WO2020063875A1 - 配送范围的确定 - Google Patents

配送范围的确定 Download PDF

Info

Publication number
WO2020063875A1
WO2020063875A1 PCT/CN2019/108572 CN2019108572W WO2020063875A1 WO 2020063875 A1 WO2020063875 A1 WO 2020063875A1 CN 2019108572 W CN2019108572 W CN 2019108572W WO 2020063875 A1 WO2020063875 A1 WO 2020063875A1
Authority
WO
WIPO (PCT)
Prior art keywords
candidate
delivery
distribution
range
distribution range
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2019/108572
Other languages
English (en)
French (fr)
Inventor
张润丰
丁雪涛
邢轲
殷瑞
贾东
何仁清
郭振刚
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Sankuai Online Technology Co Ltd
Original Assignee
Beijing Sankuai Online Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Sankuai Online Technology Co Ltd filed Critical Beijing Sankuai Online Technology Co Ltd
Publication of WO2020063875A1 publication Critical patent/WO2020063875A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • G06Q10/083—Shipping
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management

Definitions

  • This application relates to the field of Internet technologies, and in particular, to the determination of the scope of distribution.
  • each delivery requester has its own delivery scope.
  • the delivery range of the delivery requester is usually manually drawn offline.
  • the delivery range of the delivery requester can be drawn one by one according to human experience. Take the instant delivery scenario of takeaway delivery as an example, the delivery requester can be a merchant; the delivery platform staff draws the delivery scope for each merchant one by one based on experience.
  • the present application provides a method, a device, a computer storage medium, and an electronic device for determining a distribution range.
  • a method for determining a delivery range includes:
  • an optimal distribution range is determined from the plurality of candidate distribution ranges.
  • a device for determining a delivery range includes:
  • a drawing unit configured to draw a plurality of candidate delivery ranges for a target delivery requester
  • a calculation unit configured to estimate the distribution efficiency data of each candidate distribution range according to the historical order information in the candidate distribution range
  • the determining unit is configured to determine an optimal distribution range from the plurality of candidate distribution ranges according to the distribution efficiency data of each candidate distribution range.
  • a computer-readable storage medium stores a computer program.
  • a processor is configured to perform the foregoing method.
  • An electronic device includes: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to execute the method for determining the above-mentioned distribution range.
  • the embodiment of the present application provides a method for determining a distribution range, and provides multiple candidate distribution ranges for a target distribution request, and uses historical distribution data of each candidate distribution range to calculate distribution efficiency data for each candidate distribution range. Finally, according to the distribution efficiency data of each candidate distribution range, an optimal candidate distribution range is selected from a plurality of candidate distribution ranges as the distribution range of the target distribution requester. In this way, the delivery range of the target delivery requester can be automatically generated.
  • the optimal distribution range scheme based on the historical distribution data is objective, so that the influence of subjective factors can be eliminated; on the other hand, the historical distribution data truly reflects the candidate distribution range Historical orders are active, so the distribution range determined based on historical distribution data is more accurate; on the other hand, it is more efficient to draw the distribution range automatically than to manually draw the distribution range.
  • FIG. 1 is a flowchart of a method for determining a delivery range, according to an exemplary embodiment of the present application
  • FIG. 2a is a schematic plan view of a certain city area according to an exemplary embodiment of the present application.
  • FIG. 2b is a schematic diagram of a candidate distribution range of a navigation radius according to an exemplary embodiment of the present application.
  • FIG. 3a is a schematic diagram of a hotspot block according to an exemplary embodiment of the present application.
  • 3b is a schematic diagram of a new candidate distribution range obtained after the candidate distribution range in FIG. 3a is optimized;
  • FIG. 4a is a schematic diagram of a candidate distribution range boundary crossing a block block according to an exemplary embodiment of the present application.
  • 4b is a schematic diagram of a road boundary closest to the boundary of the candidate distribution range in FIG. 4a;
  • 4c is a schematic diagram of a candidate distribution range redrawn according to the road boundary determined in FIG. 4b;
  • FIG. 5 is a hardware structural diagram of a device for determining a delivery range, according to an exemplary embodiment of the present application.
  • Fig. 6 is a schematic block diagram of a device for determining a delivery range, according to an exemplary embodiment of the present application.
  • first, second, third, etc. may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other.
  • first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information.
  • word “if” as used herein can be interpreted as “at” or "when” or "in response to determination”.
  • each delivery requester has its own delivery scope.
  • determining the delivery scope of a delivery requester is usually done manually offline.
  • the distribution scope of the distribution requester may be drawn one by one based on human experience.
  • timely delivery such as takeaway delivery
  • the delivery requester can be a merchant; the delivery platform staff draws the delivery scope for each merchant one by one based on experience.
  • the manual drawing of the distribution range is mainly based on the work experience of the staff. Not only is the subjective factor large, prone to deviation, but also time-consuming and labor-intensive, and the efficiency is low; a more accurate, efficient and objective distribution needs to be provided Scoping plan.
  • the distribution range is an area in the geographic concept. On the takeaway platform, the merchant is only visible to users who are located within the distribution range of the merchant. In other words, the order relationship only occurs between the merchant and the users within the scope of distribution. And the delivery range is directly related to the merchant's revenue. Therefore, it can be seen that the merchant's delivery range is a hard constraint and directly affects the order volume, delivery efficiency and user experience. If the merchant's distribution range is set too small, the potential user group will be small, and the order flow and revenue will be small; if the merchant's distribution range is set too large, although the potential user group is large, the order flow generated may be improved to a certain extent , But the average distribution efficiency may be greatly affected, which in turn affects the user experience.
