Service request quantity prediction method and related equipment

Abstract

The invention provides a service request quantity prediction method and related equipment, and relates to the technical field of computers. The method comprises the steps that interaction parameter information is obtained in response to interaction operation, and the interaction parameter information comprises data parameter information, feature adjustment information, model type information and mode adjustment information; obtaining target data related to the service request quantity according to the data parameter information, and performing feature processing on the target data to obtain initial data features; performing feature adjustment on the initial data features according to the feature adjustment information to obtain target data features; determining a target prediction model according to the model type information; and based on the target prediction model, predicting the service request quantity according to the mode adjustment information and the target data features to obtain a target service request quantity. The method can adapt to prediction requirements in different scenes, and ensures the accuracy of service request quantity prediction, so that the customer service staff can be more effectively scheduled.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a service request volume prediction method, a service request volume prediction device, a computer-readable storage medium, an electronic device, and a computer program product. Background Art

[0002] With the rapid development of the logistics and express delivery industry, more and more logistics companies have established their own customer service centers, aiming to understand and handle various problems encountered by users through communication methods such as telephone.

[0003] Traffic forecasting is a key part of customer service center operations management. For example, the traffic volume for the next week or several weeks is forecasted every day, and customer service personnel are arranged reasonably based on this forecast. The accuracy of traffic forecasting affects the matching degree between resources and demand, which in turn affects the answering efficiency, service quality and customer experience.

[0004] The existing traffic volume forecasting methods mainly include manual forecasting methods and intelligent forecasting methods. Among them, the manual forecasting method relies on manual experience and has low forecasting accuracy; the intelligent forecasting method relies on the quality of input data. When the data is abnormal or has great uncertainty, the forecast results will be seriously biased.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0006] The purpose of the present disclosure is to provide a service request volume prediction method, a service request volume prediction device, a computer-readable storage medium, an electronic device and a computer program product, so as to overcome the above problems or at least partially solve the above problems.

[0007] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0008] According to one aspect of the present disclosure, a method for predicting a service request volume is provided, comprising: obtaining interaction parameter information in response to an interaction operation; the interaction parameter information comprises data parameter information, feature adjustment information, model type information and mode adjustment information; obtaining target data related to the service request volume according to the data parameter information, performing feature processing on the target data to obtain initial data features; performing feature adjustment on the initial data features according to the feature adjustment information to obtain target data features; determining a target prediction model according to the model type information; based on the target prediction model, predicting the service request volume according to the mode adjustment information and the target data features to obtain a target service request volume.

[0009] In some embodiments of the present disclosure, the target data includes target historical data, target predicted data and target event data, and the initial data features include initial basic features, initial predicted features and initial event features; the feature processing of the target data to obtain the initial data features includes: feature processing of the target historical data to obtain the initial basic features; feature processing of the target predicted data to obtain the initial predicted features; feature processing of the target event data to obtain the initial event features.

[0010] In some embodiments of the present disclosure, the feature adjustment information is used to indicate adjustment of data features; the feature adjustment information includes feature addition information and / or feature modification information.

[0011] In some embodiments of the present disclosure, the feature adjustment of the initial data feature according to the feature adjustment information to obtain the target data feature includes: in response to the feature adjustment information including the feature addition information, feature processing is performed according to the feature addition information to obtain a newly added feature, and the newly added feature is added to the initial data feature to obtain the target data feature; in response to the feature adjustment information including the feature modification information, the initial data feature is modified according to the feature modification information to obtain the target data feature.

[0012] In some embodiments of the present disclosure, the target historical data includes target historical service request volume data and target historical waybill volume data; the target prediction data includes target waybill volume prediction data; and the target event data refers to target external factor data that affects the service request volume.

[0013] In some embodiments of the present disclosure, the model type information is used to indicate the selected model type; determining the target prediction model based on the model type information includes: selecting the target prediction model from a model library based on the model type information; wherein the model library includes statistical models and artificial intelligence models; the statistical model includes a time series model and a regression model, the time series model is obtained by training based on historical data of service request volume, and the regression model is obtained by modeling based on the correlation between the service request volume and the waybill volume; the artificial intelligence model includes a machine learning model and a deep learning model.

[0014] In some embodiments of the present disclosure, the pattern adjustment information is used to indicate an adjustment to a model prediction result; based on the target prediction model, the service request amount is predicted according to the pattern adjustment information and the target data characteristics to obtain a target service request amount, including: based on the target prediction model, the service request amount is predicted according to the target data characteristics to obtain an initial service request amount; and the initial service request amount is adjusted according to the pattern adjustment information to obtain the target service request amount.

[0015] In some embodiments of the present disclosure, the pattern adjustment information is used to indicate the reconstruction of existing data features; based on the target prediction model, the service request volume is predicted according to the pattern adjustment information and the target data features to obtain the target service request volume, including: if the target prediction model is a machine learning model obtained by training based on the existing data features, the existing data features are reconstructed according to the pattern adjustment information to obtain reconstructed data features; the target prediction model is trained according to the reconstructed data features to obtain a new target prediction model; the service request volume is predicted based on the new target prediction model to obtain the target service request volume.

[0016] In some embodiments of the present disclosure, there are multiple target prediction models; the method also includes: obtaining the target service request volume of each target prediction model; weighting the target service request volume of each target prediction model to obtain the weighted service request volume, and determining the weighted service request volume as the prediction result.

[0017] According to another aspect of the present disclosure, a service request volume prediction device is provided, including: an interaction module, used to obtain interaction parameter information in response to an interaction operation; the interaction parameter information includes data parameter information, feature adjustment information, model type information and mode adjustment information; a data acquisition module, used to obtain target data related to the service request volume according to the data parameter information; a feature processing module, used to perform feature processing on the target data to obtain initial data features; a feature adjustment module, used to perform feature adjustment on the initial data features according to the feature adjustment information to obtain target data features; a model determination module, used to determine a target prediction model according to the model type information; a prediction module, used to predict the service request volume based on the target prediction model, according to the mode adjustment information and the target data features, to obtain a target service request volume.

[0018] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned service request volume prediction method is implemented.

[0019] According to another aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the above-mentioned service request volume prediction method by executing the executable instructions.

[0020] According to another aspect of the present disclosure, a computer program product is provided, including a computer program / instruction, wherein when the computer program / instruction is executed by a processor, the above-mentioned service request volume prediction method is implemented.

