Method and apparatus for forecasting delivery cost for predetermined order for cargo delivery service and determining dynamic discount based on batch processing likelihood prediction

By using multiple algorithms and the sigmoid function to predict scheduled delivery costs and batch processing rates, the accuracy issues of scheduled delivery costs and discount mechanisms are resolved, thereby improving resource utilization efficiency and customer satisfaction.

CN118946900APending Publication Date: 2024-11-12GRABTAXI HOLDINGS PTE LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202380030125.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-04-22
Filing Date
2023-03-21
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

Existing technologies struggle to accurately predict pre-delivery costs and design effective discount mechanisms, leading to improper customer behavior and inefficient resource utilization.

Method used

Algorithms such as quantile regression, linear regression, lasso regression, support vector regression, multilayer perceptron neural network, and long short-term memory neural network are used, combined with the S-shaped sigmoid function, to predict delivery costs and batch processing rates, and dynamically adjust discounts to optimize pre-delivery costs.

Benefits of technology

It enables accurate prediction of scheduled delivery costs and precise forecasting of batch processing rates, provides a flexible discount mechanism, and improves resource utilization efficiency and customer satisfaction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118946900A_ABST
    Figure CN118946900A_ABST
Patent Text Reader

Abstract

Aspects relate to a method for forecasting a delivery cost for a predetermined order for a cargo delivery service and determining a dynamic discount based on a batch processing likelihood prediction, the method comprising: predicting a delivery cost for a predetermined delivery in an available time period; and predicting a batch processing rate at which the predetermined delivery is to be subjected to batch processing with at least one other order in the available time period. The method further includes determining a discount for the predetermined delivery in an available time period based on the predicted batch processing rate; based on the predicted delivery cost and the determined discount, determining a final delivery cost of the predetermined delivery in the available time period; and displaying the determined final delivery cost for the user to select the predetermined delivery.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] Various aspects of the present disclosure relate to methods and apparatus for forecasting delivery costs for scheduled orders of goods delivery services and determining dynamic discounts based on batch processing likelihood predictions. Background Art

[0002] Scheduled delivery is a service delivery provided to customers who may not need the service delivery urgently and provides customers with more flexibility to schedule the service delivery in advance. For example, a customer can order a scheduled delivery for his dinner or grocery delivery on his way home.

[0003] Customers may be allowed to order scheduled deliveries for weeks, as multiple delivery time slots may be available during each day. Delivery charges may be calculated for multiple time slots at the same time, not just for a single time slot. There are a few things to consider when calculating delivery charges. First, the delivery charges calculated for a given time slot should not be too far away from the delivery charges for live orders for the same time slot. This is to avoid gaming behavior by customers. Also, customers should be encouraged to place orders during off-peak hours whenever possible, so that delivery drivers can be fully utilized when they are less occupied. Furthermore, customers should be pushed to place orders during certain time slots, so that order batching rates can be increased to improve network efficiency.

[0004] However, with the above considerations come some challenges. First, the delivery costs for real-time orders may need to be forecasted several days in advance. The accuracy of forecasting these delivery costs is a big challenge because real-time signals for understanding market conditions may not be available. In addition, on top of the cost forecast, a discount mechanism may need to be designed, which can be quite complex.

[0005] To address the above challenges, a naive approach may be to assume that future delivery costs and batch processing rates are the same or very close to their historical distributions, thereby calculating the mean of future delivery costs and batch processing rates in a specific area and a specific time period, and giving a discount based on the calculated mean. However, the performance of this discount depends heavily on the forecast accuracy, which may not be met by the naive approach. Summary of the invention

[0006] Various embodiments relate to a method for forecasting delivery costs for scheduled orders of a goods delivery service and determining a dynamic discount based on a prediction of batch processing possibility, the method comprising: predicting the delivery costs of scheduled deliveries in an available time period; and predicting a batch processing rate at which the scheduled deliveries are to be batch processed with at least one other order in the available time period. The method further comprises: determining a discount for the scheduled deliveries in the available time period based on the predicted batch processing rate; determining a final delivery cost for the scheduled deliveries in the available time period based on the predicted delivery costs and the determined discount; and displaying the determined final delivery cost for a user to select the scheduled deliveries.