  • a plurality of candidate distribution ranges are first provided to the target distribution requester, and then the historical distribution data of each candidate distribution range is used to calculate the distribution efficiency data of each candidate distribution range.
  • the distribution efficiency data of the candidate distribution range selects an optimal candidate distribution range from the multiple candidate distribution ranges as the distribution range of the target distribution requester. In this way, the delivery range of the target delivery requester can be automatically generated. Because the historical distribution data exists objectively, the distribution efficiency, revenue and other parameters included in the historical distribution data can objectively reflect the true distribution situation of the candidate distribution range, thus effectively guaranteeing objectivity and accuracy, and automatically drawing the distribution range relative to manual It is more efficient to map the distribution scope.
  • an optimal candidate distribution range is selected from a plurality of candidate distribution ranges as the distribution range of the target distribution requester.
  • the distribution efficiency data can be used to maximize the prediction of the candidate distribution range. Revenue situation, and then select the candidate distribution range with the most revenue as the final distribution range. In this way, both the distribution efficiency of the distribution scope is taken into consideration, and the revenue of the distribution scope can be maximized.
  • Fig. 1 is a flow chart of a method for determining a delivery range, according to an exemplary embodiment.
  • the method may be applied to a server (hereinafter referred to as a server) for determining a delivery range.
  • the server may refer to Servers, server clusters, or cloud platforms built from server clusters.
  • the method may specifically include the following steps:
  • Step 110 Draw multiple candidate delivery ranges for the target delivery requester.
  • the target delivery requester may be an object whose delivery scope is to be determined.
  • a merchant whose delivery scope is to be determined in a takeaway business scenario; or a courier outlet whose delivery scope is to be determined in a courier business scenario.
  • the step 110 may specifically include: drawing a plurality of candidate delivery ranges with a plurality of different radii taking the position of the target delivery requester as the center.
  • the plurality of different radii may refer to two or more radii.
  • the radius may be preset. For example, 1 km, 3 km, 5 km, etc. Generally, the radius can be flexibly changed based on actual business needs.
  • FIG. 2a A schematic plan map of a certain city area as shown in FIG. 2a.
  • point P represents the location of the target delivery requester. With point P as the center, three different radius distribution ranges are plotted, and the corresponding radii are r1 km, r2 km, and r3 km, respectively.
  • the radius may refer to a straight line radius.
  • the delivery range of the straight-line radius may refer to a range where the straight-line distance from the point P (the location of the target delivery requester) is R (a preset radius value). Referring to the distribution range shown in FIG. 2a, it can be seen that all are regular circles, that is, the straight line distance from any point on the boundary of the same distribution range to the point P is the same.
  • the radius may refer to a navigation radius.
  • the delivery range of the navigation radius may refer to a range in which the navigation distance is R (a preset radius value) from the point P (the location of the target delivery requester).
  • the navigation distance refers to the distance of the real path from point to point. Because the logistics distribution process requires the distribution personnel to distribute along the route, and the distribution route is not always straight, it may be necessary to detour from one location to another, so the actual delivery distance is often greater than the straight distance. Using the navigation radius can truly measure the actual distance between two points and the distribution cost, so that it is more reasonable and accurate in the real geographical location scenario.
  • FIG. 2b is a schematic diagram of a candidate distribution range of a navigation radius. It can be seen that the candidate distribution range is not a regular circle as shown in Figure 2a. Although the straight line distance between the point P and the point on the boundary of the same distribution range in Figure 2b may be different, the navigation distance is the same, and the navigation distances are both the navigation radius .
  • the method may further include: performing optimization processing on the candidate distribution range.
  • optimizing the candidate distribution range may specifically include: obtaining hotspot blocks within a preset radius outside the candidate distribution range; and using the hotspot blocks as part of the candidate distribution range To form a new candidate distribution range.
  • the hot block may refer to a geographic region where orders perform well. In one embodiment, whether to belong to a hot block can be determined according to the number of historical orders. Specifically, a clustering algorithm can be used to cluster out blocks whose historical orders exceed a certain threshold, and such blocks can be referred to as hot blocks.
  • the preset radius outside the candidate distribution range is not the same radius value as the preset radius.
  • the aforementioned preset radius of the planned candidate distribution range may be referred to as a first radius
  • the preset radius outside the candidate distribution range may be referred to as a second radius.
  • the second radius is larger than the first radius.
  • the second radius may be obtained by adding a preset value to the first radius.
  • Figure 3a is a schematic diagram of a hot block.
  • a small range there are two ranges of a first radius range (referred to as a small range) and a second radius range (referred to as a large range).
  • a large range In a large area outside a small area, there are hotspot blocks A and hotspot regions B, so hotspot blocks A and hotspot blocks B can be used as part of the candidate distribution range to form new candidates as shown in Figure 3b Delivery Area.
  • the candidate distribution range can cover as many user-exposed places as possible, so that the candidate distribution range of the distribution requester can be generated more reasonably.
  • the boundary of the candidate distribution range drawn in the above steps may be inaccurate, for example, it is not a road boundary, as shown in the schematic diagram of the candidate distribution range boundary 401 crossing the block (road 402) in FIG.
