Parcel quantity prediction method and device, computer device, and storage medium

Abstract

The application provides a parcel quantity prediction method and device, computer equipment and a storage medium. The method comprises the following steps: obtaining a parcel attribute prediction value of a target area in a first time period; obtaining a parcel attribute true value of the target area in a second time period; determining a parcel attribute error according to the parcel attribute prediction value and the parcel attribute true value; and determining a first quantity prediction value of the target area in the first time period based on the parcel attribute error. The method can effectively improve the accuracy of parcel prediction and save manpower and resources.

Description

Technical Field

[0001] This application relates to the field of logistics technology, specifically to a method, apparatus, computer equipment, and storage medium for predicting the volume of express parcels. Background Technology

[0002] In the actual logistics field, in order to reduce costs, it is often necessary to predict the volume of express parcels in different delivery areas, including the volume of parcels received and the volume of parcels delivered, and to dispatch couriers to collect and deliver the parcels.

[0003] Traditional single-model predictions of parcel volume are often inaccurate. Deviations in parcel volume prediction often lead to deviations in the accuracy of subsequent scheduling algorithms, resulting in error accumulation and increased costs for delivery personnel.

[0004] Therefore, existing methods for predicting express delivery volume suffer from the technical problem of low accuracy in volume prediction. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, device, computer equipment, and storage medium for predicting the volume of express parcels to address the aforementioned technical problems, so as to improve the accuracy of express parcel volume prediction.

[0006] Firstly, this application provides a method for predicting the volume of express parcels, including:

[0007] Obtain predicted values ​​of parcel attributes for the target area within the first time period;

[0008] Obtain the actual values ​​of the parcel attributes in the target area during the second time period;

[0009] The error in the express delivery attribute is determined based on the predicted value of the express delivery attribute and the actual value of the express delivery attribute.

[0010] Based on the express delivery attribute error, the predicted value of the first shipment volume in the target area within the first time period is determined; wherein, the first time period is after the second time period.

[0011] In some embodiments of this application, obtaining the predicted values ​​of express mail attributes for the target area within a first time period includes:

[0012] Obtain the actual quantity of items within a third time period, which is prior to the first time period;

[0013] Based on multiple preset quantity prediction models and the actual quantity value within the third time period, obtain the third quantity prediction value corresponding to each quantity prediction model in the multiple quantity prediction models, and obtain multiple third quantity prediction values.

[0014] Based on the multiple predicted values ​​of the third quantity, the predicted value of the second quantity within the first time period is obtained;

[0015] Based on the second predicted quantity value, the predicted value of the express item attribute is calculated.

[0016] In some embodiments of this application, obtaining the second quantity prediction value within the first time period based on the plurality of third quantity prediction values ​​includes:

[0017] Obtain the initial regression function for the preset part quantity prediction;

[0018] Obtain the average quantity of items within the second time period;

[0019] The average quantity of items and the multiple third quantity prediction values ​​are used as feature inputs to the initial regression function to obtain the second quantity prediction value within the first time period.

[0020] In some embodiments of this application, obtaining the predicted values ​​of express mail attributes for the target area within a first time period includes:

[0021] Get the maximum number of packages received, the maximum number of packages delivered, and the courier attributes for each courier.

[0022] Obtain the state equation corresponding to the preset package attribute prediction;

[0023] Substituting the second predicted quantity of packages, the maximum number of packages received, the maximum number of packages delivered, and the courier attributes into the state equation, the predicted value of the package attributes is calculated.

[0024] In some embodiments of this application, obtaining the true value of the number of parcels within the first time period to determine the true value of the parcel attributes includes:

[0025] Substituting the actual value of the number of packages, the maximum number of packages received by the courier, the maximum number of packages delivered, and the courier attributes into the state equation, the actual value of the package attributes is calculated.

[0026] In some embodiments of this application, obtaining the true values ​​of the parcel attributes of the target area within the second time period includes:

[0027] Obtain the actual quantity of items within the second time period;

[0028] Substituting the actual number of parcels received, the maximum number of parcels delivered, and the courier's attributes into the state equation during the second time period, the actual parcel attribute values ​​for the target area during the second time period are calculated.

[0029] In some embodiments of this application, determining the predicted quantity of the first shipment in the target area within the first time period based on the shipment attribute error includes:

[0030] Obtain the true values ​​of express mail attributes for the target area within multiple second time periods to obtain multiple express mail attribute errors;

[0031] By utilizing the multiple express delivery attribute errors, the initial regression function is trained to obtain the target regression function;

[0032] The predicted value of the first quantity is calculated using the target regression function.

[0033] In some embodiments of this application, training the initial regression function using the plurality of express delivery attribute errors to obtain the target regression function includes:

[0034] The initial regression function is trained using the multiple express delivery attribute errors to obtain the trained regression function;

[0035] Determine whether the loss function corresponding to the trained regression function meets the preset requirements;

[0036] If the loss function does not meet the preset requirements, the new express delivery attribute error is calculated using the trained regression function;

[0037] The regression function is trained using the new express delivery attribute error until the loss function meets the preset requirements to obtain the target regression function.

[0038] Secondly, this application provides a device for predicting the volume of express parcels, comprising:

[0039] The first acquisition module is used to acquire the predicted values ​​of express mail attributes in the target area within the first time period.

[0040] The second acquisition module is used to acquire the actual values ​​of the express mail attributes of the target area within the second time period.

[0041] The error calculation module is used to determine the error of the express delivery attribute based on the predicted value of the express delivery attribute and the actual value of the express delivery attribute.

[0042] The shipment volume prediction module is used to determine the first shipment volume prediction value of the target area within the first time period based on the shipment attribute error; wherein the first time period is after the second time period.

[0043] Thirdly, this application also provides a computer device, the computer device comprising:

[0044] One or more processors;

[0045] Memory; and

[0046] 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 express delivery volume prediction method.

[0047] Fourthly, 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 express delivery volume prediction method.

[0048] Fifthly, embodiments of this application provide a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method provided in the first aspect described above.