[0007] Predicting delivery costs may include: obtaining the coordinates of the starting point of the scheduled delivery, the coordinates of the destination of the scheduled delivery, the available time period, the basic fee or tiered basic fee for the scheduled delivery, and the range of surge rates; and predicting the delivery cost of the scheduled delivery in the available time period based on the obtained coordinates of the starting point, the obtained coordinates of the destination, the obtained available time period, the obtained basic fee or tiered basic fee, and the obtained range of surge rates.

[0008] Delivery costs can be predicted using at least one of the following: quantile regression, linear regression, lasso regression, support vector regression, multilayer perceptron neural network, long short-term memory neural network, and decision tree-based algorithm.

[0009] Each of the quantile regression neural networks may include a multi-hidden layer feed-forward neural network trained using historical data and quantile loss.

[0010] Predicting the batch processing rate may include obtaining: the coordinates of the starting point of the scheduled delivery, the coordinates of the destination of the scheduled delivery, the available time period, the number of completed orders in the area of ​​the scheduled delivery and during the same time period as the available time period in the previous week(s), and the batch processing rate of the completed orders; and based on the obtained coordinates of the starting point, the obtained coordinates of the destination, the obtained available time period, the obtained number of completed orders, and the obtained batch processing rate of completed orders, predicting the batch processing rate of the scheduled deliveries to be batched with at least one other order in the available time period.

[0011] Predicting a batching rate may also include determining a bearing for the scheduled delivery, the bearing representing a direction between an origin and a destination of the scheduled delivery, and may also predict a batching rate at which the scheduled delivery is to be batched with at least one other order in an available time period based on the determined bearing.

[0012] Predicting the batching rate also includes determining a number of completed orders in the area of ​​the scheduled delivery, having the determined location, and during the same time period as the available time period in the previous week, and determining a batching rate for the completed orders. The batching rate at which the scheduled delivery is to be batched with at least one other order in the available time period may also be predicted based on the determined number of completed orders and the determined batching rate for the completed orders.

[0013] Based on the available time period including off-peak times, the predicted batch processing rate may be lower, and based on the available time period including peak times, the predicted batch processing rate may be higher.

[0014] The batch rate may be predicted using a neural network with a long short-term memory (LSTM) layer trained using historical data.

[0015] Determining the discount may include obtaining: a predicted batch processing rate, coordinates of a starting point for scheduled delivery, coordinates of a destination for scheduled delivery, an available time period, a distance for scheduled delivery, and a duration for scheduled delivery; and determining a discount for scheduled delivery in the available time period based on the obtained predicted batch processing rate, the obtained coordinates of the starting point, the obtained coordinates of the destination, the obtained available time period, the obtained distance for scheduled delivery, and the obtained duration for scheduled delivery.

[0016] The discount may be determined using an S-shaped sigmoid function that has the predicted batch rate as input and the discount as output.

[0017] The S-shaped sigmoid function can include where a, b, c, d and e are coefficients and BR is the predicted batch rate.

[0018] Each of the coefficients a, b, c, d and e is a function of the length of the available time period.

[0019] Based on a shorter available time period, the determined discount may be smaller, and based on a longer available time period, the determined discount may be larger.

[0020] Based on the predicted batch processing rate being less than a predetermined minimum value, the determined discount may be a preset minimum value, and based on the predicted batch processing rate being greater than or equal to a predetermined maximum value, the determined discount may be a preset maximum value.

[0021] The final delivery fee may be determined using an equation including Final Fare = Predicted Fare x (1 - Discount), where Predicted Fare is the predicted delivery fee and Discount is the determined discount.

[0022] The determined final delivery cost may be displayed on the display along with the available time periods and highlighted when selected by the user.

[0023] A server may be configured to perform the method.

[0024] A computer program element may comprise program instructions which, when executed by one or more processors, cause the one or more processors to perform the method.