  • GMV Geographical Merchandise Volume
  • optimizing the candidate distribution range may specifically include: determining a road boundary closest to a boundary of the candidate distribution range; and drawing an optimized candidate distribution range according to the road boundary .
  • FIG. 4a it can be seen that the boundary 401 of the candidate distribution range crosses the road 402; at this time, the road boundary closest to the boundary 401 needs to be determined.
  • the road boundary 403 closest to the boundary 401 is shown as a dashed line segment; therefore, the optimized candidate distribution range is drawn based on the determined road boundary 403 as shown in FIG. 4c.
  • performing optimization processing on the candidate distribution range may specifically include: performing data compression on a set of distribution reference points constituting the candidate distribution range.
  • the data compression on the set of distribution reference points constituting the candidate distribution range may specifically include: traversing the set of distribution reference points constituting the candidate distribution range; and if the distance between two adjacent distribution reference points is less than a threshold value , The two adjacent distribution reference points are combined into one.
  • the candidate distribution range may be composed of a series of distribution reference points on a boundary of the candidate distribution range.
  • a border line can be planned by connecting the distribution reference points.
  • the navigation paths between two adjacent delivery reference points are connected to form a closed polygon, and the area within the polygon is the candidate delivery range.
  • the distribution reference points are stored in the form of a collection.
  • the distribution reference set can be compressed.
  • the method of merging the two adjacent delivery reference points into one may include: deleting any one of the two adjacent delivery reference points from the set of delivery reference points; or calculating the adjacent For an average delivery reference point of two delivery reference points, replace two adjacent delivery reference points in the set of delivery reference points with the average delivery reference point.
  • the amount of stored data can be effectively reduced without affecting the size of the candidate distribution range.
  • Step 120 Estimate the distribution efficiency data of each candidate distribution range according to the historical order information in the candidate distribution range.
  • the distribution efficiency data of the candidate distribution range can be estimated based on historical order information within the candidate distribution range.
  • the distribution efficiency data may be used to indicate a revenue situation that a distribution requester can generate after adopting a candidate distribution range.
  • the delivery efficiency data may include an order quantity and / or an average delivery duration.
  • the step 120 may specifically include: estimating an order quantity and / or an average delivery time in the candidate delivery range according to historical order information in the candidate delivery range.
  • the historical order information may include statistics of merchant information within the candidate distribution range, distribution difficulty information, and the like.
  • the merchant information may include characteristics of a merchant dimension, such as the number of merchants within a candidate distribution range, historical order amounts, merchant ratings, merchant categories, and the like.
  • the distribution difficulty information may include the no-load ratio of the distribution staff within the candidate distribution range, the number of distribution staff, and the like.
  • the server may pre-train a machine learning model for estimating the order quantity and / or average delivery time of the candidate delivery range based on the machine learning technology. Estimating the order quantity and / or average delivery time in each candidate delivery range specifically includes: estimating the order quantity and / or average delivery time in each candidate delivery range based on a machine learning algorithm.
  • the server can collect a large number of historical order information samples in different distribution ranges in advance, and train an order quantity estimation model based on machine learning algorithms; through continuous learning, the order quantity estimation model can be continuously improved, and when the order When the volume estimation model reaches the estimation (for example, the accuracy of the estimation meets the business requirements), you can go online and use the order volume estimation model.
  • the machine learning model may use, for example, Logistic Regression (LR), GBDT (Gradient Boosting Decision Tree), and the like.
  • the historical order information may include statistics of delivery staff in the candidate delivery range.
  • the deliveryman information may include a delivery duration of a deliveryman's delivery history order.
  • the server can collect a large number of historical order information samples in different delivery ranges in advance, and train an average delivery duration estimation model based on a machine learning algorithm; the continuous delivery can continuously improve the average delivery duration estimation model.
  • the average delivery time estimation model reaches the prediction (for example, the estimation accuracy meets the business requirements)
  • the online delivery time estimation model can be used online.
  • the average delivery time within the candidate delivery range can be calculated.
  • the machine learning model may adopt, for example, logistic regression (LR), GBDT (gradient enhanced decision tree), and the like.
  • Step 130 According to the distribution efficiency data of each candidate distribution range, determine an optimal distribution range from the plurality of candidate distribution ranges.
  • an optimal candidate distribution range can be selected from the distribution efficiency data as the distribution range of the target distribution requester.
  • such an optimal item selected from a plurality of optional items can be implemented by using an optimization algorithm. Specifically, after optimizing the optimization target according to the business requirements, and using the optimization target as the jump-out condition, the optimization algorithm is called for calculation. When the calculation result meets the jump-out condition and the optimization target is reached, the optional items at the time of the jump can be used as the optimal items .
  • the step 130 may specifically include: determining an optimal distribution range from the plurality of candidate distribution ranges based on a combination optimization algorithm.
  • the goal of the combined optimization algorithm is to maximize the revenue of the target delivery requester in the candidate delivery range; the constraint is that the order-weighted area of the candidate delivery range meets a preset condition.
  • the revenue may include a transaction amount of a single average delivery time.
  • the transaction amount can be expressed by GMV (Gross Merchandise Volume). That is, the transaction amount of the single average delivery time is expressed as the GMV of the single average delivery time.
  • the GMV that maximizes the average delivery time per unit of the target delivery requester in the candidate delivery range can be calculated by the following formula:
  • c mn indicates whether the merchant m uses the 0-1 identification of the candidate delivery range n (also a variable to be solved by the combination optimization algorithm); order_num mn indicates the estimated order of the merchant m predicted by the machine learning model when the candidate delivery range n is adopted.