[0021] The service request volume prediction method provided by the embodiment of the present disclosure interacts with the user to obtain interactive parameter information such as data parameter information, feature adjustment information, model type information and mode adjustment information, and then uses the obtained interactive parameter information to predict the service request volume. It can adapt to the prediction needs in different scenarios and ensure the accuracy of the service request volume prediction, so as to more effectively schedule customer service personnel; select appropriate target data according to the data parameter information, introduce feature adjustment and mode adjustment mechanisms, which can avoid the influence of data anomalies or uncertainties on the prediction results, and make the prediction results accurate and reliable; obtain the target prediction model suitable for the business scenario through the model type information, and predict the service request volume based on the target prediction model, so as to further improve the prediction accuracy.

[0022] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0024] Figure 1 A schematic diagram of a system architecture to which the service request volume prediction method disclosed in the present invention can be applied is shown;

[0025] Figure 2 is a flow chart of a method for predicting service request volume according to an embodiment of the present disclosure;

[0026] Figure 3 is a framework diagram of a service request volume prediction system according to an embodiment of the present disclosure;

[0027] Figure 4 is an interaction diagram between a user and a service request volume prediction system according to an embodiment of the present disclosure;

[0028] Figure 5 is a comparison chart of prediction accuracy between the existing method and the service request volume prediction method of the embodiment of the present disclosure;

[0029] Figure 6 is a block diagram of a service request volume prediction device according to an embodiment of the present disclosure; and

[0030] Figure 7 It is a structural block diagram of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0031] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein.

[0033] The following example embodiments can be implemented in a variety of forms and should not be construed as being limited to the examples set forth herein; on the contrary, these embodiments are provided so that the present disclosure will be more comprehensive and complete, and the concepts of the example embodiments are fully conveyed to those skilled in the art. The described features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced while omitting one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known technical solutions are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

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

[0035] It should be noted that the collection, collection, updating, analysis, processing, use, transmission, storage and other aspects of user personal information involved in the technical solution of this disclosure are in compliance with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain the security of user personal information, network security and national security.

[0036] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings.

[0037] Figure 1 A schematic diagram of a system architecture to which the service request volume prediction method disclosed in the present invention can be applied is shown.

[0038] like Figure 1 As shown, the system architecture may include a server 101, a network 102, a terminal device 103, a terminal device 104, and a terminal device 105. The network 102 is used to provide a medium for a communication link between the terminal device 103, the terminal device 104, or the terminal device 105 and the server 101. The network 102 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0039] The server 101 may be a server that provides various services, such as a background management server that provides support for devices operated by users using the terminal device 103, the terminal device 104, or the terminal device 105. The background management server may analyze and process the received request and other data, and feed back the processing results to the terminal device 103, the terminal device 104, or the terminal device 105.

[0040] Terminal device 103, terminal device 104 and terminal device 105 can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a wearable smart device, a virtual reality device, an augmented reality device, etc., but are not limited thereto.

[0041] It should be understood that Figure 1 The number of terminal devices 103, terminal devices 104, terminal devices 105, networks 102 and servers 101 are merely illustrative. Server 101 may be a physical server, a server cluster consisting of multiple servers, or a cloud server. Depending on actual needs, any number of terminal devices, networks and servers may be provided.

[0042] Figure 2 It is a flowchart of a method for predicting service request volume according to an embodiment of the present disclosure. Figure 2 The method provided in the embodiment can be executed by any electronic device, such as the above Figure 1 The terminal device in Figure 1 Servers in, or Figure 1 The terminal device and the server in the embodiment are jointly executed, but the present disclosure does not limit this. Figure 2 The service request volume prediction method provided by the embodiment of the present disclosure may include the following steps.

[0043] Step S210, in response to the interactive operation, obtaining interactive parameter information; wherein the interactive parameter information includes data parameter information, feature adjustment information, model type information and mode adjustment information.

[0044] Through user interaction operations, such as clicking, inputting, selecting, etc., the interaction parameter information required for predicting the service request volume is obtained. For the customer service center, the service request volume refers to the call volume. In the embodiment of the present disclosure, the interaction parameter information includes data parameter information, feature adjustment information, model type information, and mode adjustment information. Through these interaction parameter information, the call volume is predicted.

[0045] The data parameter information is used to define the data used to predict the service request volume, including the data range (such as time period), the business type corresponding to the data, the data format, the location of the data source, etc.

[0046] Feature adjustment information refers to information that adjusts the acquired data features in order to better adapt to business scenarios, including feature addition and feature modification.

[0047] The model type information is used to specify the model used to predict the service request volume, such as linear regression, decision tree, random forest, neural network and other models.

[0048] Mode adjustment information refers to the adjustment strategy for the model prediction results or internal feature structure in order to optimize model performance or make the model prediction results more in line with business scenario requirements. Mode adjustment information includes adjustment of model prediction results and feature reconstruction. Among them, adjustment of model prediction results is applicable to all models, and feature reconstruction is applicable to machine learning models based on feature engineering.

[0049] Step S220, acquiring target data related to the service request quantity according to the data parameter information, performing feature processing on the target data, and obtaining initial data features.

[0050] Target data refers to data used for service request volume prediction. After obtaining data parameter information, target data can be obtained according to the data parameter information. For example, if the data parameter information includes a data range (such as a time period), data within the data range is selected for service request volume prediction; for another example, if the data parameter information includes a business type corresponding to the data, data corresponding to the business type is selected for service request volume prediction.

[0051] After obtaining the target data, feature processing can be performed on the target data, such as data cleaning (removing missing values ​​and outliers), data conversion (such as date format conversion), feature extraction, etc., to obtain initial data features. These initial data features will serve as the basis for subsequent model predictions.

[0052] Step S230, adjusting the initial data features according to the feature adjustment information to obtain target data features.

[0053] Feature adjustment information includes feature addition and feature modification. Feature adjustment information aims to optimize the initial data features to make them more suitable for the subsequent model prediction process.

[0054] Adding new features means adding new features to the initial data features. New features may come from further processing of the target data, such as adding the 7-day average feature of the target data; or they may come from external data, such as adding a feature description of the promotion degree of a promotion activity when the business scenario includes promotion activities.

[0055] Feature modification refers to the adjustment of initial data features. Feature modification may be based on the need for data cleaning, such as processing missing values, outliers, etc., or it may be to improve the performance of features in the model, such as through normalization, standardization, encoding, etc.