[0025] A computer readable medium may include program instructions that, when executed by one or more processors, cause the one or more processors to perform the method. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The invention will be better understood with reference to the detailed description when considered in conjunction with the non-limiting examples and accompanying drawings, in which:

[0027] [ Figure 1 ] shows a schematic diagram illustrating a communication device using a goods delivery service, the communication device including a smartphone and a server;

[0028] [ Figure 2 ] shows a block diagram of an architecture for forecasting delivery costs for scheduled orders of goods delivery services and determining dynamic discounts based on batch processing possibility predictions according to an embodiment;

[0029] [ Figure 3 ] shows a graph illustrating a mapping function between a predicted batching rate as an input and a discount as an output according to an embodiment;

[0030] [ Figure 4 ] shows a flow chart of a method for forecasting delivery costs for scheduled orders for goods delivery services and determining dynamic discounts based on batch processing possibility predictions according to an embodiment; and

[0031] [ Figure 5 ] shows the implementation of [ Figure 2 ] of the architecture Figure 1 ]block diagram of the server. DETAILED DESCRIPTION

[0032] The following specific embodiments refer to the accompanying drawings, which illustrate the specific details and embodiments in which the present disclosure can be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the present disclosure. Other embodiments may also be utilized and structural and logical changes may be made without departing from the scope of the present disclosure. Various embodiments are not necessarily mutually exclusive, as some embodiments may be combined with one or more other embodiments to form new embodiments.

[0033] Embodiments described in the context of one of the apparatus or method are analogously valid for the other apparatus or method. Likewise, embodiments described in the context of the apparatus are analogously valid for the vehicle or method, and vice versa.

[0034] Features described in the context of one embodiment may be correspondingly applicable to the same or similar features in other embodiments. Features described in the context of one embodiment may be correspondingly applicable to other embodiments, even if not explicitly described in these other embodiments. In addition, additions and / or combinations and / or substitutions as described for features in the context of one embodiment may be correspondingly applicable to the same or similar features in other embodiments. In the context of various embodiments, the articles "a", "an" and "the" as used with respect to features or elements include references to one or more features or elements in the features or elements.

[0035] As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0036] Hereinafter, embodiments will be described in detail.

[0037] Goods delivery applications commonly used on smartphones allow their users to order delivery of goods (e.g., food and / or packages) via his or her smartphone.

[0038] [ Figure 1 ] shows a schematic diagram of a communication device 100 for using a cargo delivery service, the communication device comprising a smartphone 105 and a server 115 (computer).

[0039] Goods delivery services may provide for the delivery of goods, which may include, for example, food, packages, groceries, etc.

[0040] Smartphone 105 includes a screen showing a graphical user interface (GUI) 106 of a goods delivery application that the user of smartphone 105 previously installed on his or her smartphone and opened or started to order a delivery of goods.

[0041] GUI 106 includes a box for a starting point 107 for the delivery of goods, which the user can touch to enter the starting point 107, for example, to open a list of possible starting points (including, for example, grocery stores and / or restaurants). GUI 106 also includes a box for a destination 108 for the delivery of goods, which the user can touch to enter the destination 108, for example, to open a list of possible destinations (including, for example, the user's home or office). There may also be a menu (not shown) that allows the user to select various options, such as a payment method (cash, credit card, credit balance of the goods delivery service). When the user selects a destination, GUI 106 shows a list of selectable dates 109 during which the delivery of goods can be scheduled and a list of selectable time periods 110 during which the delivery of goods can be scheduled. As described below, the delivery cost for each selectable time period 110 is determined and displayed. In this example, as shown in [ Figure 1 ], select "Best" from the list of selectable dates 109, and select "Today, free from 8 to 9 a.m." from the list of selectable time periods 110. Once one of the selectable time periods 110 is selected, an order for a goods delivery may be initiated by sending the order from the smartphone 105 to a server 115 of the goods delivery service.