  • Price m represents the average customer unit price of the merchant m (that is, the sum of historical order amounts divided by the number of historical orders); time mn represents the estimated average delivery time of the merchant m predicted by the machine learning model when the candidate delivery range n is used; ⁇ represents a smoothing factor (to avoid the denominator being 0); A mn represents the distribution area of the candidate distribution range n of the merchant m; Represents a manually set scaling factor (used to regulate the expansion / reduction of order-weighted average distribution area).
  • the delivery efficiency data of each candidate delivery range is calculated, and finally based on the delivery efficiency data of each candidate delivery range, a variety of Among the candidate distribution ranges, an optimal candidate distribution range is selected as the distribution range of the target distribution requester. In this way, the delivery range of the target delivery requester can be automatically generated.
  • an embodiment of a device for determining a delivery range is also provided.
  • the means for determining the delivery range may be applied on a server.
  • the device embodiments may be implemented by software, or by hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by reading the corresponding computer program instructions in the non-volatile memory into the memory through its processor.
  • FIG. 5 it is a hardware structure diagram of a device for determining a delivery range in an embodiment, except for the processor, memory, network interface, and non-volatile memory shown in FIG. 5.
  • the actual functions usually determined according to the distribution scope may also include other hardware, which will not be described again.
  • the apparatus for determining a delivery range may include:
  • a drawing unit 610 configured to draw a plurality of candidate delivery ranges for a target delivery requester
  • the calculation unit 620 is configured to estimate the distribution efficiency data of each candidate distribution range according to the historical order information in the candidate distribution range;
  • the determining unit 630 is configured to determine an optimal distribution range from the plurality of candidate distribution ranges according to the distribution efficiency data of each candidate distribution range.
  • the drawing unit 610 is specifically configured to draw a plurality of candidate delivery ranges with the target delivery requestor as the center and a plurality of different navigation radii.
  • the delivery efficiency data includes an order quantity and / or an average delivery time.
  • the apparatus further includes: an optimization unit configured to optimize the candidate distribution range; and the calculation unit 620 is specifically configured to: based on historical order information in the candidate distribution range, predict Estimate the order quantity and / or average delivery time in each candidate delivery range after optimization.
  • the optimization unit specifically includes:
  • a road boundary determination subunit configured to determine a road boundary closest to a boundary of the candidate distribution range
  • the redrawing sub-unit is configured to draw an optimized candidate distribution range according to the road boundary.
  • the optimization unit specifically includes:
  • the data compression subunit is configured to perform data compression on a set of distribution reference points constituting the candidate distribution range.
  • the data compression subunit specifically includes:
  • the merging subunit is configured to traverse the set of distribution reference points constituting the candidate distribution range; if the distance between two adjacent distribution reference points is less than a threshold value, the two adjacent distribution reference points are merged into one.
  • merging the two adjacent delivery reference points into one specifically including: deleting any one of the two adjacent delivery reference points from the set of delivery reference points; or, calculating the phase An average delivery reference point adjacent to two delivery reference points, and replacing two adjacent delivery reference points in the set of delivery reference points with the average delivery reference point.
  • the optimization unit specifically includes:
  • a hotspot block acquisition subunit configured to obtain hotspot blocks within a preset radius outside the candidate distribution range
  • the redrawing subunit is configured to use the hot block as a part of the candidate distribution range to form an optimized candidate distribution range.
  • estimating the order quantity and / or average delivery time in each candidate delivery range specifically includes: estimating the order quantity and / or average delivery time in each candidate delivery range based on a machine learning algorithm.
  • the determining unit 630 specifically includes:
  • the determining subunit is configured to determine an optimal delivery range from the plurality of candidate delivery ranges based on a combination optimization algorithm based on the delivery efficiency data of each candidate delivery range.
  • the goal of the combined optimization algorithm is to maximize the revenue of the target delivery requester in the candidate delivery range.
  • the revenue includes the transaction amount of the average delivery time of each order.
  • the determining subunit is specifically configured to: maximize the transaction amount of the average delivery time of each order in the candidate delivery range according to the order quantity and / or average delivery time of each candidate delivery range; The candidate delivery range with the largest transaction amount is determined as the optimal delivery range.
  • the relevant part may refer to the description of the method embodiment.
  • the device embodiments described above are only schematic, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, may be located One place, or it can be distributed across multiple network elements. Some or all of these modules can be selected to achieve the purpose according to actual needs. Those of ordinary skill in the art can understand and implement without creative efforts.
  • FIG. 6 describes the schematic diagram of the internal functional modules and structure of the device for determining the delivery range.
  • the substantial execution subject may be an electronic device, including: a processor; a memory for storing processor-executable instructions; The processor is configured to:
  • an optimal distribution range is determined from the plurality of candidate distribution ranges.
  • drawing multiple candidate delivery ranges for the target delivery requester specifically includes drawing multiple candidate delivery ranges with the target delivery requestor as the center and multiple different navigation radii.
  • the delivery efficiency data includes an order quantity and / or an average delivery time.
  • the method further includes: optimizing the candidate distribution range; estimating the distribution efficiency data of each candidate distribution range, specifically including: estimating the order quantity and / or average of each candidate distribution range after optimization; Delivery time.
