer of the second call center and the call transfer information to reconstruct the first historical call information of the second call center and obtain the reconstructed target historical call data.

[0042] The first construction unit is used to construct a target call volume prediction model based on the target historical call data.

[0043] In one possible implementation of this application, the first time period includes a second time period and a third time period, wherein the second time period is the length of the time range within the first time period in which call transfer information can be statistically analyzed, and the third time period is the length of the time range within the first time period in which call transfer information cannot be statistically analyzed.

[0044] The second acquisition unit is specifically used for:

[0045] Obtain the first transfer percentage value of the first call center during the second time period;

[0046] Calculate the estimated transfer rate of the first call center during the third time period using the first transfer rate value;

[0047] Based on the first transfer percentage value and the estimated transfer percentage value, the transfer percentage value of the first call center within the first time period is determined.

[0048] In one possible implementation of this application, the first construction unit specifically includes:

[0049] The second determining unit is used to determine training sample data from the target historical call data;

[0050] The first establishment unit is used to establish a primary traffic volume prediction model based on the training sample data.

[0051] The first adjustment unit is used to adjust the parameters in the primary traffic volume prediction model to obtain the target traffic volume prediction model.

[0052] In one possible implementation of this application, the first adjustment unit specifically includes:

[0053] The third acquisition unit is used to acquire the second historical call traffic information within the fourth time period;

[0054] The first preprocessing unit is used to preprocess the second historical call information to generate the third historical call information;

[0055] The fourth acquisition unit is used to acquire pre-generated model candidate parameters;

[0056] The second adjustment unit is used to adjust the parameters in the primary traffic volume prediction model based on the third historical traffic information and the model candidate parameters to obtain the target traffic volume prediction model.

[0057] In one possible implementation of this application, the second adjustment unit specifically includes:

[0058] The third determining unit is used to determine the optimal parameters of the model based on the third historical call information and the model candidate parameters;

[0059] The third adjustment unit is used to adjust the parameters in the primary traffic volume prediction model based on the optimal parameters to obtain the target traffic volume prediction model.

[0060] In one possible implementation of this application, the third determining unit is specifically used for:

[0061] Based on the aforementioned third historical call traffic information, an intermediate call traffic prediction model is established;

[0062] The traffic volume in the fifth time period is predicted using the intermediate traffic volume prediction model to obtain the first traffic volume prediction data.

[0063] Based on the first call volume prediction data, calculate the rolling prediction error data within the fourth time period;

[0064] Based on the rolling prediction error data and the candidate parameters of the model, the optimal parameters of the model are determined.

[0065] On the other hand, this application also provides a computer device, the computer device comprising:

[0066] One or more processors;

[0067] Memory; and

[0068] One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the processor to implement the traffic volume prediction model construction method.

[0069] On the other hand, this application also provides a computer-readable storage medium having a computer program stored thereon, the computer program being loaded by a processor to execute the steps in the traffic volume prediction model construction method.

[0070] This application addresses the issue of decreased call volume in call volume prediction models caused by business changes by first obtaining the total call volume before the transfer from the first call center and the total call volume before the transfer from the second call center; then, it obtains the transfer percentage of the first call center within a first time period, which is the ratio of the target call volume to the total call volume before the transfer from the first call center; finally, based on the total call volume before the transfer from the first call center and the transfer percentage, it determines the call transfer information from the first call center to the second call center within the first time period; next, it sums the total call volume before the transfer from the second call center and the call transfer information to reconstruct the first historical call volume information of the second call center, obtaining the reconstructed target historical call volume data; and finally, it constructs a target call volume prediction model based on the target historical call volume data. Therefore, it can repair historical call volume data according to business changes and automatically and quickly adapt to the latest data distribution without requiring additional manual adjustments, thus solving the problem of decreased prediction accuracy of call volume prediction models caused by business changes, resulting in higher prediction accuracy for the reconstructed call volume prediction model. Attached Figure Description

[0071] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0072] Figure 1 This is a schematic diagram of a traffic volume prediction model construction system provided in an embodiment of this application.

[0073] Figure 2 This is a schematic flowchart of an embodiment of the traffic volume prediction model construction method provided in this application.

[0074] Figure 3 This is a schematic flowchart of an embodiment of step 202 provided in this application;

[0075] Figure 4 This is a schematic flowchart of an embodiment of step 205 provided in this application;

[0076] Figure 5 This is a schematic flowchart of an embodiment of step 403 provided in this application;

[0077] Figure 6 This is a schematic flowchart of an embodiment of step 504 provided in this application;

[0078] Figure 7 This is a schematic flowchart of an embodiment of step 601 provided in this application;

[0079] Figure 8 This is a schematic flowchart of an embodiment of the traffic volume prediction method provided in this application.