[0049] The aforementioned method, apparatus, computer equipment, and storage medium for predicting parcel volume involve a server that acquires predicted parcel attribute values ​​for the target area and actual parcel attribute values ​​corresponding to the actual parcel volume over a historical period. This data is then used to derive the parcel attribute error between the actual parcel attribute values ​​and the parcel cost attribute values. Based on this error, the parcel volume for the target area is re-predicted. Finally, the parcel attribute error between the actual and predicted parcel attribute values ​​is used to further predict the parcel volume, thereby improving the accuracy of parcel prediction. Attached Figure Description

[0050] 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a schematic diagram of a scenario for the express delivery volume prediction method in the embodiments of this application;

[0052] Figure 2 This is a flowchart illustrating the express delivery volume prediction method in the embodiments of this application;

[0053] Figure 3 This is a flowchart illustrating the express mail attribute prediction step in an embodiment of this application;

[0054] Figure 4 This is a flowchart illustrating the second step of obtaining the predicted quantity value in an embodiment of this application.

[0055] Figure 5 This is a schematic diagram of the structure of the GRU model in the embodiments of this application;

[0056] Figure 6 This is a flowchart illustrating the second step of obtaining the predicted quantity value in an embodiment of this application.

[0057] Figure 7 This is a flowchart illustrating the steps for obtaining predicted values ​​of express mail attributes in an embodiment of this application.

[0058] Figure 8 This is a flowchart illustrating the quantity prediction process in an embodiment of this application;

[0059] Figure 9 This is a schematic diagram of the express parcel volume prediction device in the embodiments of this application;

[0060] Figure 10 This is a schematic diagram of the structure of the computer device in the embodiments of this application. Detailed Implementation

[0061] 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.

[0062] In the description of this application, 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, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0063] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" 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 the invention. 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 the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention 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.

[0064] In this application embodiment, the express delivery volume prediction method mainly involves computer vision (CV) technology within artificial intelligence (AI). Artificial intelligence utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to obtain optimal results—theories, methods, technologies, and application systems. In other words, artificial intelligence is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine capable of reacting in a manner similar to human intelligence.

[0065] In the embodiments of this application, it should be noted that since the express delivery volume prediction method provided in this application is executed in a computer device, the processing objects of each computer device exist in the form of data or information, such as time, which is essentially time information. It can be understood that if size, quantity, location, etc. are mentioned in subsequent embodiments, they are all corresponding data that exist so that the computer device can process them. The specifics will not be elaborated here.

[0066] In the embodiments of this application, it should also be noted that the express delivery volume prediction method provided in the embodiments of this application can be applied to, for example, Figure 1 The parcel volume prediction system shown includes a terminal 100 and a server 200. The terminal 100 can be a device that includes both receiving and transmitting hardware, meaning it has receiving and transmitting hardware capable of performing bidirectional communication over a two-way communication link. Such a device can include cellular or other communication equipment with a single-line display, a multi-line display, or no multi-line display. Specifically, the terminal 100 can be a desktop terminal or a mobile terminal, and can also be a mobile phone, tablet computer, laptop computer, etc. The server 200 can be a standalone server or a server network or server cluster, including but not limited to computers, network hosts, single network servers, multiple network server sets, or cloud servers composed of multiple servers. The cloud server consists of a large number of computers or network servers based on cloud computing.

[0067] 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 those that are more specific to this application. Figure 1 The number of computer devices shown is more or less, for example Figure 1 Only one server (200) is shown in the image. It is understandable that this parcel volume prediction system may include one or more other servers; specific details are not specified here. Additionally, as... Figure 1 As shown, the express delivery volume prediction system may also include a memory for storing data, such as express delivery waybill data.

[0068] It should also be noted that, Figure 1 The schematic diagram of the express delivery volume prediction system shown is merely an example. The express delivery volume prediction system and scenario described in this embodiment are for the purpose of more clearly illustrating the technical solutions of this embodiment and do not constitute a limitation on the technical solutions provided by this embodiment. As those skilled in the art will know, with the evolution of express delivery volume prediction systems and the emergence of new business scenarios, the technical solutions provided by this embodiment are also applicable to similar technical problems.

[0069] See Figure 2 This application provides a method for predicting the volume of express parcels, mainly applied to the above-mentioned... Figure 1 Taking server 200 as an example, the method includes steps S201 to S204, as follows:

[0070] S201. Obtain the predicted values ​​of express mail attributes for the target area in the first time period.

[0071] In specific implementation, the embodiments of this application predict the number of parcels within a fixed area, and the parcel prediction in the embodiments of this application may include prediction of parcel receipt and delivery. When predicting parcel attributes, the actual number of parcels over a historical period is usually used to predict the number of parcels in a future period, thereby determining the parcel attributes corresponding to the predicted number of parcels.

[0072] In this embodiment of the application, the express delivery attribute can be the cost corresponding to the express delivery, that is, the express delivery cost corresponding to the predicted quantity of express delivery.

[0073] More specifically, the predicted value of the second quantity of packages in the target area within the first time period can be obtained to determine the predicted value of the express delivery cost corresponding to the predicted value of the second quantity of packages.

[0074] S202. Obtain the actual values ​​of the parcel attributes in the target area within the second time period.

[0075] In specific implementation, this application embodiment proposes that after obtaining the predicted value of the express attribute corresponding to the predicted express quantity, it is also necessary to obtain the actual value of the quantity over a historical period to determine the actual value of the express attribute over a historical period; and compare the predicted value of the express attribute with the actual value of the express attribute.

[0076] Similarly, the package attribute can be the package cost, so it is necessary to obtain the actual package cost value of the target area in the second time period.

[0077] More specifically, the actual number of parcels in the second historical time period can be obtained. The actual number of parcels in the second time period can usually be obtained directly from cloud platforms or databases. After obtaining the actual number of parcels in the second time period, the actual cost of the parcels can be obtained.

[0078] It should be noted that in the embodiments of this application, the first time period is after the second time period, and the duration of the first and second time periods is usually the same. That is, the comparison is between the actual value of the express delivery cost and the predicted value of the express delivery cost within the same duration.