[0056] By adding and modifying features, we can find target data features that are more suitable for model prediction. These target data features not only contain useful information in the target data, but also improve the data quality and the model's predictive ability through adjustment and optimization, thereby obtaining more accurate and reliable prediction results.

[0057] Step S240: determining the target prediction model according to the model type information.

[0058] Different business scenarios are suitable for different models. For example, there are multiple models. For the prediction of telephone hotline volume in area A, model a has a better prediction effect, while for the prediction of online call volume in area B, model b has a better prediction effect. Users can interact with each model based on experience or the business scenarios applicable to each model to obtain model type information, and then select an appropriate model as the target prediction model based on the model type information.

[0059] Step S250 , based on the target prediction model, the service request volume is predicted according to the mode adjustment information and the target data characteristics to obtain the target service request volume.

[0060] In this step, the target prediction model determined in step S240 is used to predict the service request volume in combination with the mode adjustment information in step S210 and the target data features obtained in step S230 to obtain the target service request volume.

[0061] In the disclosed embodiment, in the process of service request volume prediction, interaction with users to obtain interaction parameter information is a very important step. User interaction can enhance the flexibility and dynamism of the prediction process and make the prediction results more in line with actual business needs.

[0062] The interactive operations that users may perform include: (1) Users can view and filter the target data used for prediction in real time, including historical service request volume data, historical waybill volume data, and data on other external influencing factors (such as holidays, promotions, weather, etc.), and can also adjust the weights of different data according to business needs; (2) Users can propose new feature combinations or feature optimizations based on business scenario requirements, and view the impact of these adjustments on the model prediction results in real time; (3) Users can view the comparison between the prediction results and the actual service request volume in real time, evaluate the accuracy and reliability of the model business, and adjust the model's prediction results or data processing process in real time based on the evaluation results.

[0063] In addition to the above situations, the user's interactive operations may also include other situations, such as adjusting model type information, which is not limited in the embodiments of the present disclosure.

[0064] The service request volume prediction method provided by the embodiment of the present disclosure interacts with the user to obtain interactive parameter information such as data parameter information, feature adjustment information, model type information and mode adjustment information, and then uses the obtained interactive parameter information to predict the service request volume. It can adapt to the prediction needs in different scenarios and ensure the accuracy of the service request volume prediction, so as to more effectively schedule customer service personnel; select appropriate target data according to the data parameter information, introduce feature adjustment and mode adjustment mechanisms, which can avoid the influence of data anomalies or uncertainties on the prediction results, and make the prediction results accurate and reliable; obtain the target prediction model suitable for the business scenario through the model type information, and predict the service request volume based on the target prediction model, so as to further improve the prediction accuracy.

[0065] Figure 3 is a framework diagram of a service request volume prediction system according to an embodiment of the present disclosure. Figure 3 As shown in the figure, the system architecture includes data layer, feature layer, parameter adjustment layer, model layer and fusion layer.

[0066] The data layer is used to represent the basic data source required for service request volume prediction. The feature layer is used to process and process features based on different data to prepare for model input. The parameter adjustment layer is used to add and modify data features, which can directly or indirectly affect the prediction output of the model.

[0067] The model layer is used to provide a variety of models, mainly including statistical models and artificial intelligence models. Statistical models can also be called black box models or transparent models, and artificial intelligence models can also be called white box models or non-transparent models. For different business scenarios, predictions can be made through different models. The fusion layer is used to operate the prediction results, such as selection operations and weighted operations. Since different business types and different prediction scenarios may be suitable for different prediction models, the prediction results can be selected and combined according to the prediction objectives.

[0068] Reference Figure 3 ,The data layer includes internal data and external data. Internal data refers to the data generated by internal factors, and external data refers to the data generated by external factors.

[0069] Internal data can be understood as data that has a direct impact on the prediction of service request volume, including historical data of service request volume. There are multiple customer service centers, each of which is responsible for the customer service business of the corresponding area, which can include telephone hotline business and online business. From a functional perspective, telephone hotline business mainly provides services through telephone, and online business provides services through online text and voice communication. The data format of historical data of service request volume is shown in Table 1.

[0070] Table 1 Service request volume data table of customer service center

[0071] Customer Service Center area Business Type time Service Request Volume Customer Service Center A Telephone 2024-01-01 10000 Customer Service Center 2 B Telephone 2024-01-02 20000 Customer Service Center 3 C Online 2024-01-01 10000 Customer Service Center 4 D Online 2024-01-02 20000

[0072] In Table 1, Customer Service Center 1 corresponds to Area A, and the number of its telephone hotlines on January 1, 2024 is 10,000; Customer Service Center 2 corresponds to Area B, and the number of its telephone hotlines on January 2, 2024 is 20,000; Customer Service Center 3 corresponds to Area C, and the number of its online services on January 1, 2024 is 10,000; Customer Service Center 4 corresponds to Area D, and the number of its online services on January 2, 2024 is 20,000.

[0073] It should be understood that Table 1 is only an example of the data format of the service request volume. In the embodiments of the present disclosure, the customer service business may include other customer service businesses, such as email customer service business and on-site customer service business, in addition to telephone hotline business and online business, and the embodiments of the present disclosure do not limit this. In addition, each customer service center may correspond to all businesses in a region, including telephone hotline business and online business, or a customer service center may be set to undertake a certain business in all regions. Of course, it can also be in other forms, and the embodiments of the present disclosure do not limit this.

[0074] External data can be understood as data that has an indirect impact on the forecast of service request volume. Since the number of waybills is strongly correlated with the number of service requests, when the number of waybills is large, the number of service requests will increase accordingly. Therefore, external data includes data related to the number of waybills, such as historical data on the number of waybills and forecast data on the number of waybills. For the customer service center, the number of service requests is strongly correlated with the number of waybills, and an increase in the number of waybills will affect the increase in the number of service requests. In addition to analyzing the historical data on the number of waybills, the forecast data on the number of waybills, that is, the forecast data on the number of waybills for a period of time in the future, will also indirectly affect the number of service requests. The data format of the historical data and forecast data on the number of waybills is shown in Table 2.

[0075] Table 2 Historical data and forecast data of waybill volume

[0076]

[0077]

[0078] In Table 2, the historical number of waybills for small and medium-sized express delivery in area A on January 1, 2024 was 1,000,000, and the predicted number of waybills was 1,200,000; the historical number of waybills for small and medium-sized express delivery in area B on January 2, 2024 was 1,100,000, and the predicted number of waybills was 1,150,000; the historical number of waybills for small and medium-sized express delivery in area C on January 1, 2024 was 900,000, and the predicted number of waybills was 930,000; the historical number of waybills for small and medium-sized express delivery in area D on January 2, 2024 was 880,000, and the predicted number of waybills was 850,000.