[0080] Figure 9 This is a schematic diagram of an embodiment of the traffic volume prediction model construction device provided in this application.

[0081] Figure 10 This is a schematic diagram of an embodiment of the computer device provided in this application. Detailed Implementation

[0082] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0083] In the description of this application, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified.

[0084] In this application, the term "exemplary" is used to mean "used as an example, illustration, or description." Any embodiment described as "exemplary" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that this application can be made without using these specific details. In other instances, well-known structures and processes are not described in detail to avoid obscuring the description of this application with unnecessary detail. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0085] This application provides a method, apparatus, and computer-readable storage medium for constructing a traffic volume prediction model, which will be described in detail below.

[0086] like Figure 1 As shown, Figure 1 This is a schematic diagram of a traffic volume prediction model building system provided in an embodiment of this application. The system may include multiple terminals 100 and a server 200, which are connected via a network. The server 200 integrates a traffic volume prediction model building device, such as... Figure 1 In the server, terminal 100 can access server 200.

[0087] In this embodiment, server 200 is mainly used to obtain the total call volume before the transfer from the first call center and the total call volume before the transfer from the second call center; obtain the transfer ratio of the first call center within a first time period, where the transfer ratio is the ratio of the target call volume to the total call volume before the transfer from the first call center; determine the call transfer information from the first call center to the second call center within the first time period based on the total call volume before the transfer from the first call center and the transfer ratio; sum the total call volume before the transfer from the second call center and the call transfer information to reconstruct the first historical call information of the second call center, thereby obtaining the reconstructed target historical call data; and construct a target call volume prediction model based on the target historical call data.

[0088] In this embodiment, the server 200 can be a standalone server, a server network, or a server cluster. For example, the server 200 described in this embodiment includes, but is not limited to, computers, network terminals, single network servers, sets of multiple network servers, or cloud servers composed of multiple servers. The cloud server is composed of a large number of computers or network servers based on cloud computing. In this embodiment, communication between the server and the terminal can be achieved through any communication method, including but not limited to, mobile communication based on the 3rd Generation Partnership Project (3GPP), Long Term Evolution (LTE), and Worldwide Interoperability for Microwave Access (WiMAX), or computer network communication based on the TCP / IP Protocol Suite (TCP / IP) and User Datagram Protocol (UDP).

[0089] It is understood that the terminal 100 used in this application embodiment can be a device that includes both receiving and transmitting hardware, i.e., a device with receiving and transmitting hardware capable of performing bidirectional communication on a bidirectional communication link. Such a terminal may include: cellular or other communication devices having a single-line display, a multi-line display, or a cellular or other communication device without a multi-line display. Specifically, the terminal 100 may be a desktop terminal or a mobile terminal, and the terminal 100 may also be one of a mobile phone, tablet computer, laptop computer, etc.

[0090] Those skilled in the art will understand that Figure 1 The application environment shown is merely one application scenario of the solution in this application and does not constitute a limitation on the application scenario of the solution in this application. Other application environments may include more than one application scenario. Figure 1 The number of more or fewer terminals, or server network connections shown, for example Figure 1 Only one server and two terminals are shown in the diagram. It is understood that this traffic prediction model building system may also include one or more other servers, and / or one or more terminals connected to the server network, which is not limited here.

[0091] In addition, such as Figure 1 As shown, the traffic volume prediction model building system may also include a memory 300 for storing data, such as storing user traffic volume prediction data and traffic volume prediction model building data, for example, traffic volume prediction model building data during the operation of the traffic volume prediction model building system.

[0092] It should be noted that, Figure 1 The schematic diagram of the traffic volume prediction model construction system shown is merely an example. The traffic volume prediction model construction system and scenario described in this application are for the purpose of more clearly illustrating the technical solutions of this application, and do not constitute a limitation on the technical solutions provided in this application. As those skilled in the art will know, with the evolution of the traffic volume prediction model construction system and the emergence of new business scenarios, the technical solutions provided in this application are also applicable to similar technical problems.

[0093] Next, we will introduce the traffic volume prediction model construction method provided in the embodiments of this application.

[0094] In this embodiment of the call volume prediction model construction method, the call volume prediction model construction device is used as the execution subject. For simplification and ease of description, the execution subject will be omitted in subsequent method embodiments. The call volume prediction model construction device is applied to a computer device. The method includes: obtaining call transfer information from a first call center to a second call center within a first time period; reconstructing the first historical call information of the second call center based on the call transfer information to obtain the reconstructed target historical call data; and constructing a target call volume prediction model based on the target historical call data.