  • performing optimization processing on the candidate distribution range specifically includes: determining a road boundary closest to a boundary of the candidate distribution range; and drawing an optimized candidate distribution range according to the road boundary.
  • performing optimization processing on the candidate distribution range specifically includes: compressing data on a set of distribution reference points constituting the candidate distribution range.
  • performing data compression on the set of distribution reference points constituting the candidate distribution range specifically includes: traversing the set of distribution reference points constituting the candidate distribution range; and if the distance between two adjacent distribution reference points is less than a threshold value , The two adjacent distribution reference points are combined into one.
  • merging the two adjacent delivery reference points into one specifically including: deleting any one of the two adjacent delivery reference points from the set of delivery reference points; or, calculating the phase An average delivery reference point adjacent to two delivery reference points, and replacing two adjacent delivery reference points in the set of delivery reference points with the average delivery reference point.
  • performing optimization processing on the candidate distribution range specifically includes: obtaining hotspot blocks within a preset radius outside the candidate distribution range; and using the hotspot blocks as part of the candidate distribution range to form an optimized distribution range.
  • Candidate delivery range specifically includes: obtaining hotspot blocks within a preset radius outside the candidate distribution range; and using the hotspot blocks as part of the candidate distribution range to form an optimized distribution range.
  • estimating the order quantity and / or average delivery time in each candidate delivery range specifically includes: estimating the order quantity and / or average delivery time in each candidate delivery range based on a machine learning algorithm.
  • determining an optimal distribution range from the multiple candidate distribution ranges specifically includes: determining an optimal distribution range from the multiple candidate distribution ranges based on a combination optimization algorithm.
  • the goal of the combined optimization algorithm is to maximize the revenue of the target delivery requester in the candidate delivery range.
  • the revenue includes the transaction amount of the average delivery time of each order.
  • determining the optimal distribution range from the plurality of candidate distribution ranges specifically including: according to the order quantity and / Or average the delivery time to maximize the transaction amount of the average delivery time in the candidate delivery range; determine the candidate delivery range with the largest transaction amount in the average delivery time as the optimal delivery range.