[0079] It should be understood that Table 2 is only an example of the data format of the waybill quantity. The waybill quantity data table may also include the customer service center corresponding to the region, such as region A corresponding to customer service center 1. In addition to being differentiated by the size of the express, waybill types may also be differentiated by other dimensions, such as by the category of items included in the express. Waybill types include home appliance type express and clothing type express, which is not limited in the embodiments of the present disclosure.

[0080] In addition to data related to the number of waybills, external data also includes data on external factors that have an indirect impact on the forecast of service request volume. These external factors include promotional activities, holidays, abnormal weather, major events, and organizational structure adjustments of customer service centers. Among them, data on external factors such as promotional activities and holidays can be quantified and configured into the system in advance. External factors such as abnormal weather and organizational structure adjustments of customer service centers cannot be effectively quantified and can be reasonably used through manual interaction.

[0081] In the disclosed embodiment, the target data includes target history data, target prediction data and target event data. The target history data includes target history service request volume data and target history waybill volume data; the target prediction data includes target waybill volume prediction data; and the target event data refers to target external factor data that affects the service request volume.

[0082] Target data refers to data used to predict the volume of service requests, including target historical data, target prediction data, and target event data. The data range and business type of the target historical data can be set to obtain data from the data layer according to the set information. For example, if the data range is set to the most recent week and the business type is set to all business types, the historical data of the service request volume and the historical data of the waybill volume in the most recent week are obtained as the target historical data. It is also possible to interact with the user to obtain data parameter information, and obtain the target historical data from the data layer according to the data parameter information. For example, if the data range is determined to be the most recent 5 days and the business type is the telephone hotline business through the data parameter information, the historical data of the service request volume of the telephone hotline business in the most recent 5 days and the historical data of the waybill volume in the most recent 5 days are obtained as the target historical data.

[0083] The target forecast data includes the target waybill volume forecast data. For the customer service center, the number of service requests is strongly correlated with the number of waybill volumes. The increase in the number of waybill volumes will affect the increase in the number of service requests to a certain extent. Among them, the selection time of the target waybill volume forecast data can be determined according to the forecast time of the service request volume. For example, if the service request volume is predicted every day in the next week, the waybill volume forecast data for every day in the next week is obtained as the target forecast data.

[0084] Target event data refers to target external factor data that has an impact on the service request volume. As mentioned above, in addition to data related to the waybill volume, external data also includes external factor data that has an indirect impact on the forecast of the service request volume. These external factors include promotional activities, holidays, abnormal weather, major events, and organizational structure adjustments of the customer service center. Among them, the external factors can be determined based on the predicted time of the service request volume. For example, if the service request volume is predicted within the time range of a certain promotional activity, the target time data includes the external factor data corresponding to the promotional activity. It is also possible to interact with the user to obtain data parameter information, and obtain target event data based on the data parameter information. If the data parameter information includes abnormal weather factors, the external factor data caused by the abnormal weather will be added to the target event data.

[0085] In the disclosed embodiment, the target data includes target historical data, target prediction data and target event data, and these target data are used to predict the service request volume. Among them, the target historical data (such as target historical service request volume data and target historical waybill volume data) provides rich historical data and trend information for prediction, which helps the model learn and understand the seasonality, periodicity and other laws of service request volume; the target prediction data (such as target waybill volume prediction data) as a preliminary estimate of possible future situations can assist the model to more accurately predict the changes in service request volume; the target event data is the target external factor data that affects the service request volume, which can capture the impact of external introduction on the service request volume, and further improve the accuracy and timeliness of the prediction.

[0086] In the embodiments of the present disclosure, the initial data features include initial basic features, initial prediction features and initial event features; feature processing is performed on the target data to obtain the initial data features, including: feature processing is performed on the target historical data to obtain the initial basic features; feature processing is performed on the target prediction data to obtain the initial prediction features; feature processing is performed on the target event data to obtain the initial event features.

[0087] After acquiring the target data, the target data is feature processed through the feature layer to obtain initial data features. In the disclosed embodiment, feature processing is performed on the target historical data, target prediction data, and target event data to obtain corresponding initial basic features, initial prediction features, and initial event features.

[0088] Reference Figure 3 The feature layer can perform feature processing and feature fusion operations on internal and external data to obtain processed data features. Among them, the processed data features are divided into basic features and configuration features.

[0089] Basic features are objective descriptions of various historical data. Under most requests, users are not allowed to modify basic features. Basic features mainly include time features, time series historical features, and time series composite features, as shown in Table 3.

[0090] Table 3 Basic feature information table

[0091]

[0092] In Table 3, time features are divided into continuous time features and discrete time features. Among them, continuous time can include holiday duration, interval time, etc., and discrete time can include the beginning of the month, the end of the month, during the week, weekends, whether it is a holiday, etc. Time series historical features are divided into statistical value features and contemporaneous value features. Among them, statistical values ​​include 7-day averages, medians, maximum values, minimum values, skewness, kurtosis, etc., and contemporaneous values ​​include the values ​​of the same period of the previous month and the values ​​of the same period of last week. Time series composite features can also be called time series cross features, including continuous-continuous feature composites and continuous-category feature composites. Among them, continuous-continuous feature composites are generated by processing two continuous variables and combining them to generate new variables. For example, the difference between the customer service center's call volume this week and last week is xy, where x represents this week's call volume, y represents last week's call volume, and xy represents the month-on-month change. The continuous-categorical feature composite is achieved by processing a combination of a continuous variable and a categorical feature. For example, the weekend average value is [if (y==6or y==7,x)for x,y in data] / 2, which means that the call volume for the two days of the weekend is selected under the condition of y=6or7, and then divided by 2 to get the average.

[0093] For configuration features, data can be modified and configured according to user needs and actual scenarios. Configuration features are also called adjustment features, which can be understood as features that can be adjusted. Configuration features are divided into quantitative features and qualitative features. They are uncertain and may belong to features of future predictions or features of special events and scenarios.