[0079] S203. Determine the error of the express delivery attributes based on the predicted value and the actual value of the express delivery attributes.

[0080] In specific implementation, the express delivery attribute involved in this application embodiment can be the express delivery cost. Therefore, after determining the predicted value and the actual value of the express delivery cost, the express delivery cost error between the two can be obtained. Specifically, the absolute value of the difference between the two can be taken as the express delivery cost error; the express delivery attribute error is the express delivery cost error.

[0081] S204. Based on the error of the express delivery attributes, determine the predicted value of the first quantity of the first item in the target area within the first time period.

[0082] In specific implementation, the second predicted shipment quantity obtained in this embodiment is not the actual predicted shipment quantity for the target area. It is necessary to train the preset loss function multiple times based on the shipment cost error between the actual and predicted shipment costs until the first predicted shipment quantity for the target area within the first time period is obtained. This first predicted shipment quantity is the actual predicted shipment quantity, and couriers are dispatched to collect and deliver the shipments based on this first predicted shipment quantity.

[0083] The aforementioned method for predicting parcel volume involves the server obtaining predicted parcel attribute values ​​for the target area and actual parcel attribute values ​​corresponding to historical parcel volumes over a specific period. This allows for the calculation of the parcel attribute error between the actual parcel attribute values ​​and the parcel cost attribute values. Based on this error, the parcel volume for the target area is re-predicted. Finally, the parcel attribute error between the actual and predicted parcel attribute values ​​is used to further predict the parcel volume, thus improving the accuracy of parcel prediction.

[0084] In one embodiment, such as Figure 3 As shown, to obtain the predicted values ​​of express mail attributes for the target area within the first time period, the following are included:

[0085] S301. Obtain the actual quantity of items within the third time period.

[0086] S302. Based on the preset multiple quantity prediction models and the actual quantity value in the third time period, obtain the third quantity prediction value corresponding to each quantity prediction model in the multiple quantity prediction models, and obtain multiple third quantity prediction values.

[0087] In practice, there are multiple methods or models that can predict the volume of express parcels. Each parcel volume prediction model can be used to predict the volume of express parcels within a fixed area over a certain period of time. However, the accuracy of a single prediction method / model in predicting the volume of express parcels is not high. Therefore, this application proposes to integrate multiple parcel volume prediction models to improve the accuracy of express parcel volume prediction.

[0088] More specifically, in the embodiments of this application, multiple preset quantity prediction models are first obtained and then integrated. The number of quantity prediction models can be determined according to the actual situation; and the specific quantity prediction model can also be selected according to the actual situation.

[0089] Furthermore, in the embodiments of this application, integrating multiple quantity prediction models is intended to obtain more accurate quantity prediction values. Integrating multiple quantity prediction models actually requires first training each model using the obtained actual quantity values ​​within the third time period, to obtain the third quantity prediction value for the target area corresponding to each model within the third time period. There are multiple third quantity prediction values; and the actual quantity values ​​within the third time period are historical data prior to the first time period.

[0090] Meanwhile, for different quantity prediction models, the methods for obtaining the third quantity prediction value are different, and the obtained third quantity prediction value is also different.

[0091] Generally speaking, when using the actual quantity of items in the third time period to predict the third quantity, the actual quantity of items in the third time period can be the actual quantity of items over a historical period of one year. The more historical data available for prediction, the more accurate the predicted third quantity will be.

[0092] S303. Based on multiple predicted values ​​of the third quantity, obtain the predicted value of the second quantity within the first time period.

[0093] In specific implementation, the second quantity prediction value in this application embodiment is obtained by integrating multiple third quantity prediction values ​​and then performing a second prediction. That is to say, the aforementioned integration of multiple quantity prediction models is actually the integration of multiple third quantity prediction values ​​corresponding to multiple quantity prediction models; based on the multiple third quantity prediction values, the second quantity prediction value is further predicted.

[0094] S304. Based on the second quantity prediction value, calculate the predicted value of the express item attributes.

[0095] In practice, after obtaining the first predicted quantity value, the predicted cost of express delivery within the target area in the first time period can be calculated; the predicted value of express delivery attributes is the predicted cost of express delivery. The specific calculation method for the predicted cost of express delivery will be explained later and will not be limited here.

[0096] In some embodiments of this application, there can be three preset quantity prediction models, therefore there are also three third quantity prediction values. Furthermore, the number of preset quantity prediction models can be changed according to actual needs, and different quantity prediction models can be selected based on actual requirements.

[0097] In one embodiment, such as Figure 4 As shown, the step of obtaining the predicted quantity of the second item within the first time period based on multiple predicted quantities of the third item includes:

[0098] S401. Obtain the initial regression function for the preset part quantity prediction.

[0099] Specifically, in the embodiments of this application, multiple third-party express delivery prediction values ​​are integrated to predict the second quantity prediction value of the target area within the first time period. In fact, multiple third-party express delivery prediction values ​​are used as inputs to the initial regression function of the preset quantity prediction, so as to use the initial regression function of the quantity prediction to predict the second quantity prediction value of the target area within the first time period.

[0100] In the embodiments of this application, an initial regression function can be used to predict the quantity of parts; that is, regression prediction is performed using the initial regression function, and the predicted result is the predicted quantity of parts. The initial regression function includes independent variables and dependent variables; calculations are needed based on the independent and dependent variables to establish the initial regression function (regression equation).

[0101] In the embodiments of this application, the independent variables of the initial regression function can be multiple third-order predicted values, and the dependent variable is the second-order predicted value. The specific steps for establishing the initial regression function between the multiple third-order predicted values ​​and the second-order predicted value will be described in detail later, and are not limited here.

[0102] S402, Obtain the average quantity of items within the second time period.

[0103] In specific implementation, the actual number of parcels in this application embodiment also needs to be used as the input of the initial regression function. Therefore, the average number of parcels in the second time period can be used as the independent variable of the initial regression function. The average number of parcels will also affect the relevant parameters in the initial regression function, so as to affect the final predicted value of the second parcel number.