[0042] For the above, the cargo delivery application communicates with the server 115 of the cargo delivery service via a radio connection. The server 115 can query the memory or data storage device 121 of the server 115, which stores information about the current location of the registered delivery vehicles 111, when they are expected to be idle, traffic jams, etc. As a result, the processor of the server 115 selects the most suitable vehicle (if available, that is, if the order can be completed), and can provide an estimate of the time when the vehicle will pick up the goods at the starting point 107, the price of the goods delivery, and how long it will take to reach the destination 108. The server 115 transmits this information back to the smartphone 105, and the smartphone 105 can display this information on the GUI 106. The user can then accept (that is, arrange the order) by touching the corresponding button. If the user accepts, the server 115 notifies the selected one of the vehicles 111 (or, equivalently, its driver), that is, the vehicle assigned by the server 115 to complete the cargo delivery.

[0043] It should be noted that although server 115 is described as a single server, its functionality (e.g., for providing cargo delivery services for an entire city) is typically provided by an arrangement of multiple server computers in actual applications (e.g., implementing a cloud service). Therefore, the functionality described below provided by server 115 may be understood to be provided by an arrangement of servers or server computers.

[0044] The data storage device 121 may, for example, be part of a cloud-based system 120 provided by a cloud storage provider for storing and accessing data that it can use to make decisions, such as information about the location of the origin, destination and vehicles, their history (earlier bookings and routes taken), etc.

[0045] Server 115 provides cargo delivery service together with vehicle 111. It should be noted that although [ Figure 1 ]'s example relates to a cargo delivery service in which cargo is transported and delivered, but a transportation system that provides a transportation service for transporting people may be similarly provided.

[0046] When a user orders a shipment of goods, server 115 may provide smartphone 105 with an estimate of when the shipment will be completed, e.g., when the shipment will be delivered to the user.

[0047] Embodiments described herein include a hybrid dynamic discount system for scheduled orders of goods delivery services. First, embodiments can use quantile regression models as a convenient tool to capture the long-term trend of delivery costs to adjust the forecast of delivery costs in a proportional and reasonable manner. Moreover, embodiments can accurately predict the batch processing probability of scheduled orders and provide dynamic discounts for scheduled orders based on the predicted batch processing probability.

[0048] [ Figure 2 ] shows a block diagram of an architecture 200 for forecasting delivery costs for scheduled orders for goods delivery services and determining dynamic discounts based on batch processing possibility predictions according to an embodiment.

[0049] refer to[ Figure 2 ], architecture 200 includes a cost prediction model 205, a batch processing possibility prediction model 210 and a discount model 215.

[0050] In the goods delivery service, users can request or order ASAP delivery or scheduled delivery. Scheduled delivery can include advance same-day delivery or future delivery several days in advance. Scheduled delivery can also have a longer delivery time period (cycle) than ASAP delivery.

[0051] When a user is ordering a scheduled delivery of goods, the goods delivery service may need to provide delivery fees and discounts (rates) for each available time period via the goods delivery application, for example, as shown in [ Figure 1 ] is shown in the GUI 106 of FIG. The architecture 200 provides a final delivery cost for each available time period based on the delivery cost and discount for each available time period.

[0052] At a high level, the architecture 200 can determine the final delivery costs for scheduled deliveries in an available time period by performing at least the following operations. First, the architecture 200 can predict the delivery costs for scheduled deliveries in the available time period. Next, the architecture 200 can predict the batching rate for scheduled deliveries in the available time period. Then, the architecture 200 can determine a discount for scheduled deliveries in the available time period based on the predicted batching rate. Finally, the architecture 200 can determine the final delivery costs based on the predicted delivery costs and the determined discount, and can display the determined final delivery costs for the user to select a scheduled delivery.

[0053] In detail, the cost prediction model 205 obtains the coordinates (e.g., latitude and longitude) of the starting point (e.g., merchant) of the scheduled delivery, the coordinates (e.g., latitude and longitude) of the destination of the scheduled delivery, the available time period (day of the week, hour and minute of the day) of the scheduled delivery, the basic fee or graded basic fee of the scheduled delivery, and the surge rate range (between the minimum and maximum surge rates). The cost prediction model 205 can also obtain or determine the distance of the scheduled delivery, the duration of the scheduled delivery, the merchant type of the starting point of the scheduled delivery, the merchant surcharge of the starting point of the scheduled delivery, the volume and weight of the scheduled delivery, the forecast weather information at the available time period, and / or whether the available time period is a public holiday. Based on these obtained features, the cost prediction model 205 predicts or forecasts the long-term delivery cost of the scheduled delivery in the available time period.