[0094] Quantitative features refer to the future predicted values ​​of continuous data (external factors), which can be integrated into feature engineering as the input of the model. As shown in Table 4, the attributes of quantitative features are predictive features, which can include waybill quantity prediction, weather prediction, customer service complaint rate prediction, etc. Among them, weather prediction can be understood as a continuous feature or a categorical feature. For example, temperature is a continuous feature, and severe weather such as rain, snow, ice and frost is an enumerated categorical feature. This feature has no direct relationship with the service request volume, but in some extreme scenarios, such as severe rain and snow weather, high temperature in summer, etc., logistics services may have some problems of untimely service, which indirectly leads to a surge in user complaints and consultations, resulting in an increase in service requests.

[0095] Qualitative features refer to the features corresponding to special events and scenarios, which cannot be accurately quantified or the data dimension is insufficient to adapt to the input mode of the prediction model, or the user wants to evaluate the prediction results in different scenarios. As shown in Table 4, the attributes of qualitative features are event features, which can include the degree of traffic control, the degree of promotion, etc.

[0096] Table 4 Configuration feature information table

[0097] Feature Category Feature attributes Feature Example Description Quantitative features Predictive features Waybill volume forecast, weather forecast, customer service complaint rate forecast, etc. Qualitative characteristics Event Characteristics Degree of traffic control, degree of promotion, etc.

[0098] In the disclosed embodiment, the target data includes target historical features, target prediction features and target event features, and these features are processed to obtain initial basic features, initial prediction features and initial event features. Among them, the initial basic features are features that cannot be adjusted by the user, and the initial prediction features and initial event features are configuration features, that is, the user can adjust the initial prediction features and initial event features.

[0099] In the disclosed embodiment, feature adjustment information is used to indicate adjustment of data features; the feature adjustment information includes feature addition information and / or feature modification information. Feature adjustment is performed on the initial data features according to the feature adjustment information to obtain target data features, including: in response to the feature adjustment information including feature addition information, feature processing is performed according to the feature addition information to obtain newly added features, and the newly added features are added to the initial data features to obtain target data features; in response to the feature adjustment information including feature modification information, the initial data features are modified according to the feature modification information to obtain target data features.

[0100] like Figure 3 As shown, the parameter adjustment layer can perform feature addition and feature modification operations on data features. In the disclosed embodiment, interactive parameter information is obtained by interactive operations with the user, and the interactive parameter information includes feature adjustment information. After receiving the feature adjustment information, the parameter adjustment layer performs feature adjustment on the initial data features according to the feature adjustment information.

[0101] If the feature adjustment information includes feature addition information, the parameter adjustment layer obtains the newly added features according to the feature addition information and adds the newly added features to the initial data features. For example, according to the feature addition information, it is determined that the feature description of the promotion degree needs to be added, and the data features corresponding to the promotion degree are added.

[0102] If the feature adjustment information includes feature adjustment information, the parameter adjustment layer modifies the initial data features according to the feature adjustment information and determines the modified features as the target data features. For example, if the user believes that the predicted value of the waybill volume in the model input has a large deviation, the predicted value of the waybill volume is corrected.

[0103] In the disclosed embodiment, the introduction of feature adjustment information allows the user to adjust the initial data features according to needs, including adding new features and modifying the initial data features, so that the obtained prediction results can better meet different business scenarios and needs.

[0104] In the disclosed embodiment, the model type information is used to indicate the selected model type; determining the target prediction model based on the model type information includes: selecting the target prediction model from a model library based on the model type information; the model library includes statistical models and artificial intelligence models.

[0105] like Figure 3 As shown, the model library can also be called the model layer, which includes statistical models and artificial intelligence models. Among them, statistical models can also be called black box models and transparent models, and artificial intelligence models can also be called white box models and non-transparent models. Statistical models include time series models and regression models. The time series model is obtained by training based on the historical data of service request volume, and the regression model is obtained by modeling based on the correlation between the service request volume and the waybill volume. Artificial intelligence models include machine learning models and deep learning models.

[0106] Table 5 shows the multiple models and model features included in the model library. In Table 5, the statistical model includes a time series model and a causal regression model. Among them, the time series model is predicted by fitting the time curve according to the recent trend characteristics, and the causal regression model is based on external factor data and uses causal or correlation relationships for prediction. Artificial intelligence models include machine learning models and deep learning models. Among them, the machine learning model is a supervised learning model established based on the constructed feature engineering, and the deep model is based on an end-to-end neural network model.

[0107] Table 5 Models and model characteristics

[0108]

[0109] In the embodiment of the present disclosure, the time series model can be a third-order exponential smoothing model based on empty seasonality and trend. The third-order exponential smoothing model is trained based on the historical data of the service request volume, and its model principle is described as follows:

[0110]

[0111] In formula (1), the first term is the prediction output equation, It means that the service request volume at time (t+h) is predicted when the service request volume at time t is known. t+h-m(k+1)The calculation process of the subscript t+hm*(k+1) can be interpreted as: first, calculate the integer multiple of seasonal cycles k from the current time to the prediction time; then, subtract a complete seasonal cycle m to ensure that the predicted seasonal factor does not exceed the known seasonal pattern range; finally, add the time index t+h of the prediction time to obtain the index of the seasonal factor at the prediction time.

[0112] For example, the current time is t = 200, and the prediction y after h = 5 is 205 , the part that needs to depend on s (seasonal estimate) needs to take the subscript of the seasonal estimate as t+hm(k+1). If the seasonal cycle is m=90, k=floor((h-1) / m)=floor(4 / 90)=0, m(k+1)=90, t+hm(k+1)=205-90=115, which means that the subscript of the seasonal estimate of the previous seasonal cycle needs to be 115.

[0113] The last three terms in formula (1) are the baseline prediction state equation, the trend prediction state equation, and the seasonal prediction state equation, respectively. h represents the prediction step, m represents the number of seasonal cycles, and k represents the kth seasonal cycle. α, β, and γ are the smoothing coefficients of the baseline prediction, trend prediction, and seasonal prediction, respectively.

[0114] The steps of model construction are as follows: determining the prediction time length, i.e., determining the prediction time length h according to whether the prediction target is a short-term prediction or a long-term prediction; determining the seasonal cycle, i.e., determining the seasonal parameters of historical data through data analysis or seasonal intensity calculation, such as weekly seasonality m=7 and monthly seasonality m=30; determining the smoothing coefficient, i.e., directly selecting a fixed value through manual experience configuration, or setting the parameter range and evaluation criteria, and automatically selecting the value through the degree of model fitting; constructing the model, i.e., inputting the model parameters h, m, k and other related parameters into formula (1) respectively, and obtaining the service request volume forecast under different prediction step sizes.