[0104] S403. Use the average quantity of parts and multiple third-order quantity predictions as feature inputs to the initial regression function to obtain the second-order quantity predictions for the first time period.

[0105] In a specific implementation, the embodiments of this application involve using multiple third-order quantity prediction values ​​and the average quantity as inputs to an initial regression function to adjust the parameters in the initial regression function, thereby obtaining the final regression function, and further obtaining the second-order quantity prediction value within the first time period based on the regression function.

[0106] In one specific embodiment of this application, the multiple quantity prediction models can be three, which can be CatBoost, Gated Recurrent Unit (GRU), and Prophet, respectively. The process of obtaining three third quantity prediction values ​​using the three quantity prediction models is as follows:

[0107] Taking the Gated Recurrent Unit (GRU) as an example, GRU is a variant of Long Short-Term Memory (LSTM) networks, which can effectively eliminate the gradient vanishing problem during neural network model training. LSTM introduces three gate functions: the input gate, the forget gate, and the input gate again to control the model's input, memory, and output values, respectively. In contrast, the GRU model only includes two gates: the update gate and the reset gate. These two gate vectors determine which information can ultimately be used as input to the gated recurrent unit; and the update and reset gates can retain information from the time series for a long time, without being cleared over time or removed because it is irrelevant to the prediction.

[0108] Please see Figure 5 , Figure 5 This is a schematic diagram of an embodiment of the GRU model provided in this application. Figure 5 In this context, zt represents the update gate, and rt represents the reset gate. The update gate controls the extent to which state information from the previous time step is incorporated into the current time step; that is, it controls how much past information can be passed to the future, or how much information from the previous time step needs to be passed to the next time step. Generally, the larger the value of zt, the more state information from the previous time step is transmitted to the current time step. The reset gate mainly controls how much state information from the previous time step can be written into the candidate set ht of the current state.

[0109] For this application, the data input into the GRU model includes, but is not limited to: the date corresponding to each region, and the region attributes under a given date; wherein, the region attributes may include: region area, region function, etc. The state information is the result obtained after performing complex calculations on the aforementioned data and the functions in the GRU model.

[0110] In the embodiments of this application, the sizes of the update gate and the reset gate are both between 0 and 1. For the update gate, when the size of the update gate is 0, it means that the state information from the previous moment has not been incorporated into the current moment at all; while when the size of the update gate is 1, it means that the state information from the previous moment has been incorporated into the current moment completely. Generally speaking, the smaller the update gate, the less state information from the previous moment is incorporated into the state information of the current moment.

[0111] The reset gate is determined by comparing the current reset gate result with the previous hidden state h. t-1 Perform element-wise multiplication; if the result is close to 0, it means that the hidden state h from the previous time step needs to be discarded. t-1 If the calculation result is close to 1, then retain the hidden state h from the previous time step. t-1 The update gate, on the other hand, is formed by calculating the candidate hidden states. Compared to the candidate hidden state at the previous time step Combine the results to generate the hidden state h at the current time. t .

[0112] In the embodiments of this application, Xt represents the input of the GRU model at time t; σ and tanh are both correlation functions in the GRU neural network model. For the GRU model, the input of the GRU model is Xt and the candidate set h. t-1 The output of the GRU model is h. t .

[0113] Specifically, for the parcel volume prediction in this application, the parcel volume of the previous date is used as the hidden state h of the GRU model corresponding to the previous date. t The current date and region attributes are passed to the GRU model corresponding to the current date. The current date and region attributes serve as the input xt to the GRU model corresponding to the current date, along with the hidden state h from the previous date. t First, perform a calculation to obtain the candidate hidden state for the current date. Hide the candidate status for the current date The hidden state h under the previous date t-1 Perform calculations to obtain the hidden state h for the current date. t The model outputs the hidden state h for the current date. t This is the predicted value of the third quantity for the current date.

[0114] In some other embodiments of this application, the Prophet model is used as an example to predict the quantity of data. Prophet is an open-source data prediction tool developed by Facebook using Python and R. It is a process of predicting time series data based on nonlinear trends and additional models of annual, weekly, and daily seasonality as well as holiday effects. The basic model is as follows:

[0115] y(t)=g(t)+s(t)+h(t)+ε t

[0116] Where t represents time, g(t) is the growth function used to fit the non-periodic changes in the model, s(t) represents periodic changes, such as weekly, monthly, or yearly variations, and h(t) represents changes caused by special events such as holidays and festivals. ε t Noise represents random and unpredictable fluctuations.

[0117] In the above embodiments, the growth function g(t) can be:

[0118]

[0119] Where C represents the quantity of packages, k is the package growth rate (the rate of increase in the quantity of packages on the next date relative to the quantity on the previous date), and b is the offset, the value of which is usually randomly defined. From the formula, it can be determined that as time t increases, g(t) approaches C more closely; and the larger k is, the faster the growth rate of the package quantity. However, in real time series, the trend of the above function is not always constant. At certain specific times or with changes in certain potential cycles, the curve corresponding to the function may fluctuate significantly. Therefore, the point where the growth rate k suddenly changes, causing a significant fluctuation in the curve corresponding to the function, is defined as the change point.

[0120] In actual logistics scenarios, the growth rate k of parcel volume changes continuously over time. Therefore, an inflection point can be used to represent the node where the growth rate k of parcel volume changes; at this point, k indicates that the growth rate of parcels has undergone a sudden change. Furthermore, in actual logistics scenarios, there can be multiple inflection points, and within the time range between any two adjacent inflection points, the trend of the function curve will change according to the changes in the inflection points.