[0095] Please see Figures 2 to 10 , Figure 2 This is a schematic flowchart of an embodiment of the traffic volume prediction model construction method provided in this application. The traffic volume prediction model construction method includes steps 201 to 205:

[0096] 201. Obtain the total call volume before the transfer of the first call center and the total call volume before the transfer of the second call center.

[0097] A call center is a service organization located in a relatively centralized location, staffed by a group of service personnel. It typically utilizes computer communication technology to handle telephone inquiries from businesses and customers, possessing the ability to handle a large volume of calls simultaneously. It also features caller ID, automatically assigning calls to personnel with the appropriate skills, and recording and storing all call information. Call centers, also known as customer service centers, originated in the 1930s, initially transferring user calls to an answering machine or specialist. Subsequently, as the number of transferred calls and responses increased, interactive voice response systems were established. These systems allow automated "operators" to answer and handle some common customer questions. Traditionally, a call center refers to a call response center primarily handling telephone inquiries, providing various telephone response services to customers.

[0098] For some large enterprises, whose service scope may cover the entire country, a single call center can no longer meet the actual needs. Furthermore, due to differences in customs and habits across different regions of the country, services require corresponding adaptation and adjustments. Therefore, they often establish call centers across the country according to certain regional divisions. Thus, even a single enterprise may have multiple call centers. For example, call center A handles non-complaint calls in Central and South China, call center B handles complaint calls in Central and South China, call center C handles non-complaint calls in North and East China, and call center D handles complaint calls in North and East China. However, due to changes in the enterprise's operational strategy, call traffic may be transferred between call centers. For instance, non-complaint calls in the Central China region handled by one call center may be transferred to call center C.

[0099] Each call center stores call volume information for a preset time period. This call volume information can be stored in a corresponding database. Therefore, the total call volume before the transfer of the first call center and the total call volume before the transfer of the second call center can be obtained from the database.

[0100] 202. Obtain the transfer rate of the first call center within the first time period.

[0101] The transfer percentage is the ratio of the target call volume to the total call volume before the transfer to the first call center.

[0102] 203. Based on the total call volume and transfer percentage of the first call center before the transfer, determine the call transfer information from the first call center to the second call center within the first time period.

[0103] 204. Sum the total call volume and call transfer information of the second call center before the transfer to reconstruct the first historical call volume information of the second call center and obtain the reconstructed target historical call volume data.

[0104] The reconstructed target historical call data includes the original historical call data of the second call center during the first time period and the call transfer information of the target call volume transferred from the first call center to the second call center during the first time period.

[0105] Because machine learning algorithms (such as LightGBM) require that the historical data (i.e., the prediction sample or the target historical call traffic data) and the data to be predicted (future call traffic data) be independently and identically distributed, the model will have prediction errors when the historical data and the data to be predicted are not distributed. The reasons for this disparity in data distribution are as described above, including call transfers caused by changes in the call center mapping relationship (i.e., call transfers between call centers), which need to be addressed through historical data reconstruction.

[0106] For ease of understanding, we will refer to the first call center as Call Center A and the second call center as Call Center B. When a portion of the total call volume of Call Center A is about to be transferred to Call Center B, the most direct solution is to add the portion of the historical call data of A that is about to be transferred to B to the historical call data of B.

[0107] For the historical call data of call center B under the new mapping relationship (i.e., call transfer has occurred) ,satisfy:

[0108] (1)

[0109] in, The call data of call center B under the old mapping relationship (i.e., the mapping relationship between the call center and its call data before call transfer occurred). This refers to the call traffic data of call center A under the old mapping relationship. Let t be the percentage of call data to be changed within the total call data at time t. That is, the target call volume in the call transfer information is... .

[0110] 205. Based on the target's historical call volume data, construct a target call volume prediction model.

[0111] This application addresses the issue of decreased call volume in call volume prediction models caused by business changes by first obtaining the total call volume before the transfer from the first call center and the total call volume before the transfer from the second call center; then, it obtains the transfer percentage of the first call center within a first time period, which is the ratio of the target call volume to the total call volume before the transfer from the first call center; finally, based on the total call volume before the transfer from the first call center and the transfer percentage, it determines the call transfer information from the first call center to the second call center within the first time period; next, it sums the total call volume before the transfer from the second call center and the call transfer information to reconstruct the first historical call volume information of the second call center, obtaining the reconstructed target historical call volume data; and finally, it constructs a target call volume prediction model based on the target historical call volume data. Therefore, it can repair historical call volume data according to business changes and automatically and quickly adapt to the latest data distribution without requiring additional manual adjustments, thus solving the problem of decreased prediction accuracy of call volume prediction models caused by business changes, resulting in higher prediction accuracy for the reconstructed call volume prediction model.