  • the processor may be a central processing unit (English: Central Processing Unit, CPU for short), or other general-purpose processors, digital signal processors (English: Digital Signal Processor). , Referred to as DSP), application specific integrated circuit (English: Application Specific Integrated Circuit, referred to as ASIC), etc.
  • a general-purpose processor may be a microprocessor, or the processor may be any conventional processor, and the foregoing memory may be a read-only memory (English: read-only memory (abbreviation: ROM)), a random access memory (English : Random access memory (abbreviation: RAM), flash memory, hard disk or solid state hard disk.
  • the steps of the method disclosed in combination with the embodiments of the present invention may be directly implemented by a hardware processor, or may be performed by a combination of hardware and software modules in the processor.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Quality & Reliability (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Development Economics (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本申请提供配送范围的确定方案，其包括：针对目标配送请求方绘制多个候选配送范围；根据所述候选配送范围内的历史订单信息，预估每个候选配送范围的配送效率数据；根据每个候选配送范围的配送效率数据，从所述多个候选配送范围中确定最优的配送范围。

Description

配送范围的确定 技术领域
本申请涉及互联网技术领域，尤其涉及配送范围的确定。
背景技术
在即时配送场景中，每个配送请求方都有独自的配送范围。
在相关技术中，配送请求方的配送范围通常是线下人工进行绘制，在一实施例中，可以根据人工经验逐个绘制配送请求方的配送范围。以外卖配送这一即时配送场景为例，配送请求方可以为商家；配送平台工作人员基于经验逐个为每个商家绘制配送范围。
发明内容
有鉴于此，本申请提供一种配送范围的确定方法、装置及计算机存储介质和电子设备。
具体地，本申请是通过如下技术方案实现的：
一种配送范围的确定方法，所述方法包括：
针对目标配送请求方绘制多个候选配送范围；
根据所述候选配送范围内的历史订单信息，预估每个候选配送范围的配送效率数据；
根据每个候选配送范围的配送效率数据，从所述多个候选配送范围中确定最优的配送范围。
一种配送范围的确定装置，所述装置包括：
绘制单元，被配置为针对目标配送请求方绘制多个候选配送范围；
计算单元，被配置为根据所述候选配送范围内的历史订单信息，预估每个候选配送范围的配送效率数据；
确定单元，被配置为根据每个候选配送范围的配送效率数据，从所述多个候选配送范围中确定最优的配送范围。
一种计算机可读存储介质，存储有计算机程序，当执行所述计算机程序时，处理器用于执行上述方法。
一种电子设备，包括：处理器；和用于存储处理器可执行指令的存储器；其中，所述处理器被配置为执行上述配送范围的确定方法。
本申请实施例，提供了一种配送范围的确定方案，针对目标配送请求提供多种候选配送范围，并利用每个候选配送范围的历史配送数据，计算出每个候选配送范围的配送效率数据，最后根据每个候选配送范围的配送效率数据从多个候选配送范围中选取一个最优的候选配送范围作为目标配送请求方的配送范围。如此，可以自动生成目标配送请求方的配送范围。一方面，由于历史配送数据是客观存在的，因此基于历史配送数据确定的最优配送范围方案是客观的，这样，可消除主观因素的影响；另一方面由于历史配送数据真实反映了候选配送范围历史订单活跃情况，因此基于历史配送数据确定的配送范围更为准确；再一方面，自动绘制配送范围相对于人工绘制配送范围更有效率。
附图说明
图1是本申请一示例性实施例示出的一种配送范围的确定方法的流程图；
图2a是本申请一示例性实施例示出的某个城市区域的平面地图示意图；
图2b是本申请一示例性实施例示出的导航半径的候选配送范围的示意图；
图3a是本申请一示例性实施例示出的热点区块的示意图；
图3b对图3a中的候选配送范围进行优化处理后得到的新的候选配送范围的示意图；
图4a是本申请一示例性实施例示出的候选配送范围边界穿越街区的示意图；
图4b是图4a中距离候选配送范围边界最近的道路边界的示意图；
图4c是根据图4b中所确定的道路边界重新绘制的候选配送范围的示意图；
图5是本申请一示例性实施例示出的一种配送范围的确定装置的硬件结构图；
图6是本申请一示例性实施例示出的一种配送范围的确定装置的模块示意图。
具体实施方式
这里将详细地对示例性实施例进行说明，其示例表示在附图中。下面的描述涉及附图时，除非另有表示，不同附图中的相同数字表示相同或相似的要素。以下示例性实施例中所描述的实施方式并不代表与本申请相一致的所有实施方式。相反，它们仅是与如所附权利要求书中所详述的、本申请的一些方面相一致的装置和方法的例子。
在本申请使用的术语是仅仅出于描述特定实施例的目的，而非旨在限制本申请。在本申请和所附权利要求书中所使用的单数形式的“一种”、“所述”和“该”也旨在包括多数形式，除非上下文清楚地表示其他含义。还应当理解，本文中使用的术语“和/或”是指并包含一个或多个相关联的列出项目的任何或所有可能组合。