[0115] In the disclosed embodiment, the regression model can be a linear regression model based on the prediction of the waybill quantity, that is, the regression model is modeled based on the strong correlation between the waybill quantity and the service request quantity, and the service request quantity is predicted from the waybill quantity prediction data. The model construction steps are: basic data input, obtaining the waybill quantity historical data and prediction data of the relevant business from the system; prediction evaluation, the prediction data of the waybill quantity may have deviations, and the credibility of the current prediction value can be judged by calculating the prediction accuracy over a period of history; prediction correction, if it is determined through the prediction evaluation result that the credibility of the current prediction value of the waybill quantity is low, the prediction result can be corrected by manual adjustment or historical error correction logic, and then substituted into the model calculation; model training and prediction, considering the strong correlation between the service request quantity and the waybill quantity, a polynomial linear regression model is used for model fitting, and the model formula is as follows:

[0116]

[0117] In formula (2), f(x) is the prediction correction function, is the forecast data of the corrected shipping order quantity, X t is the initial forecast data of the waybill quantity, is the predicted service request volume, a1, a2, ... a n as a parameter.

[0118] In the disclosed embodiments, the machine learning model may include models such as random forest, support vector machine, and K-nearest neighbor. Among them, random forest is an integrated learning method that improves prediction accuracy by building multiple decision trees. Support vector machine (SVM) is a supervised learning algorithm that performs classification or regression by finding the maximum boundary between different categories. K-nearest neighbors (KNN) is an instance-based learning method that makes predictions by finding the training instance that is most similar to the new instance.

[0119] In the disclosed embodiments, the deep learning model may include models such as recurrent neural networks, long short-term memory networks, convolutional neural networks and Transformer. Among them, recurrent neural networks (RNNs) are suitable for processing sequence data, such as time series prediction. Long Short-Term Memory (LSTM) is a special type of RNN that can better handle long-term dependency problems. Convolutional Neural Networks (CNNs) are commonly used for image processing and can also be used for one-dimensional signal processing, such as service request volume prediction. Transformer is commonly used in natural language processing and is also applied to time series prediction. It uses a self-attention mechanism to capture long-term dependencies in the sequence.

[0120] In the disclosed embodiment, the model library includes different types of models, such as time series models, regression models, machine learning modules, and deep learning models. These models can all be used to predict the service request volume. Different models are suitable for different data characteristics and prediction scenarios. The most appropriate model can be selected according to the specific situation, thereby further improving the prediction accuracy of the service request volume.

[0121] Figure 3The system shown also provides a mode adjustment function, that is, obtaining mode adjustment information through interactive operations with users, and using the mode adjustment information to adjust the prediction of service request volume. Among them, mode adjustment includes simple interaction mode and feature interaction mode. The simple interaction mode mainly makes adjustments based on the model prediction results, which are divided into addition calculation mode and multiplication calculation mode. The idea of ​​the feature interaction mode is to reconstruct the existing feature engineering based on the configuration features, input the prediction model for retraining and prediction, and is divided into single feature calculation mode and multi-feature calculation mode.

[0122] Table 6 Adjustment mode information table

[0123]

[0124] In Table 6, the adjustment level corresponding to the simple interaction model is the model output, which is used to adjust the model prediction results, including addition and multiplication. Among them, addition is applicable to additive adjustment on the model prediction results, and multiplication is applicable to multiple adjustment on the model prediction results. For example, if the model prediction result is 10,000 service request data for the next day, considering the recent abnormal weather, the service request data is adjusted to 1.2 times the prediction result, that is, 12,000, and customer service personnel are arranged according to the adjusted data. The adjustment level corresponding to the feature interaction model is feature input, that is, reconstructing the existing feature engineering, retraining the model, and using the trained model for prediction, including single feature calculation and multi-feature calculation. Among them, single feature calculation is applicable to models based on single feature modeling, and multi-feature calculation is applicable to models based on multi-feature modeling.

[0125] It should be noted that different adjustment modes have corresponding model selection ranges. For example, the simple interaction mode can be applied to all prediction models, and the feature interaction mode is applicable to machine learning models based on feature engineering.

[0126] In the disclosed embodiment, the pattern adjustment information is used to indicate adjustment to the model prediction result; based on the target prediction model, the service request amount is predicted according to the pattern adjustment information and the target data characteristics to obtain the target service request amount, including: based on the target prediction model, the service request amount is predicted according to the target data characteristics to obtain the initial service request amount; the initial service request amount is adjusted according to the pattern adjustment information to obtain the target service request amount.

[0127] If the mode adjustment information includes a simple interaction mode, that is, it is used to instruct adjustment of the model prediction results, the service request volume is first predicted using the target prediction model to obtain the initial service request volume, and then the initial service request volume is adjusted according to the specific instructions in the model adjustment information, such as adjusting the service request volume data to 1.2 times the predicted result, to obtain the target service request volume.

[0128] In the disclosed embodiment, pattern adjustment information is used to indicate reconstruction of existing data features; based on a target prediction model, the service request volume is predicted according to the pattern adjustment information and the target data features to obtain a target service request volume, including: if the target prediction model is a machine learning model obtained by training based on existing data features, the existing data features are reconstructed according to the pattern adjustment information to obtain reconstructed data features; the target prediction model is trained according to the reconstructed data features to obtain a new target prediction model; the service request volume is predicted based on the new target prediction model to obtain a target service request volume.

[0129] If the mode adjustment information includes a feature interaction mode, that is, it is used to instruct the reconstruction of existing data features. In this case, the target prediction model includes a machine learning model trained based on the existing data features. The existing data features are first reconstructed according to the mode adjustment information to obtain reconstructed data features. Then, the machine learning model is retrained using the reconstructed data features to obtain a new prediction model. The service request volume is predicted based on the new prediction model to obtain the target service request volume.

[0130] In the disclosed embodiments, the initial service request volume is adjusted through pattern adjustment information (such as a simple interaction pattern), so that the model prediction results can be fine-tuned based on business experience or specific needs to be more in line with business conditions and improve prediction accuracy; when the pattern adjustment information indicates the reconstruction of existing data features (such as a feature interaction pattern), more complex relationships between data can be captured by reconstructing the data features and retraining the model, thereby further improving the accuracy of the prediction model; the diversity of pattern adjustment information allows the system to select different adjustment strategies according to different situations, thereby enhancing the flexibility and adaptability of the system.