[0054] In the example, the longer the distance and / or duration of the scheduled delivery, the higher the predicted delivery fee. During peak lunch or dinner times, the surge rate may be high, and therefore, the predicted delivery fee may also be high. Certain merchant types may add a merchant surcharge or discount to all bookings, which is taken into account when predicting delivery fees.

[0055] Moreover, the cost prediction model 205 may include but is not limited to providing a series of multiple quantile regression neural networks with different quantile forecast results, which can be used as a convenient tool for adjusting forecast performance. Each of the quantile regression neural networks may include a multi-hidden layer feedforward neural network trained with historical data and quantile loss. Quantile regression can be used as a convenient tool to adjust the predicted delivery costs higher or lower as a whole in a proportional and reasonable manner when necessary. In this example, a quantile value of 0.5 can be used to perform standard prediction. It is also possible to use, for example, various regression methods (e.g., linear regression, lasso regression, support vector regression, quantile regression, etc.), various neural networks (e.g., multilayer perceptron, LSTM, etc.), and / or decision tree-based algorithms (e.g., XGBoost, LightGBM, etc.) to predict delivery costs.

[0056] The batch processing possibility prediction model 210 obtains the coordinates of the starting point of the scheduled delivery, the coordinates of the destination of the scheduled delivery, the available time period of the scheduled delivery, the number of completed orders in the area of ​​the scheduled delivery and during the same time period as the available time period in the previous week(s), and the batch processing rate (percentage) of the completed orders. Based on these obtained features, the batch processing possibility prediction model 210 predicts the batch processing rate (percentage of possibility) of the scheduled delivery to be batched with at least one other order in the available time period.

[0057] Furthermore, the batch processing possibility prediction model 210 can determine the orientation of the scheduled delivery, which indicates the direction between the start point and the destination of the scheduled delivery, and can use the determined orientation as one of the most important factors for predicting the batch processing rate of the scheduled delivery. For example, the batch processing possibility prediction model 210 can determine the number of completed orders in the area of ​​the scheduled delivery, with the determined orientation, and during the same time period as the available time period in the previous week, and determine the batch processing rate of the completed orders. The batch processing possibility prediction model 210 can use these determined features to predict the batch processing rate of the scheduled delivery in the available time period.

[0058] In addition, the batch processing possibility prediction model 210 may include a neural network with multiple long short-term memory (LSTM) layers to process time series features and thereby provide accurate batch processing possibility prediction results. The neural network with multiple LSTM layers can be trained with historical data.

[0059] Features used to predict the batch processing rate of scheduled deliveries can be composed of two parts: (1) time series (TS) features; and (2) non-TS features. TS features can include both the number of completed orders and the batch processing rate of completed orders. In general, historical data from the previous week within the same time period can be used as TS features. Non-TS features can include both geographical and calendar information, including the coordinates of the starting point of the scheduled delivery, the coordinates of the destination of the scheduled delivery, and the available time period. For a neural network, the TS features can be first analyzed by the LSTM layer and then merged with the non-TS features to train a classifier.

[0060] For non-peak times or time periods, demand may be lower (e.g., historical data indicates a lower number of completed orders), which means that the predicted batch processing rate may be determined to be lower. During peak times or time periods, due to high demand (e.g., historical data indicates a higher number of completed orders), the predicted batch processing rate may be determined to be higher.

[0061] The discount model 215 obtains the predicted batch processing rate of the scheduled delivery in the available time period, the coordinates of the starting point of the scheduled delivery or the location of the delivery vehicle, the coordinates of the destination of the scheduled delivery or the user, the available time period of the scheduled delivery, the distance of the scheduled delivery, and the duration of the scheduled delivery. The discount model 215 can also obtain the lead time of the scheduled delivery (the time difference between the current cost and / or discount time and the scheduled delivery time) and the remaining capacity of the available time period. Based on these obtained characteristics, the discount model 215 determines the discount (rate or percentage) for the scheduled delivery in the available time period.