[0112] In the embodiments of this application, such as Figure 3 As shown, the first time period includes a second time period and a third time period. The second time period is the length of the time range within the first time period for which call transfer information can be statistically analyzed, and the third time period is the length of the time range within the first time period for which call transfer information cannot be statistically analyzed. Step 202: Obtain the transfer percentage of the first call center within the first time period, specifically including steps 301 to 303:

[0113] 301. Obtain the percentage of first transfers from the first call center during the second time period.

[0114] 302. Calculate the estimated transfer rate of the first call center in the third time period using the first transfer rate value.

[0115] 303. Based on the first transfer percentage value and the estimated transfer percentage value, determine the transfer percentage value of the first call center within the first time period.

[0116] In practice, records of historical mapping changes are often not saved; only call data subject to change can be statistically analyzed within a relatively short timeframe. The time range of all available historical call data. To allow for the statistical analysis of the time range of changes in call traffic data, then when hour, Unknown; and hour, Given that the first time period corresponds to T, and the second time period corresponds to... The third time period corresponds to .

[0117] The first transfer percentage in step 301 is known and can be directly calculated based on the transferred call volume during the second time period and the total call volume before the first call center transfer. Furthermore, to obtain an accurate... It is necessary to within To make an accurate estimate, it is necessary to obtain... Estimate , making Estimated amount of change in call traffic data at that time and Estimated amount of change traffic data The distribution is consistent. To address this, this application employs adversarial verification to determine the consistency of the distribution. The specific steps are as follows:

[0118] 1. Estimated value of the arbitrary percentage Calculate the historical call data under this percentage.

[0119] 2. Construct a historical call sample of length k. Mark it as an obfuscation class, where ;

[0120] 3. Construct a real-world call sample of length k. Mark it as a real class, where ;

[0121] 4. Randomize the samples in steps 2 and 3, and extract feature vectors as training and validation samples using n-fold cross-validation while ensuring that the proportion of the confused class remains unchanged. Construct classifier C and train the classification model to distinguish between the confused class and the true class.

[0122] 5. Iterate through all possible values ​​of the estimated value to obtain a... ,satisfy:

[0123] ;

[0124] in, , These represent the weights of the approximation degree and the degree of confusion in the optimization objective, respectively. Considering that a higher approximation degree is desirable while maximizing the degree of confusion, therefore... Discrimination error Calculated using the following formula:

[0125]

[0126] Where n represents the total number of cross-validations, and s represents the number of samples in the validation set in the j-th cross-validation.

[0127] The optimization objective in step 5 indicates that the optimal proportion estimate is... The generated historical data should be so complex that classifier C cannot correctly distinguish whether the data belongs to newly constructed historical data (confusing class) or recent actual historical data (true class). That is, when the classifier's discrimination error is close to 0.5, it indicates that the classifier has confused the data in determining its category, and at this point, the distribution of historical data and actual data is consistent.

[0128] In this embodiment, a prediction model is constructed based on reconstructed historical data, enabling stable predictions without mean shift in future data distribution. However, in actual operation, data distribution changes frequently, and the call volume mean constantly shifts. Therefore, further improvements to the basic call volume prediction model are needed, such as... Figure 4 As shown, step 205, based on the target historical call traffic data, constructs a target call traffic prediction model, including:

[0129] 401. Determine training sample data from the target's historical call data.

[0130] 402. Based on the training sample data, establish a basic call volume prediction model.

[0131] The training sample data should be feature samples that are strongly correlated with the target date. The input features used in this application are as follows:

[0132] 1. Historical call characteristics of [tk, t-1], where k represents the range of historical data. Specifically, k can be a multiple of 7.

[0133] 2. The characteristics of [tk,t-1] number of items and the predicted number of items show a strong correlation between changes in number of items and changes in call volume.

[0134] 3. Operational indicators, such as transshipment rate, are also strongly correlated with changes in business volume.

[0135] 4. The growth forecast characteristics for special dates, such as the estimated decrease in call volume on New Year's Eve, are obtained through decomposition using the fbProphet model;

[0136] 5. The difference features of the above features include the first difference of the historical n days (n={1,3,7,14,30}) and the second difference of the historical n days.

[0137] 6. The rolling features of the above features are based on different time windows w={3,5,7,14,30}, and the maximum, minimum, average, decayed average, and variance are statistically analyzed within the time window range.

[0138] 7. Date characteristics, including weekday, season, quarter, year, date, and whether it is a holiday;

[0139] Considering the periodicity of date features, preprocessing of date features is required, constructing two auxiliary features as follows:

[0140]

[0141] Where T is the period of the feature, and for the weekday feature, T=7; t is the specific weekday, such as Saturday, t=6.