[0131] In the disclosed embodiment, there are multiple target prediction models; the service request volume prediction method also includes: obtaining the target service request volume of each target prediction model; weighting the target service request volume of each target prediction model to obtain the weighted service request volume, and determining the weighted service request volume as the prediction result.

[0132] Reference Figure 3 The fusion layer is used to perform prediction combination and prediction selection operations on the prediction results. In the disclosed embodiment, the number of target prediction models can be multiple. After obtaining the target service request volume predicted by each target prediction model, the fusion layer can be used to select a suitable prediction result. The fusion layer can also perform weighted processing on the target service request volume predicted by each target prediction model to obtain the weighted service request volume, that is, the final prediction result.

[0133] In the disclosed embodiment, there are multiple target prediction models, and the service request volume can be predicted from different angles and dimensions. Different models may capture different patterns and features in the data. By integrating the prediction results of multiple target prediction models, the possible deviations and uncertainties of a single model can be reduced, thereby improving the stability and reliability of the overall prediction results.

[0134] Figure 4 FIG. 1 is an interaction diagram between a user and a service request volume prediction system according to an embodiment of the present disclosure. Figure 4 , the interaction process can include:

[0135] (1) Parameter configuration: The user selects parameters and obtains interactive parameter information. The selected parameters include time range, service type, feature configuration, mode selection, and model selection.

[0136] The time range may include the time range of the selected historical data; the business type may include the business type corresponding to the selected service request volume data, such as a telephone hotline; feature configuration may be understood as feature adjustment, including feature addition and feature modification; mode selection refers to adjusting the model selection, including a simple interaction mode and a feature interaction model; model selection refers to the model used for prediction.

[0137] (2) Parameter transmission: Transmit the interactive parameter information to the parameter adjustment calculation engine.

[0138] (3) Information acquisition: The parameter adjustment calculation engine obtains relevant data (i.e., target data), feature adjustment, and model information (i.e., target prediction model used for prediction) based on the interaction parameter information to generate a prediction task.

[0139] (4) Scheduling calculation: Based on the prediction task, the feature calculation module and the model prediction module are called to calculate the prediction results.

[0140] (5) Result update: The prediction results are returned and the system data display is updated in real time.

[0141] Figure 5 : is a comparison chart of the prediction accuracy of the existing method and the service request volume prediction method of the embodiment of the present disclosure. Figure 5 It can be seen that the average prediction accuracy of the existing manual prediction method is 84.75%, and the average prediction accuracy of the method provided by the embodiment of the present disclosure is 90.5%. It can be seen that compared with the existing manual prediction method, the prediction accuracy of the method provided by the embodiment of the present disclosure is higher. Figure 5 It can be seen that the prediction accuracy of the method provided by the embodiment of the present disclosure is steadily improving.

[0142] It should be noted that, although the steps of the method of the present invention are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0143] Figure 6 is a block diagram of a service request volume prediction device according to an embodiment of the present disclosure. Figure 6 The device 600 includes an interaction module 610, a data acquisition module 620, a feature processing module 630, a feature adjustment module 640, a model determination module 650 and a prediction module 660.

[0144] Among them, the interaction module 610 is used to: obtain interaction parameter information in response to the interaction operation; wherein the interaction parameter information includes data parameter information, feature adjustment information, model type information and mode adjustment information. The data acquisition module 620 is used to: obtain target data related to the service request amount according to the data parameter information. The feature processing module 630 is used to: perform feature processing on the target data to obtain initial data features. The feature adjustment module 640 is used to: perform feature adjustment on the initial data features according to the feature adjustment information to obtain target data features. The model determination module 650 is used to: determine the target prediction model according to the model type information. The prediction module 660 is used to: predict the service request amount based on the target prediction model according to the mode adjustment information and the target data features to obtain the target service request amount.

[0145] In some embodiments of the present disclosure, the target data includes target historical data, target prediction data and target event data, and the initial data features include initial basic features, initial prediction features and initial event features. The feature processing module 630 is also used to: perform feature processing on the target historical data to obtain initial basic features; perform feature processing on the target prediction data to obtain initial prediction features; perform feature processing on the target event data to obtain initial event features.

[0146] In some embodiments of the present disclosure, feature adjustment information is used to indicate adjustment of data features; the feature adjustment information includes feature addition information and / or feature modification information.

[0147] In some embodiments of the present disclosure, the feature adjustment module 640 is also used to: in response to feature adjustment information including feature addition information, perform feature processing according to the feature addition information to obtain newly added features, add the newly added features to initial data features, and obtain target data features; in response to feature adjustment information including feature modification information, modify the initial data features according to the feature modification information to obtain target data features.

[0148] In some embodiments of the present disclosure, the target historical data includes target historical service request volume data and target historical waybill volume data; the target prediction data includes target waybill volume prediction data; and the target event data refers to target external factor data that affects the service request volume.

[0149] In some embodiments of the present disclosure, the model type information is used to indicate the selected model type. The model determination module 650 is also used to: select a target prediction model from a model library according to the model type information; wherein the model library includes statistical models and artificial intelligence models; the statistical model includes a time series model and a regression model, the time series model is obtained by training based on the historical data of the service request volume, and the regression model is obtained by modeling based on the correlation between the service request volume and the waybill volume; the artificial intelligence model includes a machine learning model and a deep learning model.

[0150] In some embodiments of the present disclosure, the mode adjustment information is used to indicate the adjustment of the model prediction result. The prediction module 660 is also used to: predict the service request amount according to the target data characteristics based on the target prediction model to obtain the initial service request amount; adjust the initial service request amount according to the mode adjustment information to obtain the target service request amount.

[0151] In some embodiments of the present disclosure, the mode adjustment information is used to instruct to reconstruct the existing data features. The prediction module 660 is also used to: if the target prediction model is a machine learning model obtained by training the existing data features, reconstruct the existing data features according to the mode adjustment information to obtain the reconstructed data features; train the target prediction model according to the reconstructed data features to obtain a new target prediction model; predict the service request volume based on the new target prediction model to obtain the target service request volume.

[0152] In some embodiments of the present disclosure, there are multiple target prediction models. The prediction module 660 is also used to: obtain the target service request volume of each target prediction model; perform weighted processing on the target service request volume of each target prediction model to obtain the weighted service request volume, and determine the weighted service request volume as the prediction result.