[0062] The discount determined for same-day delivery can be moderate for off-peak times and higher for peak times. However, surge rates during peak times may also be higher, so the final delivery fee may still be higher than the fee determined during off-peak times. Advantageously, the number of same-day delivery orders can be increased while shifting demand to off-peak times.

[0063] The determined discount may increase monotonically with the predicted batch processing rate to provide discounts to scheduled delivery orders that have a higher probability of being batched within the available time period. The discount model 215 may use an S-shaped sigmoid function to implement a mapping function between the input predicted batch processing rate and the output discount, which has adjustable parameters for further optimization. For example, the following sigmoid function may be used to calculate the discount based on the predicted batch processing rate:

[0064]

[0065] where a, b, c, d, and e are coefficients, and BR is the predicted batch rate.

[0066] Moreover, in order to encourage users to select longer delivery time periods to further improve the goods delivery service, a larger discount can be provided for scheduled deliveries with longer delivery time windows. Therefore, in the example, each of the coefficients a, b, c, d, and e can be a function of a possible time window (which can be selected by the user), as shown below:

[0067]

[0068] b = 1;

[0069] c = 20-log (time window × 2);

[0070] d = 0.6; and

[0071] e=0.005×log(time window×2)(2)

[0072] In addition, when the predicted batch processing rate is low (e.g., below 0.3), a near minimum discount can be provided. For scheduled deliveries within the same time period, the discount can level off after the predicted batch processing rate reaches a predetermined high threshold, e.g., greater than or equal to 0.8.

[0073] Furthermore, the discount may range from 0 to 0.5. Alternatively, the range may be preset via gamma, and the determined discount may be linearly mapped to a final discount within the preset range. The range may include an absolute minimum and maximum value, or a minimum and maximum percentage value.

[0074] [ Figure 3 ] shows graphs illustrating mapping functions 305, 310 and 315 between predicted batching rates for inputs and output discounts according to an embodiment.

[0075] refer to[ Figure 3 ], mapping function 305 is for a time period having a duration of 2 hours, mapping function 310 is for a time period having a duration of 4 hours, and mapping function 315 is for a time period having a duration of 6 hours. Each of mapping functions 305, 310, and 315 is an S-shaped sigmoid function, where the discount increases monotonically as the batching rate of the prediction increases.

[0076] Furthermore, the mapping function 315 for a time period having a duration of 6 hours has a maximum discount, while the mapping function 305 for a time period having a duration of 2 hours has a minimum discount, to provide a larger discount for a larger time window.

[0077] Furthermore, in each of the mapping functions 305, 310, and 315, when the predicted batch processing rate is low (e.g., less than 0.3), the discount is at or near a predetermined minimum value. When the predicted batch processing rate is high (e.g., greater than or equal to 0.8), the discounts of the mapping functions 305, 310, and 315, respectively, are at or near a predetermined maximum value.

[0078] Again refer to [ Figure 2 ], the architecture 200 may determine the final delivery cost by, for example, the following equation:

[0079] Final cost = predicted cost × (1 – discount) (3)

[0080] The architecture 200 can also control devices (e.g., Figure 1] smartphone 105) displays the determined final delivery fee via, for example, a display (such as a smartphone screen). Based on the user selecting the final delivery fee displayed together with the available time period (e.g., scheduled delivery), the device can highlight the displayed final delivery fee and the available time period, and can control the server (e.g., [ Figure 1 ] server 115) to initiate a scheduled delivery (e.g., by controlling [ Figure 1 ]'s communication device 100 in the delivery vehicle 111).

[0081] In summary, the dynamic discount system can accurately predict the delivery costs of scheduled deliveries and the batch processing possibility of scheduled deliveries several days in advance. The dynamic discount system can also provide optimized discounts in a flexible manner based on the predicted batch processing possibility.

[0082] [ Figure 4 ] shows a flowchart of a method 400 for forecasting delivery costs for scheduled orders for goods delivery services and determining dynamic discounts based on batch processing possibility predictions according to an embodiment.

[0083] At operation 405 , method 400 includes predicting delivery costs for scheduled deliveries during available time periods.