[0142] The prediction model uses LightGBM as the solution algorithm to calculate the mapping function between the input and the predicted traffic volume. Specifically, historical data samples are divided into training and validation sets according to a certain ratio (e.g., 0.8:0.2) and chronological order. The LightGBM model is trained using the training samples, and the model error is validated on the validation set. By adjusting the model parameters, an optimal model is trained. The input distribution and parameters of the model largely determine the accuracy of the prediction model. The prediction accuracy in this application is defined by relative error percentage (MAPE), as follows:

[0143]

[0144] in, This represents the true value of the i-th sample. This represents the predicted value of the i-th sample, where n is the total number of samples.

[0145] 403. Adjust the parameters in the primary call volume prediction model to obtain the target call volume prediction model.

[0146] In the embodiments of this application, such as Figure 5 As shown, step 403, adjusting the parameters in the primary traffic volume prediction model to obtain the target traffic volume prediction model, includes:

[0147] 501. Obtain the second historical call information within the fourth time period.

[0148] The fourth time period can be the first time period or a period after the first time period. Therefore, the historical call data in the second historical call information can be the target historical call data obtained through the above reconstruction.

[0149] 502. Preprocess the second historical call information to generate the third historical call information.

[0150] Preprocessing can include time-series alignment and moving average padding. It's important to note that, unlike traditional prediction processes, the input vector for constructing the model differs in that, if predictive features (such as future order volumes or holiday characteristics) are involved in the preprocessing, they need to be generated on a rolling basis to simulate a day-by-day prediction process, thus avoiding the leakage of future information.

[0151] 503. Obtain the pre-generated model candidate parameters.

[0152] Bayesian optimization algorithms assume a mapping relationship between the parameter space and the model error. After continuously generating parameters to build a prediction model, this mapping relationship can be fitted based on the parameters and prediction error, with the fitting accuracy continuously improving with the number of iterations. Through the fitted mapping relationship and random sampling probabilities, new parameters can be continuously generated to achieve joint optimization of the mapping relationship and the generated parameters, ultimately converging to the optimal parameter space.

[0153] 504. Based on the third historical call volume information and model candidate parameters, adjust the parameters in the primary call volume prediction model to obtain the target call volume prediction model.

[0154] In steps 501 to 504, a dynamic parameter tuning algorithm based on Bayesian optimization is proposed to address the issue of traffic mean drift. This algorithm avoids the need for manual parameter tuning and allows the model to automatically adapt to new data distributions.

[0155] In the embodiments of this application, such as Figure 6 As shown, step 504, based on the third historical call volume information and model candidate parameters, adjusts the parameters in the primary call volume prediction model to obtain the target call volume prediction model, including:

[0156] 601. Based on the third historical call traffic information and model candidate parameters, determine the optimal parameters of the model.

[0157] 602. Based on the optimal parameters, adjust the parameters in the primary traffic volume prediction model to obtain the target traffic volume prediction model.

[0158] In the embodiments of this application, such as Figure 7 As shown, step 601, based on the third historical call traffic information and model candidate parameters, determines the optimal parameters of the model, including:

[0159] 701. Based on third-party historical call traffic information, establish an intermediate call traffic prediction model.

[0160] 702. The call volume in the fifth time period is predicted using an intermediate call volume prediction model to obtain the first call volume prediction data.

[0161] 703. Based on the first call volume prediction data, calculate the rolling prediction error data for the fourth time period.

[0162] 704. Determine the optimal parameters of the model based on the rolling prediction error data and candidate model parameters.

[0163] In steps 701 and 704, historical call data from the third historical call information is used as input data, model candidate parameters are adopted, and the call volume within the fifth time period is used as output to establish an intermediate call volume prediction model (such as LightGBM) and simulate the prediction process within the preset period range, for example, the length of the fifth time period is 15 days.

[0164] Then, the rolling prediction error within the fourth time period is calculated and weighted according to a certain rule (e.g., the closer to the target date, the greater the weight), to fit recent characteristics so that the prediction model can adapt to the latest data distribution. Finally, the average value of the rolling prediction error within this preset period range is taken as the score of this set of parameters.

[0165] Among all parameters, select the one that best fits the most recent data distribution and has the highest score, with the smaller the error, the higher the score.

[0166] To better implement the traffic volume prediction model construction method in the embodiments of this application, based on the traffic volume prediction model construction method, the embodiments of this application also provide a traffic volume prediction method, such as... Figure 8 As shown, the call volume prediction method specifically includes steps 801 and 802:

[0167] 801. Obtain historical call data for the sixth time period.

[0168] 802. Input historical call volume data into the target call volume prediction model to predict the call volume in the seventh time period.

[0169] The seventh time period is after the sixth time period, and the target call volume prediction model is the target call volume prediction model described in the above embodiments.