[0121] In the embodiments of this application, in a time series, when time t is greater than or equal to a certain timestamp s j At this point, the model will reach an inflection point:

[0122]

[0123] Since an inflection point has occurred, in order to make the above g(t) continuous, the above formula needs to be processed to obtain:

[0124]

[0125] Since there can be multiple inflection points, let's assume the timestamp corresponding to each inflection point is s. j δ l This represents timestamp s j The change in the growth rate over time; if we assume the initial growth rate is k, then at time s... j The growth rate at that time was:

[0126]

[0127] It should be noted that the functions mentioned above are all transition functions during the training process of the Prophet model. Finally, we can obtain:

[0128]

[0129] In the above embodiments, for s(t), a Fourier series can be used to represent the periodic change, that is, the Fourier series can be used to describe the periodic change in the number of express parcels over time, specifically as follows:

[0130]

[0131] In other embodiments, for variables such as holidays in the Prophet model, the same past and future holidays can be set as the same dummy variable, as follows:

[0132]

[0133] In this system, the same holiday is set as the same dummy variable, while different holidays correspond to different dummy variables. For example, the Spring Festival (Chinese New Year) is defined as the same dummy variable regardless of when it occurs; while the Spring Festival and National Day correspond to different dummy variables. Typically, multiple holidays throughout the year can be set as different dummy variables Di according to their chronological order.

[0134] If the current time t is a holiday, then Di is 1; otherwise, Di is 0. To represent the holiday effect, a corresponding indicator function is needed, along with a parameter k. i This represents the scope of the impact of holidays. The parameter k... i Satisfies a normal distribution: k ~ Normal(0, v) 2Furthermore, this normal distribution is affected by the holiday effect (holidays_prior_scale). The larger the value of the holiday effect, the greater the impact of holidays on the model; the smaller the value, the smaller the effect of holidays on the model.

[0135] Among them, different holidays affect parameter k i The impact also varies; that is, the magnitude of the holiday effect differs for different holidays. For example, the volume of express deliveries surges during Singles' Day (November 11th), while the volume of express deliveries on a regular weekend remains at an average level. Furthermore, in the embodiments of this application, the value corresponding to the holiday effect can be set manually; however, the value of the holiday effect for different holidays is usually different. In one specific embodiment, the value of the holiday effect can be 15.

[0136] In the above embodiment, it is necessary to first determine whether the current date t is a holiday; if it is a holiday, then Di can be set to 1; then, the parameter k can be used to determine the date. i The normal distribution it follows determines the impact of holidays on express delivery forecasting. If the current date t does not fall within a holiday period, then Di is 0; since the above function is a function of the impact of holidays, if the current period does not fall within a holiday period, it will not have any impact on express delivery volume forecasting, therefore (t) is 0.

[0137] In the above embodiments, the functions corresponding to g(t), s(t), and h(t), as well as the range of values ​​of the relevant parameters in the functions, can be obtained to obtain the quantity prediction function corresponding to the Prophet model; further, the third quantity prediction value of the target area in the first time period under the prediction of the Prophet model can be obtained.

[0138] In other embodiments, taking the CatBoost model as an example, the CatBoost model can be trained by using parameters such as AOI region type, AOI region area, maximum, minimum and average number of items in the AOI region as inputs.

[0139] Here, AOI (area of ​​interest), also known as information surface or interest surface, refers to a regional geographic entity in map data, meaning a clearly defined area such as a neighborhood, work park, school, or commercial area. In the embodiments of this application, it can be the aforementioned target area. That is, parameters such as the target area's area type, area, maximum, minimum, and average item quantity are used as inputs to the CatBoost model to train it, obtaining a third item quantity prediction value for the target area using the CatBoost model.

[0140] CatBoost is a machine learning method based on Gradient Boosting Decision Trees. It requires fewer parameters and supports categorical variables. CatBoost effectively addresses gradient bias and prediction shift, reducing overfitting and improving accuracy and generalization. It uses a fully symmetric tree as its base model and incorporates an innovative algorithm that automatically converts categorical features into numerical features. By statistically analyzing categorical features, it calculates the frequency of a particular category and adds hyperparameters to generate new numerical features. Furthermore, CatBoost utilizes combined categorical features, leveraging the relationships between different categories to significantly enrich the feature dimensions.

[0141] In the embodiments of this application, parameters such as date, region attributes, and region area are input as training features into the CatBoost model, and the predicted quantity is calculated according to the CatBoost model algorithm. The prediction process using the CatBoost model can refer to existing technologies and is not limited here.

[0142] In one specific embodiment, such as Figure 6 The diagram shown is a schematic flowchart illustrating the process of obtaining the predicted value of the second quantity in an embodiment of this application. Figure 6 In this study, three models are used for quantity prediction: CatBoost, GRU, and Prophet. Each of these three models yields three third-order quantity predictions. Simultaneously, the average quantity over a period preceding the date to be predicted is obtained. Using the average quantity and the three third-order quantity predictions as the four input features of the initial regression function, the final second-order quantity prediction for the target region within the first time period is obtained.

[0143] exist Figure 6 In this study, the CatBoost, GRU, and Prophet models all utilize raw data from a past period to train different feature values ​​for subsequent regression training. Additionally, it's necessary to obtain the average number of parcels received three days prior to the prediction date from the raw data; similarly, feature values ​​are extracted from this average parcel volume and used as input to the initial regression function for subsequent regression prediction.

[0144] The initial regression function can be the Logistic function. However, when using the initial regression function for prediction, it is actually necessary to predict the receipt and delivery separately, and finally obtain the predicted values ​​of the receipt and delivery respectively.

[0145] In a specific embodiment of this application, the final quantity prediction expression obtained using the above three quantity prediction models and the Logistic regression function can be:

[0146]

[0147] Where w GRU GRU(x) is the predicted value of the third quantity obtained by the GRU model, w Propet Propet(x) is the predicted value of the third item quantity obtained from the Propet model, w Catboost Catboost(x) is the third quantity prediction value obtained by the Catboost model; while b is the offset to avoid meaningless situations in the denominator of f(x).

[0148] It should be noted that after calculating f(x), f(x) needs to be inversely sigmoidized, and the final result is the predicted value of the second quantity.

[0149] In the embodiments of this application, the specific process of obtaining different third quantity prediction values ​​using different quantity prediction models can be referred to the prior art, and no limitation is made here.