[0084] At operation 410 , method 400 includes predicting a batching rate at which scheduled deliveries are to be batched with at least one other order in an available time period.

[0085] At operation 415 , method 400 includes determining a discount for scheduled deliveries during an available time period based on the predicted batching rate.

[0086] At operation 420 , method 400 includes determining a final delivery cost for a scheduled delivery during an available time period based on the predicted delivery cost and the determined discount.

[0087] At operation 425, method 400 includes displaying the determined final delivery cost for the user to select for the scheduled delivery.

[0088] [ Figure 4 ] method 400 is, for example, as follows [ Figure 5 ] is executed by server 115 shown in ].

[0089] [ Figure 5 ] shows the implementation of [ Figure 2 ]'s architecture 200 Figure 1 ]block diagram of server 115.

[0090] refer to[ Figure 5], the server 115 can be a server computer including a communication interface 505, a processor 510 and a memory 515.

[0091] The communication interface 505 can be used as a hardware and / or software interface that can, for example, transmit commands and / or data between a user and / or external device and other components of the server 115. The communication interface 505 can also establish communication between the server 115 and external devices (such as [ Figure 1 ] between the smart phone 105). The communication interface 505 can be connected to the network through a wireless or wired communication architecture to communicate with the external device. The communication interface 505 can be a wired or wireless transceiver or any other component for sending and receiving signals.

[0092] The processor 510 may include one or more of a CPU, a graphics processor unit (GPU), an accelerated processing unit (APU), a multi-integrated core (MIC), a field programmable gate array (FPGA), and / or a digital signal processor (DSP). The processor 510 may be a general controller that performs control of any one or any combination of the other components of the server 115, and / or performs operations related to communication or data processing. The processor 510 may execute one or more programs stored in the memory 515.

[0093] The memory 515 may include volatile and / or non-volatile memory. The memory 515 stores information, such as one or more of commands, data, programs (one or more instructions), applications, etc., which are related to at least one other component of the server 115 and are used to drive and control the server 115. For example, the commands and / or data may formulate an operating system (OS). The information stored in the memory 515 may be executed by the processor 510. The memory 515 may also store commands to be executed by the processor 510 to perform operations related to [ Figures 1 to 4 ] describes the functions and operations of the information.

[0094] The methods described herein may be performed, and the various processing or computing units and devices and computing entities described herein may be implemented by one or more circuits. In an embodiment, a "circuit" may be understood as any type of logic implementation entity, which may be hardware, software, firmware, or any combination thereof. Therefore, in an embodiment, a "circuit" may be a hardwired logic circuit or a programmable logic circuit, such as a programmable processor, such as a microprocessor. A "circuit" may also be software implemented or executed by a processor, such as any type of computer program, such as a computer program using virtual machine code. Any other type of implementation of the various functions described herein may also be understood as a "circuit" according to an alternative embodiment.

[0095] Although the present disclosure has been specifically shown and described with reference to specific embodiments, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention as defined by the appended claims. Therefore, the scope of the present invention is indicated by the appended claims, and it is therefore intended to cover all changes within the meaning and range of equivalents of the claims.

Claims

1. A method for forecasting delivery costs for a scheduled order of a goods delivery service and determining a dynamic discount based on a batch processing possibility prediction, the method comprising: Forecasting delivery costs for scheduled deliveries in available time slots; predicting a batching rate at which the scheduled delivery will be batched with at least one other order during the available time period; determining a discount for the scheduled delivery during the available time period based on the predicted batching rate; determining a final delivery cost for the scheduled delivery in the available time period based on the predicted delivery cost and the determined discount; as well as The determined final delivery cost is displayed for the user to select for the scheduled delivery.

2. The method according to claim 1, wherein: The estimated delivery costs include: obtaining the coordinates of the starting point of the scheduled delivery, the coordinates of the destination of the scheduled delivery, the available time period, the base fee or tiered base fees for the scheduled delivery, and a surge rate range; and Based on the obtained coordinates of the origin, the obtained coordinates of the destination, the obtained available time period, the obtained basic fee or tiered basic fee, and the obtained surge rate range, the delivery fee of the scheduled delivery in the available time period is predicted.