[0170] This application achieves more accurate call volume prediction results by using the call volume prediction model reconstructed in the above embodiments.

[0171] To better implement the traffic volume prediction model construction method in the embodiments of this application, based on the traffic volume prediction model construction method, the embodiments of this application also provide a traffic volume prediction model construction device, such as... Figure 9As shown, the traffic volume prediction model construction device 900 includes a first acquisition unit 901, a second acquisition unit 902, a first determination unit 903, a first reconstruction unit 904, and a first construction unit 905.

[0172] The first acquisition unit 901 is used to acquire the total call volume before the first call center transfer and the total call volume before the second call center transfer.

[0173] The second acquisition unit 902 is used to acquire the transfer ratio value of the first call center within a first time period, wherein the transfer ratio value is the ratio of the target call volume to the total call volume of the first call center before the transfer;

[0174] The first determining unit 903 is used to determine the call transfer information from the first call center to the second call center within the first time period based on the total call volume before the transfer at the first call center and the transfer ratio value.

[0175] The first reconstruction unit 904 is used to sum the total call volume before the transfer of the second call center and the call transfer information to reconstruct the first historical call information of the second call center and obtain the reconstructed target historical call data.

[0176] The first construction unit 905 is used to construct a target call volume prediction model based on the target historical call data.

[0177] In this embodiment of the application, the first time period includes a second time period and a third time period. The second time period is the length of the time range within the first time period for which call transfer information can be statistically analyzed, and the third time period is the length of the time range within the first time period for which call transfer information cannot be statistically analyzed.

[0178] The second acquisition unit 902 is specifically used for:

[0179] Obtain the percentage of first transfers from the first call center within the second time period.

[0180] Using the first transfer percentage value, calculate the estimated transfer percentage of the first call center in the third time period.

[0181] Based on the first transfer percentage value and the estimated transfer percentage value, the transfer percentage value of the first call center within the first time period is determined.

[0182] In this embodiment of the application, the first construction unit 905 specifically includes:

[0183] The second determining unit is used to determine training sample data from the target historical call data.

[0184] The first establishment unit is used to establish a basic traffic volume prediction model based on training sample data.

[0185] The first adjustment unit is used to adjust the parameters in the primary traffic volume prediction model to obtain the target traffic volume prediction model.

[0186] In this embodiment of the application, the first adjustment unit specifically includes:

[0187] The third acquisition unit is used to acquire the second historical call information within the fourth time period.

[0188] The first preprocessing unit is used to preprocess the second historical call information to generate the third historical call information.

[0189] The fourth acquisition unit is used to acquire pre-generated model candidate parameters.

[0190] The second adjustment unit is used to adjust the parameters in the primary traffic volume prediction model based on the third historical traffic information and model candidate parameters to obtain the target traffic volume prediction model.

[0191] In this embodiment of the application, the second adjustment unit specifically includes:

[0192] The third determining unit is used to determine the optimal parameters of the model based on the third historical call information and the model candidate parameters.

[0193] The third adjustment unit is used to adjust the parameters in the primary traffic volume prediction model based on the optimal parameters to obtain the target traffic volume prediction model.

[0194] In this embodiment of the application, the third determining unit is specifically used for:

[0195] A mid-level call volume prediction model is established based on third-party historical call traffic information.

[0196] The first volume prediction data was obtained by predicting the volume of calls in the fifth time period using an intermediate volume prediction model.

[0197] Based on the first call volume prediction data, the rolling prediction error data for the fourth time period is calculated.

[0198] Based on the rolling prediction error data and candidate parameters of the model, the optimal parameters of the model are determined.

[0199] This application achieves this by first acquiring the total call volume before the transfer from the first call center and the total call volume before the transfer from the second call center using a first acquisition unit 901; second acquiring the transfer percentage of the first call center within a first time period, where the transfer percentage is the ratio of the target call volume to the total call volume before the transfer from the first call center; then, a first determining unit 903 determines the call transfer information from the first call center to the second call center within the first time period based on the total call volume before the transfer from the first call center and the transfer percentage; then, a first reconstructing unit 904 sums the total call volume before the transfer from the second call center and the call transfer information to reconstruct the first historical call information of the second call center, obtaining the reconstructed target historical call data; finally, a first constructing unit 905 constructs a target call volume prediction model based on the target historical call data. Therefore, it can repair historical call data according to business changes and automatically and quickly adapt to the latest data distribution without additional manual adjustments, thus solving the problem of decreased prediction accuracy of the call volume prediction model caused by business changes, resulting in higher prediction accuracy of the reconstructed call volume prediction model.