[0150] In the embodiment of the application, since the predicted number of parcels is strongly correlated with the final allocation result of the courier, in order to reduce the error between the two, when training and fitting the initial regression function Logistics, the predicted value of the second parcel quantity is used as a condition for dynamic programming to schedule and allocate the courier.

[0151] For delivery drivers, each driver incurs a minimum guaranteed cost and a delivery cost. When the delivery capacity of a fixed area's driver allocation doesn't meet the actual volume demand, excess costs are incurred. In other words, each driver incurs costs, including but not limited to minimum guaranteed costs, delivery costs, and excess costs. The target area is constrained by the volume of packages, and each driver also incurs costs. To minimize the total driver cost for the target area, drivers need to be allocated more accurately based on the volume of packages in that area. In the embodiments of this application, dynamic programming can be used to determine the number of drivers for the target area.

[0152] In the embodiments of this application, dynamic programming is actually used to calculate the courier cost corresponding to different numbers of couriers and different predicted package volumes, compare different costs to determine the lowest cost, and further determine the courier configuration corresponding to the lowest cost.

[0153] In one embodiment, such as Figure 7 As shown, the step of obtaining the predicted values ​​of express mail attributes for the target area within the first time period includes:

[0154] S701. Obtain the maximum number of packages received, the maximum number of packages delivered, and the courier attributes for each courier.

[0155] In the embodiments of this application, dynamic programming can be used to calculate the courier allocation for a target area under constraints of parcel volume and cost. Dynamic programming is a process of optimizing a decision-making process; in this application, it is the process of finding the optimal courier allocation.

[0156] For the dynamic programming solution in this embodiment, the maximum number of packages received, the maximum number of packages delivered, and the courier cost for each courier are the constraints for the dynamic programming solution. The courier attribute is the courier cost.

[0157] S702. Obtain the state equation corresponding to the preset express delivery cost prediction.

[0158] Dynamic programming primarily uses state equations for calculation and solution, thus providing the state equations corresponding to express delivery cost prediction. These state equations are an intermediate step in the dynamic programming solution process.

[0159] S703. Substitute the predicted value of the second item quantity, the maximum number of items received, the maximum number of items delivered, and the courier attributes into the state equation to calculate the predicted value of the item attributes.

[0160] In the embodiments of this application, in addition to the maximum number of packages collected, the maximum number of packages delivered, and the courier cost for each courier, the number of packages collected and delivered in the target area are also constraints in the dynamic programming solution. That is, the total number of packages collected by all couriers will not exceed the total number of packages collected in the target area; the total number of packages delivered by all couriers will not exceed the total number of packages delivered in the target area. When calculating the predicted cost value corresponding to the predicted package volume, the predicted second package volume value is used as the total number of packages collected and delivered in the target area.

[0161] The predicted second number of packages, the maximum number of packages each courier can collect, the maximum number of packages each courier can deliver, and the courier cost are all relevant parameters in the state equation. Substituting these parameters into the state equation allows us to solve for the optimal courier configuration scheme.

[0162] In one embodiment, the state equation corresponding to dynamic programming can be:

[0163]

[0164] Specifically, assume that the target area has two-dimensional attributes [u, v] representing the total number of packages received and the total number of packages delivered in the target area, respectively; and each courier also has three-dimensional attributes [c, d, m] representing the maximum number of packages received, the maximum number of packages delivered, and the courier cost, respectively.

[0165] In the above state equation, dp[i][u][v] represents the minimum cost when the current i couriers are assigned to the target area, and the total number of packages received (u) and the total number of packages delivered (v) in the target area are determined. Furthermore, the number of couriers assigned to the target area is also determined at the same time as the cost.

[0166] For the above state equation, generally speaking, for each given number of couriers i, dp[i][u][v] is a step-like monotonically non-decreasing function of variables u and v. For each given i, there is usually a corresponding jump point, which divides the monotonically non-decreasing function into multiple segments. Generally, the jump point is uniquely determined, but it may repeat under different i.

[0167] If a static array is used to store jump points, a large static array needs to be defined in advance to avoid jump point overflow, which would result in a significant waste of space resources. Therefore, in the embodiments of this application, the data results of a dynamic linked list can be used to store jump points; the main task is to add non-repeating jump points to the dynamic linked list and mark each non-repeating jump point to facilitate the determination of whether subsequent jump points are repeated.

[0168] Specifically, in the process of solving the state equation, only the values ​​of the state equation before the jump point are retained; when a new jump point appears, we add the new jump point to the dynamic linked list and mark it as a Boolean variable. If the same jump point appears later, it is not stored, and the value of the new state equation generated after the jump point overwrites the previous value, thereby saving space.

[0169] In the embodiments of this application, a dynamic linked list storage structure is used to store jump points, which can effectively store the complexity of the storage space and avoid memory overflow. It should be noted that in the embodiments of this application, the above state equation is only the state transition equation in the dynamic programming solution process, and the result obtained by solving the state transition equation is the minimum cost.

[0170] In the above embodiments, dynamic programming can be used to determine the predicted and actual costs of express delivery. This requires simply replacing [u, v] in the state equation with the predicted and actual quantities of the second shipment. Specifically, replacing [u, v] in the state equation with the predicted receipt and delivery values ​​in the predicted second shipment quantity allows for the calculation of the predicted cost corresponding to the predicted second shipment quantity; replacing [u, v] with the actual quantity allows for the calculation of the actual cost.

[0171] Specifically, when calculating the true value of express delivery cost, it is necessary to first obtain the true value of the number of packages in the second time period, and then substitute the true value of the number of packages in the second time period, the maximum number of packages received by the courier, the maximum number of packages delivered, and the courier cost into the state equation to calculate the true value of the express delivery attributes in the target area in the second time period; that is, to obtain the true value of the express delivery cost in the target area in the second time period.

[0172] In the embodiments of this application, the specific steps for calculating the predicted value of express delivery cost and the actual value of express delivery cost using dynamic programming methods can be referred to the prior art, and no limitation is made here.