应当理解，尽管在本申请可能采用术语第一、第二、第三等来描述各种信息，但这些信息不应限于这些术语。这些术语仅用来将同一类型的信息彼此区分开。例如，在不脱离本申请范围的情况下，第一信息也可以被称为第二信息，类似地，第二信息也可以被称为第一信息。取决于语境，如在此所使用的词语“如果”可以被解释成为“在……时”或“当……时”或“响应于确定”。
如前所述，在即时配送场景中，每个配送请求方都有独自的配送范围。
在相关技术中，确定配送请求方的配送范围通常是线下人工进行绘制。在一些实施例中，可根据人工经验逐个绘制配送请求方的配送范围。以外卖配送这样的及时配送为例，配送请求方可以为商家；配送平台工作人员基于经验逐个为每个商家绘制配送范围。然而，人工绘制配送范围主要基于工作人员的工作经验，不仅主观因素较大，容易出现偏差，而且费时费力，效率较低；需要提供一种更为准确性，更有效率且更为客观的配送范围的确定方案。
所述配送范围是地理概念上的一块区域，在外卖平台上商家只对位于该商家配送范围内的用户可见。也就是说，订单关系只产生在商家和配送范围内的用户之间。而配送范围直接与商家的收益相关，由此可见，商家配送范围是一种硬性约束，且直接影响订单量、配送效率和用户体验。若商家配送范围设定得过小，则潜在用户群体小，订单流、收益将较小；若商家配送范围设定得过大，虽然潜在用户群体较大，产生的订单流可能有一定程度提升，但平均配送效率可能会受到较大影响，进而影响用户的体验。
因此，如何平衡配送范围、配送效率和收益之间的关系成为一个亟待解决的问题。
在一个或多个实施例中，首先提供给目标配送请求方多个候选配送范围，然后利用每个候选配送范围的历史配送数据，计算出每个候选配送范围的配送效率数据，最后根据每个候选配送范围的配送效率数据从多种候选配送范围中选取一个最优的候选配送范围作为目标配送请求方的配送范围。如此，可以自动生成目标配送请求方的配送范围。由于历史配送数据是客观存在的，因此历史配送数据包括的配送效率、收益等参数均可以客观反映候选配送范围的真实配送情况，因此有效保证了客观性和准确性，自动绘制配送范围相对于人工绘制配送范围更有效率。
其中，根据每个候选配送范围的配送效率数据从多种候选配送范围中选取一个最优的候选配送范围作为目标配送请求方的配送范围，具体可以根据配送效率数据来最大化预测候选配送范围的收益情况，然后选取收益最大的候选配送范围作为最终的配送范围。如此，既兼顾了配送范围的配送效率，也可以最大化配送范围的收益情况。
图1是一示例性实施例示出的一种配送范围的确定方法流程图，所述方法可以应用在用于确定配送范围的服务端(以下简称为服务端)中，所述服务端可以是指服务器、服务器集群或者由服务器集群构建的云平台。该方法具体可以包括如下步骤：
步骤110：针对目标配送请求方绘制多个候选配送范围。
所述目标配送请求方可以是指待确定配送范围的对象。例如在外卖业务场景下待确定配送范围的商家；或者，在快递业务场景下待确定配送范围的快递网点等等。
在一实施例中，所述步骤110，具体可以包括：以目标配送请求方的位置为中心、以多个不同的半径绘制多个候选配送范围。
在一实施例中中，所述多个不同的半径可以是指2个或者2个以上半径。所述半径可以是预先设置的。例如1公里，3公里，5公里等。一般的，所述半径可以基于实际业务需要灵活变更。
如图2a所示的某个城市区域的平面地图示意图。图2a中，P点表示目标配送请求方的位置，以P点为中心，绘制有3个不同半径的配送范围，对应的半径分别为r1公里、r2公里和r3公里。
在一实现方式中，所述半径可以是指直线半径。所述直线半径的配送范围可以是指距离P点(目标配送请求方的位置)直线距离为R(预设的半径值)的范围。参考图2a所示的配送范围，可见均是规则的圆，即同一个配送范围的边界上的任意一点距离P点的直线距离均是相同的。
在另一实现方式中，所述半径可以是指导航半径。所述导航半径的配送范围可以是指距离P点(目标配送请求方的位置)导航距离为R(预设的半径值)的范围。也就是说导航距离指的是点到点之间真实路径的距离。由于物流配送过程需要配送人员沿路径配送，而配送路径并非总是直线，从一个地点到另一个地点可能需要绕行，因此实际配送距离往往会大于直线距离。采用导航半径能够真实衡量两点之间的实际距离以及配送成本，从而在真实的地理位置场景中更加合理准确。
图2b所示是导航半径的候选配送范围的示意图。可见该候选配送范围并非图2a中那样规则的圆，虽然图2b中同一个配送范围边界上的点距离P点的直线距离可能是不同的，但是导航距离是相同的，导航距离均为导航半径。
在一实施例中，在所述步骤110之后，还可以包括：对所述候选配送范围进行优化处理。
在一种实现方式中：对所述候选配送范围进行优化处理，具体可以包括：获取所述候选配送范围外预设半径内的热点区块；和将所述热点区块作为候选配送范围的一部分，形成新的候选配送范围。
在一实施例中，所述热点区块可以是指订单表现较好的地理区域。在一实施例中，判定是否属于热点区块，可以根据历史订单数量进行判断。具体地，可以采用聚类算法，聚类出历史订单数量超过一定阈值的区块，而这样的区块即可以称为热点区块。
其中，所述候选配送范围外预设半径与前述预设半径并非同一个半径值。为了便于区分，可以将前述规划候选配送范围的预设半径称为第一半径，将该候选配送范围外预设半径称之为第二半径。通常，所述第二半径大于所述第一半径。
在一实施例中，第二半径可以是在第一半径基础上增加预设数值后得到的。
图3a所示是热点区块的示意图。图3a中，存在第一半径范围(简称为小范围)和第二半径范围(简称为大范围)这两种范围。在小范围外大范围内的区域内，存在热点区块A和热点区域B，因此可以将热点区块A和热点区块B作为候选配送范围的一部分，从而形成图3b所示的新的候选配送范围。
通过本实施例，添加用户热度高的地理位置区块，也就是让候选配送范围能够尽可能的覆盖一些用户曝光密集的地方，从而可以使得配送请求方的候选配送范围生成地更加合理。
在实际应用中，上述步骤绘制的候选配送范围边界可能不准确，例如不是以道路 为边界，如图4a所示的候选配送范围边界401穿越街区(道路402)的示意图。以即时配送业务中的外卖场景为例，商家的配送范围若穿越街区而未完全覆盖该街区，则可能会导致该街区的部分用户可以在该商家处进行点餐，而另一部分用户却无法在该商家处进行点餐，这显然是不合理的；况且，通常同一个街区点餐人数越多越好，因为对于配送员来说可以同时为该街区的用户配送，从而能够实现较高的配送效率，以获得较多收益。对于平台来说穿越街区会显著降低GMV(Gross Merchandise Volume，平台成交金额)。因此需要对候选配送范围进行优化处理，使得候选配送范围的边界是以道路为边界划分的。
为此，在一种实现方式中：对所述候选配送范围进行优化处理，具体可以包括：确定距离所述候选配送范围的边界最近的道路边界；根据所述道路边界绘制优化后的候选配送范围。