[0200] In addition to the methods and apparatus for constructing call volume prediction models described above, embodiments of this application also provide a computer device that integrates any of the call volume prediction model construction apparatuses provided in embodiments of this application. The computer device includes:

[0201] One or more processors;

[0202] Memory; and

[0203] One or more applications, wherein the one or more applications are stored in the memory and configured by the processor to perform the operation of any of the methods described in any of the embodiments of the above-described traffic volume prediction model construction method.

[0204] This application also provides a computer device that integrates any of the call volume prediction model building apparatuses provided in this application. See also... Figure 10 , Figure 10 This is a schematic diagram of the structure of one embodiment of the computer device provided in this application.

[0205] like Figure 10 As shown, it illustrates the structural diagram of the traffic volume prediction model construction device designed in the embodiments of this application. Specifically:

[0206] The traffic volume prediction model building apparatus may include components such as a processor 101 with one or more processing cores, a storage unit 102 with one or more computer-readable storage media, a power supply 103, and an input unit 104. Those skilled in the art will understand that... Figure 10The structure of the traffic volume prediction model building device shown does not constitute a limitation on the traffic volume prediction model building device. It may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein:

[0207] The processor 101 is the control center of the traffic volume prediction model building device. It connects to various parts of the device via various interfaces and lines. By running or executing software programs and / or modules stored in the storage unit 102, and by calling data stored in the storage unit 102, it performs various functions and processes data of the traffic volume prediction model building device, thereby providing overall monitoring of the device. Optionally, the processor 101 may include one or more processing cores; preferably, the processor 101 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 101.

[0208] Storage unit 102 can be used to store software programs and modules. Processor 101 executes various functional applications and data processing by running the software programs and modules stored in storage unit 102. Storage unit 102 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created by using the device based on the traffic volume prediction model. In addition, storage unit 102 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, storage unit 102 may also include a memory controller to provide processor 101 with access to storage unit 102.

[0209] The traffic volume prediction model building device also includes a power supply 103 that supplies power to the various components. Preferably, the power supply 103 can be logically connected to the processor 101 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 103 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0210] The traffic volume prediction model building device may also include an input unit 104, which can be used to receive input digital or character information, and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0211] Although not shown, the traffic volume prediction model building device may also include a display unit, etc., which will not be described in detail here. Specifically, in the embodiments of this application, the processor 101 in the traffic volume prediction model building device loads the executable files corresponding to the processes of one or more applications into the storage unit 102 according to the following instructions, and the processor 101 runs the applications stored in the storage unit 102 to realize various functions, as follows:

[0212] Obtain the total call volume before the transfer from the first call center and the total call volume before the transfer from the second call center; obtain the transfer ratio of the first call center within the first time period, which is the ratio of the target call volume to the total call volume before the transfer from the first call center; based on the total call volume before the transfer from the first call center and the transfer ratio, determine the call transfer information from the first call center to the second call center within the first time period; sum the total call volume before the transfer from the second call center and the call transfer information to reconstruct the first historical call information of the second call center, and obtain the reconstructed target historical call data; based on the target historical call data, construct a target call volume prediction model.

[0213] This application addresses the issue of decreased call volume in call volume prediction models caused by business changes by first obtaining the total call volume before the transfer from the first call center and the total call volume before the transfer from the second call center; then, it obtains the transfer percentage of the first call center within a first time period, which is the ratio of the target call volume to the total call volume before the transfer from the first call center; finally, based on the total call volume before the transfer from the first call center and the transfer percentage, it determines the call transfer information from the first call center to the second call center within the first time period; next, it sums the total call volume before the transfer from the second call center and the call transfer information to reconstruct the first historical call volume information of the second call center, obtaining the reconstructed target historical call volume data; and finally, it constructs a target call volume prediction model based on the target historical call volume data. Therefore, it can repair historical call volume data according to business changes and automatically and quickly adapt to the latest data distribution without requiring additional manual adjustments, thus solving the problem of decreased prediction accuracy of call volume prediction models caused by business changes, resulting in higher prediction accuracy for the reconstructed call volume prediction model.

[0214] Therefore, embodiments of this application provide a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), a disk, or an optical disk, etc. The computer-readable storage medium stores multiple instructions, which can be loaded by a processor to execute the steps in any of the traffic volume prediction model construction methods provided in embodiments of this application. For example, the instructions can execute the following steps:

[0215] Obtain the total call volume before the transfer from the first call center and the total call volume before the transfer from the second call center; obtain the transfer ratio of the first call center within the first time period, which is the ratio of the target call volume to the total call volume before the transfer from the first call center; based on the total call volume before the transfer from the first call center and the transfer ratio, determine the call transfer information from the first call center to the second call center within the first time period; sum the total call volume before the transfer from the second call center and the call transfer information to reconstruct the first historical call information of the second call center, and obtain the reconstructed target historical call data; based on the target historical call data, construct a target call volume prediction model.