[0173] After obtaining the predicted and actual costs of a shipment, the cost error between them can be determined. Specifically, the absolute value of the difference between the two can be taken as the cost error.

[0174] To avoid excessive discrepancies between the predicted and actual costs of parcels, in this embodiment, the cost error can be used as the training set for an initial regression function to obtain a new initial regression function, which in turn yields a new second predicted parcel quantity. When the error between the predicted parcel cost corresponding to the new second predicted parcel quantity and the actual parcel cost corresponding to the actual parcel cost reaches its minimum, the new second predicted parcel quantity becomes the final predicted first parcel quantity. At this point, couriers can be assigned based on the final predicted first parcel quantity.

[0175] Typically, the training set corresponding to a regression function usually includes multiple data points; in the embodiments of this application, the training set corresponding to the regression function Logistic usually includes multiple cost errors. Therefore, in the embodiments of this application, it is necessary to obtain multiple predicted part quantities and calculate between them and multiple actual part quantities to obtain multiple cost errors.

[0176] like Figure 8 The diagram shown is a schematic representation of the quantity prediction provided in an embodiment of this application. Figure 8In this context, dynamic programming can be used to determine the actual cost of express delivery corresponding to the actual number of packages received and delivered (i.e., the actual value of express delivery); similarly, dynamic programming can be used to determine the predicted cost of express delivery corresponding to the predicted number of packages received and delivered (i.e., the second predicted value of express delivery) obtained using multiple prediction models.

[0177] After obtaining two costs using dynamic programming, the express cost error between the two costs can be determined. Error analysis can then be performed on the express cost error to determine whether the current cost error meets the preset requirements; that is, whether the loss function corresponding to the regression function meets the preset requirements.

[0178] If the cost error obtained at this point does not meet the preset requirements, the new predicted receipt quantity and predicted delivery quantity are recalculated using the regression function, and the cost error is recalculated to perform error analysis again; this continues until the cost error meets the preset requirements. The regression function obtained at this point is the final required regression function, and the predicted receipt quantity and predicted delivery quantity predicted by the regression function are the final predicted receipt quantity and predicted delivery quantity.

[0179] It should be noted that, in the embodiments of this application, the cost error between the predicted cost and the actual cost is actually used as the training set for the initial regression function. The initial regression function is trained multiple times until the loss function corresponding to the initial regression function meets the preset requirements. During the training of the initial regression function, a new regression function is obtained after each training iteration. If the loss function corresponding to the regression function does not meet the preset requirements, a new predicted value for the parcel volume is predicted using the new regression function. This prediction is then used to obtain the predicted cost for the parcel volume, which is then compared with the actual cost to obtain the cost error. In other words, in the embodiments of this application, the cost error for each iteration of the regression function training is different.

[0180] In the embodiments of this application, the first, second, and third predicted express delivery values ​​all include both receipt and delivery; that is, when predicting express delivery, receipt and delivery can be predicted separately. The final second predicted express delivery value can also actually include both receipt and delivery predicted values.

[0181] Of course, in other embodiments of this application, only the number of packages received or only the number of packages delivered can be predicted. Only the number of packages received or delivered needs to be used as input to the initial regression function; it is not necessary to input both the number of packages received and the number of packages delivered into the initial regression function.

[0182] To better implement the express delivery volume prediction method in the embodiments of this application, based on the express delivery volume prediction method, the embodiments of this application also provide an express delivery volume prediction device, such as... Figure 9 As shown, the express shipment volume prediction device 900 includes:

[0183] The first acquisition module 910 is used to acquire the predicted values ​​of express mail attributes in the target area within the first time period;

[0184] The second acquisition module 920 is used to acquire the actual values ​​of the express mail attributes of the target area within the second time period.

[0185] Error calculation module 930 is used to determine the error of express delivery attributes based on the predicted value of the express delivery attributes and the actual value of the express delivery attributes;

[0186] The shipment quantity prediction module 940 is used to determine the first shipment quantity prediction value of the target area within a first time period based on the shipment attribute error; wherein the first time period is after the second time period.

[0187] In some embodiments of this application, the first acquisition module 910 is further configured to acquire the actual quantity value of the shipment within a third time period, the third time period being before the first time period; based on a plurality of preset shipment quantity prediction models and the actual quantity value of the shipment within the third time period, acquire the third shipment quantity prediction value corresponding to each of the plurality of shipment quantity prediction models, thereby obtaining a plurality of third shipment quantity prediction values; based on the plurality of third shipment quantity prediction values, obtain the second shipment quantity prediction value within the first time period; and based on the second shipment quantity prediction value, calculate the predicted value of the express shipment attributes.

[0188] In some embodiments of this application, the first acquisition module 910 is further configured to acquire an initial regression function for the preset quantity prediction; acquire the average quantity of parts within a second time period; and use multiple third quantity prediction values ​​and the average quantity as feature inputs to the initial regression function to obtain the second quantity prediction value within a first time period.

[0189] In some embodiments of this application, the first acquisition module 910 is further configured to acquire the maximum number of packages received, the maximum number of packages delivered, and the courier attributes corresponding to each courier; acquire the state equation corresponding to the preset package attribute prediction; and substitute the second package quantity prediction value, the maximum number of packages received, the maximum number of packages delivered, and the courier attributes corresponding to each courier into the state equation to calculate the package attribute prediction value.

[0190] In some embodiments of this application, the second acquisition module 920 is further used to substitute the actual value of the number of packages, the maximum number of packages received by the courier, the maximum number of packages delivered, and the courier attributes into the state equation to calculate the actual value of the package attributes.

[0191] In some embodiments of this application, the express delivery volume prediction module 940 is further configured to obtain multiple true values ​​of express delivery attributes in the target area within a second time period to obtain multiple express delivery attribute errors; use the multiple express delivery attribute errors to train an initial regression function to obtain a target regression function; and use the target regression function to calculate a first predicted volume value.