3. The method according to any one of claims 1 and 2, wherein: The delivery cost is predicted using at least one of the following: quantile regression, linear regression, lasso regression, support vector regression, multilayer perceptron neural network, long short-term memory neural network, decision tree-based algorithm.

4. The method according to claim 3, wherein: Each of the quantile regression neural networks includes a multi-hidden layer feed-forward neural network trained using historical data and quantile loss.

5. The method according to any one of claims 1 to 4, wherein: Predicting the batch processing rate includes: obtaining the coordinates of the starting point of the scheduled delivery, the coordinates of the destination of the scheduled delivery, the available time period, the number of completed orders in the area of ​​the scheduled delivery and during the same time period as the available time period in the previous week, and a batch processing rate of the completed orders; and Based on the obtained coordinates of the starting point, the obtained coordinates of the destination, the obtained available time period, the obtained number of completed orders, and the obtained batch processing rate of the completed orders, predict the batch processing rate at which the scheduled delivery will be batch processed with the at least one other order in the available time period.

6. The method according to claim 5, wherein: Predicting the batching rate also includes determining a bearing for the scheduled delivery, the bearing representing a direction between the origin and the destination of the scheduled delivery, and The batching rate at which the scheduled deliveries are to be batched with the at least one other order during the available time period is predicted further based on the determined location.

7. The method according to claim 6, wherein: Predicting the batch processing rate also includes determining the number of completed orders in the area of ​​the scheduled delivery, having the determined location and during the same time period as the available time period in the previous week, and determining the batch processing rate of the completed orders, and The batching rate at which the scheduled deliveries are to be batched with the at least one other order in the available time period is predicted further based on the determined number of the completed orders and the determined batching rate for the completed orders.

8. The method according to claim 5, wherein: Based on the available time period including off-peak hours, the predicted batch processing rate is low, and Based on the available time period including peak hours, the predicted batch processing rate is high.

9. The method according to any one of claims 1 to 8, wherein: The batch rate is predicted using a neural network with a long short-term memory (LSTM) layer trained using historical data.

10. The method according to any one of claims 1 to 9, wherein: Determining the discount includes: obtaining the predicted batching rate, the coordinates of the starting point of the scheduled delivery, the coordinates of the destination of the scheduled delivery, the available time period, the distance of the scheduled delivery, and the duration of the scheduled delivery; and The discount for the scheduled delivery in the available time period is determined based on the obtained predicted batch processing rate, the obtained coordinates of the starting point, the obtained coordinates of the destination, the obtained available time period, the obtained distance of the scheduled delivery, and the obtained duration of the scheduled delivery.

11. The method according to claim 10, wherein: The discount is determined using an S-shaped sigmoid function that has the predicted batch rate as input and the discount as output.

12. The method according to claim 11, wherein: The S-shaped sigmoid function includes: where a, b, c, d and e are coefficients and BR is the predicted batch rate.

13. The method according to claim 12, wherein: Each of the coefficients a, b, c, d and e is a function of the length of the available time period.

14. The method according to claim 10, wherein: Based on the short available time period, the determined discount is small, and Based on the long available time period, the determined discount is large.

15. The method according to claim 10, wherein: The determined discount is a preset minimum value based on the predicted batch processing rate being less than a predetermined minimum value, and The determined discount is a preset maximum value based on the predicted batch processing rate being greater than or equal to the predetermined maximum value.

16. The method according to any one of claims 1 to 15, wherein: The final delivery cost is determined using an equation that includes: Final cost = Predicted Fare × (1-Discount), Here, Predicted Fare is the predicted delivery fee, and Discount is the determined discount.

17. The method according to any one of claims 1 to 16, wherein: The determined final delivery cost is displayed on the display along with the available time period and is highlighted when selected by the user.

18. A server configured to execute the method according to any one of claims 1 to 17.

19. A computer program element comprising program instructions which, when executed by one or more processors, cause the one or more processors to perform the method according to any one of claims 1 to 17.

20. A computer readable medium comprising program instructions which, when executed by one or more processors, cause the one or more processors to perform the method according to any one of claims 1 to 17.