[0216] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0217] The above provides a detailed description of a traffic volume prediction model construction method, apparatus, and computer-readable storage medium provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for constructing a traffic volume prediction model, characterized in that, The method includes: Obtain the total call volume before the first call center transfer and the total call volume before the second call center transfer; Obtain the transfer percentage value of the first call center within the first time period, wherein the transfer percentage value is the ratio of the target call volume to the total call volume of the first call center before the transfer; Based on the total call volume before the transfer to the first call center and the transfer percentage, determine the call transfer information from the first call center to the second call center within the first time period; The total call volume before the transfer to the second call center and the call transfer information are summed to reconstruct the first historical call volume information of the second call center, so as to obtain the reconstructed target historical call volume data. Based on the target's historical call data, a target call volume prediction model is constructed.

2. The method for constructing a traffic volume prediction model according to claim 1, characterized in that, The first time period includes a second time period and a third time period. The second time period is the length of the time range within the first time period for which call transfer information can be statistically analyzed, and the third time period is the length of the time range within the first time period for which call transfer information cannot be statistically analyzed. The step of obtaining the transfer rate of the first call center within the first time period includes: Obtain the first transfer percentage value of the first call center during the second time period; Calculate the estimated transfer rate of the first call center during the third time period using the first transfer rate value; Based on the first transfer percentage value and the estimated transfer percentage value, the transfer percentage value of the first call center within the first time period is determined.

3. The method for constructing a traffic volume prediction model according to claim 1, characterized in that, The step of constructing a target call volume prediction model based on the target historical call data includes: Training sample data is determined from the target historical call data; Based on the training sample data, a basic call volume prediction model is established; Adjust the parameters in the primary call volume prediction model to obtain the target call volume prediction model.

4. The method for constructing a traffic volume prediction model according to claim 3, characterized in that, The step of adjusting the parameters in the primary traffic volume prediction model to obtain the target traffic volume prediction model includes: Obtain the second historical call traffic information within the fourth time period; The second historical call information is preprocessed to generate the third historical call information; Obtain pre-generated model candidate parameters; Based on the third historical call volume information and the model candidate parameters, the parameters in the primary call volume prediction model are adjusted to obtain the target call volume prediction model.

5. The method for constructing a traffic volume prediction model according to claim 4, characterized in that, The step of adjusting the parameters in the primary call volume prediction model based on the third historical call volume information and the model candidate parameters to obtain the target call volume prediction model includes: Based on the third historical call information and the candidate parameters of the model, the optimal parameters of the model are determined; Based on the optimal parameters, the parameters in the primary call volume prediction model are adjusted to obtain the target call volume prediction model.

6. The method for constructing a traffic volume prediction model according to claim 5, characterized in that, The step of determining the optimal parameters of the model based on the third historical call information and the candidate parameters of the model includes: Based on the aforementioned third historical call traffic information, an intermediate call traffic prediction model is established; The traffic volume in the fifth time period is predicted using the intermediate traffic volume prediction model to obtain the first traffic volume prediction data. Based on the first call volume prediction data, calculate the rolling prediction error data within the fourth time period; Based on the rolling prediction error data and the candidate parameters of the model, the optimal parameters of the model are determined.

7. A method for predicting call volume, characterized in that, The method includes: Obtain historical call data for the sixth time period; The historical call volume data is input into the target call volume prediction model to predict the call volume in the seventh time period, wherein the seventh time period is after the sixth time period, and the target call volume prediction model is the target call volume prediction model as described in any one of claims 1 to 6.

8. A device for constructing a traffic volume prediction model, characterized in that, The device includes: The first acquisition unit is used to acquire the total call volume before the first call center transfer and the total call volume before the second call center transfer. The second acquisition unit is used to acquire the transfer ratio of the first call center within a first time period, wherein the transfer ratio is the ratio of the target call volume to the total call volume of the first call center before the transfer. The first determining unit is used to determine the call transfer information from the first call center to the second call center within the first time period based on the total call volume before the transfer at the first call center and the transfer ratio value. The first reconstruction unit is used to sum the total call volume before the transfer of the second call center and the call transfer information to reconstruct the first historical call information of the second call center and obtain the reconstructed target historical call data. The first construction unit is used to construct a target call volume prediction model based on the target historical call data.

9. A computer device, characterized in that, The computer device includes: One or more processors; Memory; and One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the processor to implement the traffic volume prediction model construction method of any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which is loaded by a processor to execute the steps in the traffic volume prediction model construction method according to any one of claims 1 to 6.

Images