[0192] In some embodiments of this application, the express delivery volume prediction module 940 is further configured to train an initial regression function using multiple express delivery attribute errors to obtain a trained regression function; determine whether the loss function corresponding to the trained regression function meets preset requirements; if the loss function does not meet preset requirements, calculate a new express delivery attribute error using the trained regression function; and continue training the regression function using the new express delivery attribute error until the loss function meets preset requirements to obtain the target regression function.

[0193] In the above embodiments, the express delivery volume prediction device acquires the predicted values ​​of express delivery attributes for the target area and the actual values ​​of express delivery attributes corresponding to the actual express delivery volume over a historical period. It further obtains the express delivery attribute error between the actual express delivery attribute value and the express delivery cost attribute value, and then re-predicts the express delivery volume for the target area based on this error. By utilizing the express delivery attribute error between the actual and predicted express delivery attribute values, the predicted express delivery volume is further obtained, thus improving the accuracy of express delivery prediction.

[0194] In some embodiments of this application, the express delivery volume prediction device 900 can be implemented as a computer program, which can be implemented in, for example... Figure 10 The computer device shown operates on this device. The computer device's memory can store the various program modules that make up the express delivery volume prediction device 900, for example, Figure 9 The first acquisition module 910, the second acquisition module 920, the error calculation module 930, and the shipment volume prediction module 940 are shown. The computer program comprised of these modules causes the processor to execute the steps in the express shipment volume prediction methods of the various embodiments of this application described in this specification.

[0195] For example, Figure 10 The computer equipment shown can be used as follows Figure 9The first acquisition module 910 in the express delivery volume prediction device 900 shown executes step S201. The computer device can execute step S202 via the second acquisition module 920. The computer device can execute step S203 via the error calculation module 930. The computer device can execute step S204 via the volume prediction module 940. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used to communicate with external computer devices via a network connection. When the computer program is executed by the processor, it implements an express delivery volume prediction method.

[0196] Those skilled in the art will understand that Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specifically, the computer device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0197] In some embodiments of this application, a computer device is provided, including one or more processors; a memory; and one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the processors as described in the express delivery volume prediction method. The steps of the express delivery volume prediction method here may be steps from the express delivery volume prediction methods of the various embodiments described above.

[0198] In some embodiments of this application, a computer-readable storage medium is provided, storing a computer program. The computer program is loaded by a processor, causing the processor to execute the steps of the aforementioned express delivery volume prediction method. The steps of this express delivery volume prediction method may be those found in the express delivery volume prediction methods of the various embodiments described above.

[0199] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0200] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0201] The foregoing has provided a detailed description of a method, apparatus, computer device, and storage medium for predicting the volume of express parcels provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. 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 the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for predicting the volume of express parcels, characterized in that, include: Obtain the predicted values ​​of parcel attributes for the target area within the first time period; Obtain the actual values ​​of the parcel attributes in the target area during the second time period; The error in the express delivery attribute is determined based on the predicted value of the express delivery attribute and the actual value of the express delivery attribute. Based on the express delivery attribute error, determine the first predicted quantity of the first shipment in the target area within the first time period; wherein, the first time period is after the second time period; Specifically, multiple third-order quantity prediction values ​​are obtained through multiple quantity prediction models and the actual quantity values ​​within the third time period. An initial regression function for the preset quantity prediction and the average quantity within the second time period are obtained. The average quantity and multiple third-order quantity prediction values ​​are used as feature inputs to the initial regression function to obtain the second-order quantity prediction value within the first time period. The second-order quantity prediction value, the maximum number of packages received, the maximum number of packages delivered, and the courier attributes are substituted into the state equation to calculate the predicted value of the package attributes. The third time period is before the first time period. The package attributes include the cost corresponding to the package.

2. The method according to claim 1, characterized in that, The step of obtaining the actual values ​​of the parcel attributes in the target area within the second time period includes: Obtain the actual quantity of items within the second time period; Substituting the actual number of parcels received, the maximum number of parcels delivered, and the courier's attributes into the state equation during the second time period, the actual parcel attribute values ​​for the target area during the second time period are calculated.

3. The method according to claim 1, characterized in that, The step of determining the predicted quantity of the first shipment in the target area within the first time period based on the shipment attribute error includes: Obtain the true values ​​of express mail attributes for the target area within multiple second time periods to obtain multiple express mail attribute errors; By utilizing the multiple express delivery attribute errors, the initial regression function is trained to obtain the target regression function; The predicted value of the first quantity is calculated using the target regression function.

4. The method according to claim 3, characterized in that, The step of training the initial regression function using the multiple express delivery attribute errors to obtain the target regression function includes: The initial regression function is trained using the multiple express delivery attribute errors to obtain the trained regression function; Determine whether the loss function corresponding to the trained regression function meets the preset requirements; If the loss function does not meet the preset requirements, the new express delivery attribute error is calculated using the trained regression function; The regression function is trained using the new express delivery attribute error until the loss function meets the preset requirements to obtain the target regression function.

5. A device for predicting the volume of express parcels, characterized in that, include: The first acquisition module is used to acquire the predicted values ​​of express mail attributes in the target area within the first time period. The second acquisition module is used to acquire the actual values ​​of the express mail attributes of the target area within the second time period. The error calculation module is used to determine the error of the express delivery attribute based on the predicted value of the express delivery attribute and the actual value of the express delivery attribute. The shipment volume prediction module is used to determine the first shipment volume prediction value of the target area within the first time period based on the shipment attribute error; wherein the first time period is after the second time period. Multiple third-order quantity prediction values ​​are obtained by using multiple quantity prediction models and the actual quantity values ​​within the third time period. Second-order quantity prediction values ​​within the first time period are obtained by using a preset initial regression function for quantity prediction and the average quantity value within the second time period. Item attribute prediction values ​​are obtained by using a state equation based on the maximum number of items received, the maximum number of items delivered, and the attributes of each courier. The third time period is before the first time period. Item attributes include the cost corresponding to the item.

6. A computer device, characterized in that, The computer device includes: One or more processors; The 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 express delivery volume prediction method according to any one of claims 1 to 4.

7. 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 express delivery volume prediction method according to any one of claims 1 to 4.

Images