[0153] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0154] It should be noted that although several modules or units of the service request volume prediction device are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be concretized in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into multiple modules or units to be concretized.

[0155] In addition, in an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above method is also provided. Figure 7 An electronic device 700 according to such an embodiment of the present invention is described. Figure 7 The electronic device 700 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0156] like Figure 7 As shown, the electronic device 700 is in the form of a general computing device. The components of the electronic device 700 may include, but are not limited to: the at least one processing unit 710, the at least one storage unit 720, a bus 730 connecting different system components (including the storage unit 720 and the processing unit 710), and a display unit 740.

[0157] The storage unit stores program codes, which can be executed by the processing unit 710, so that the processing unit 710 performs the steps described in the "Exemplary Method" section of the present specification according to various exemplary embodiments of the present disclosure. For example, the processing unit 710 can perform the steps shown in the above embodiments.

[0158] The storage unit 720 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 721 and a cache memory unit 722 , and may further include a read-only memory unit (ROM) 723 .

[0159] The storage unit 720 may also include a program / utility 724 having a set (at least one) of program modules 725, such program modules 725 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0160] Bus 730 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0161] The electronic device 700 may also communicate with one or more external devices 800 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 700, and may communicate with any device that enables the electronic device 700 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 750. Furthermore, the electronic device 700 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and public networks, such as the Internet) via a network adapter 760. As shown, the network adapter 760 communicates with other modules of the electronic device 700 via a bus 730. It should be understood that, although not shown in the figure, other hardware and software modules may be used in conjunction with the electronic device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0162] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiment of the present disclosure.

[0163] In an exemplary embodiment, a computer-readable storage medium including instructions is also provided, such as a memory including instructions, and the above instructions can be executed by a processor of the device to complete the above method. Optionally, the computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0164] In an exemplary embodiment, a computer program product is also provided, including a computer program and instructions. When the computer program and the instructions are executed by a processor, the data processing method in the above embodiment is implemented.

[0165] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0166] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method for predicting service request volume, characterized in that: include: In response to the interactive operation, interactive parameter information is obtained; the interactive parameter information includes data parameter information, feature adjustment information, model type information and mode adjustment information; According to the data parameter information, target data related to the service request amount is acquired, and feature processing is performed on the target data to obtain initial data features; Performing feature adjustment on the initial data features according to the feature adjustment information to obtain target data features; Determining a target prediction model according to the model type information; Based on the target prediction model, the service request volume is predicted according to the mode adjustment information and the target data characteristics to obtain a target service request volume.

2. The method according to claim 1, characterized in that The target data includes target historical data, target prediction data and target event data, and the initial data features include initial basic features, initial prediction features and initial event features; The performing feature processing on the target data to obtain initial data features includes: Performing feature processing on the target historical data to obtain the initial basic features; Performing feature processing on the target prediction data to obtain the initial prediction feature; Perform feature processing on the target event data to obtain the initial event feature.

3. The method according to claim 2, characterized in that The feature adjustment information is used to indicate adjustment of data features; the feature adjustment information includes feature addition information and / or feature modification information.

4. The method according to claim 3, characterized in that: The step of adjusting the initial data features according to the feature adjustment information to obtain target data features includes: In response to the feature adjustment information including the feature adding information, performing feature processing according to the feature adding information to obtain a newly added feature, and adding the newly added feature to the initial data feature to obtain the target data feature; In response to the feature adjustment information including the feature modification information, the initial data feature is modified according to the feature modification information to obtain the target data feature.

5. The method according to any one of claims 2 to 4, characterized in that: The target historical data includes target historical service request volume data and target historical waybill volume data; the target prediction data includes target waybill volume prediction data; and the target event data refers to target external factor data that affects the service request volume.

6. The method according to claim 1, characterized in that The model type information is used to indicate the selected model type; Determining the target prediction model according to the model type information includes: Selecting the target prediction model from a model library according to the model type information; Wherein, the model library includes statistical models and artificial intelligence models; The statistical model includes a time series model and a regression model. The time series model is obtained by training based on the historical data of the service request volume, and the regression model is obtained by modeling based on the correlation between the service request volume and the waybill volume. The artificial intelligence model includes a machine learning model and a deep learning model.

7. The method according to claim 1, characterized in that The mode adjustment information is used to instruct to adjust the model prediction result; The step of predicting the service request amount based on the target prediction model and according to the mode adjustment information and the target data characteristics to obtain the target service request amount includes: Based on the target prediction model, predicting the service request volume according to the target data characteristics to obtain an initial service request volume; The initial service request amount is adjusted according to the mode adjustment information to obtain the target service request amount.

8. The method according to claim 1, characterized in that The mode adjustment information is used to instruct to reconstruct the existing data features; The step of predicting the service request amount based on the target prediction model and according to the mode adjustment information and the target data characteristics to obtain the target service request amount includes: If the target prediction model is a machine learning model obtained by training the existing data features, reconstructing the existing data features according to the pattern adjustment information to obtain reconstructed data features; Training the target prediction model according to the reconstructed data features to obtain a new target prediction model; The service request volume is predicted based on the new target prediction model to obtain the target service request volume.

9. The method according to claim 1, characterized in that: The number of the target prediction models is multiple; the method further includes: Obtain target service request volume of each target prediction model; The target service request quantities of the target prediction models are weighted to obtain the weighted service request quantities, and the weighted service request quantities are determined as the prediction results.

10. A service request volume prediction device, characterized in that: include: An interaction module, used for obtaining interaction parameter information in response to an interaction operation; The interaction parameter information includes data parameter information, feature adjustment information, model type information and mode adjustment information; A data acquisition module, used to acquire target data related to the service request amount according to the data parameter information; A feature processing module, used to perform feature processing on the target data to obtain initial data features; A feature adjustment module, used to adjust the initial data features according to the feature adjustment information to obtain target data features; A model determination module, used to determine a target prediction model according to the model type information; A prediction module is used to predict the service request volume based on the target prediction model, according to the mode adjustment information and the target data characteristics, to obtain a target service request volume.

11. A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the service request volume prediction method according to any one of claims 1 to 9 is implemented.

12. An electronic device, characterized in that: include: processor; and a memory for storing executable instructions for the processor; The processor is configured to execute the service request volume prediction method according to any one of claims 1 to 9 by executing the executable instructions.

13. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the service request volume prediction method according to any one of claims 1 to 9 is implemented.

Images