依然以图4a为例，可见候选配送范围的边界401跨越了道路402；此时，需要确定距离边界401最近的道路边界。如图4b所示，距离边界401最近的道路边界403如虚线段所示；因此，根据所确定的道路边界403绘制优化后的候选配送范围如图4c所示。
在一种实现方式中：对所述候选配送范围进行优化处理，具体可以包括：对构成所述候选配送范围的配送基准点集合进行数据压缩。
其中，对构成所述候选配送范围的配送基准点集合进行数据压缩，具体可以包括：遍历构成所述候选配送范围的配送基准点集合；且若相邻两个配送基准点之间的距离小于阈值，则将所述相邻两个配送基准点合并为一个。
在一些实施例中，所述候选配送范围可以是由所述候选配送范围边界上的一系列配送基准点构成的。将所述配送基准点相连接即可规划出一条边界线。具体地，将相邻的两个配送基准点之间的导航路径连接，形成一个封闭多边形，该多边形内的区域即为候选配送范围。
通常，所述配送基准点是以集合的形式存储的。为了提高存储空间利用率,可以对配送基准点集合进行压缩处理。
其中，将所述相邻两个配送基准点合并为一个的方式可以包括：将所述相邻两个配送基准点中任意一个从所述配送基准点集合中删除；或者，计算所述相邻两个配送基准点的平均配送基准点，将所述配送基准点集合中的相邻两个配送基准点替换为所 述平均配送基准点。
通过对构成候选配送范围的配送基准点集合进行压缩处理，在不影响候选配送范围大小的情况下，可以有效减少存储数据量。
步骤120：根据所述候选配送范围内的历史订单信息，预估每个候选配送范围的配送效率数据。在一实施例中，对于每个候选配送范围，可根据该候选配送范围内的历史订单信息，预估该候选配送范围的配送效率数据。
所述配送效率数据可以用于表示配送请求方采用候选配送范围后可以产生的收益情况。
在一实施例中，所述配送效率数据可以包括订单量和/或平均配送时长。
也就是说，所述步骤120，具体可以包括：根据所述候选配送范围内的历史订单信息，预估所述候选配送范围内的订单量和/或平均配送时长。
其中，针对预估候选配送范围内的订单量：所述历史订单信息可以包括统计出的候选配送范围内的商家信息、配送难度信息等。具体地，所述商家信息可以包括商家维度的特征，例如候选配送范围内的商家数量、历史订单量、商家评分、商家品类等。所述配送难度信息可以包括候选配送范围内配送员空载率，配送员数量等。
在一实施例中，服务端可以基于机器学习技术，预先训练用于预估候选配送范围的订单量和/或平均配送时长的机器学习模型。预估每个候选配送范围内的订单量和/或平均配送时长，具体包括：基于机器学习算法，预估每个候选配送范围内的订单量和/或平均配送时长。
具体地，服务端可以预先收集大量不同配送范围内的历史订单信息样本，并基于机器学习算法，训练订单量预估模型；通过持续地学习可以不断完善所述订单量预估模型，当该订单量预估模型达到预计(例如预估准确性符合业务要求)时，即可以上线并使用订单量预估模型。在业务使用时，当统计出候选配送范围的历史订单信息后，将统计出的历史订单信息输入订单量预估模型，就可以计算得到候选配送范围内的订单量。所述机器学习模型可以采用例如逻辑回归(Logistics Regression，LR)，GBDT(Gradient Boosting Decision Tree，梯度增强决策树)等。
其中，针对预估候选配送范围内的平均配送时长：所述历史订单信息可以包括统计出候选配送范围内的配送员信息。具体地，所述配送员信息可以包括配送员配送历史订单的配送时长。
具体地，服务端可以预先收集大量不同配送范围内的历史订单信息样本，并基于机器学习算法，训练平均配送时长预估模型；通过持续地学习可以不断完善所述平均配送时长预估模型，当该平均配送时长预估模型达到预计(例如预估准确性符合业务要求)时，即可以上线并使用平均配送时长预估模型。在业务使用时，当统计出候选配送范围的历史订单信息后，将统计出的历史订单信息输入平均配送时长预估模型，就可以计算得到候选配送范围内的平均配送时长。所述机器学习模型可以采用例如逻辑回归(LR)，GBDT(梯度增强决策树)等。
步骤130：根据每个候选配送范围的配送效率数据，从所述多个候选配送范围中确定最优的配送范围。
在预估出每个候选配送范围的配送效率数据之后，可以根据配送效率数据从中选出一个最优的候选配送范围作为目标配送请求方的配送范围。
在一些示例中，这种从多种可选项目中选出一个最优的项目，可以采用优化算法实现。具体地，根据业务需求明确优化目标后，以优化目标为跳出条件，调用优化算法进行计算，当计算结果符合跳出条件即达到优化目标后，就可以将跳出时的可选项目作为最优的项目。
在一实施例中，所述步骤130，具体可以包括：基于组合优化算法，从所述多个候选配送范围中确定最优的配送范围。
其中，所述组合优化算法的目标为：最大化候选配送范围中目标配送请求方的收益；约束条件为：候选配送范围的订单量加权面积满足预设条件。
在一实施例中，所述收益可以包括单均配送时长的成交金额。通常所述成交金额可以采用GMV(Gross Merchandise Volume,平台成交金额)表示。即所述单均配送时长的成交金额表示为单均配送时长的GMV。
具体地，最大化候选配送范围中目标配送请求方的单均配送时长的GMV，可以通过如下公式计算：
Figure PCTCN2019108572-appb-000001
Figure PCTCN2019108572-appb-000002
Figure PCTCN2019108572-appb-000003
Figure PCTCN2019108572-appb-000004
其中，c mn表示商家m是否采用候选配送范围n的0-1标识(也是组合优化算法待求解的变量)；order_num mn表示机器学习模型预测的商家m在采用候选配送范围n时的预估订单量；price m表示商家m的平均客单价(即历史订单金额之和，除以历史订单数量)；time mn表示机器学习模型预测的商家m在采用候选配送范围n时的预估平均配送时长；α表示平滑因子(避免分母为0)；A mn表示商家m的候选配送范围n的配送面积；
Figure PCTCN2019108572-appb-000005
表示人工设定的比例因子(用来调控订单量加权平均配送面积的扩大/缩小状况)。
通过针对目标配送请求提供多种候选配送范围，并利用每个候选配送范围的历史配送数据，计算出每个候选配送范围的配送效率数据，最后根据每个候选配送范围的配送效率数据从多种候选配送范围中选取一个最优的候选配送范围作为目标配送请求方的配送范围。如此，可以自动生成目标配送请求方的配送范围。
与前述配送范围的确定方法的实施例相对应，还提供了配送范围的确定装置的实施例。
在一个或多个实施例中，配送范围的确定装置可以应用在服务器上。装置实施例可以通过软件实现，也可以通过硬件或者软硬件结合的方式实现。以软件实现为例，作为一个逻辑意义上的装置，是通过其所在处理器将非易失性存储器中对应的计算机程序指令读取到内存中运行形成的。从硬件层面而言，如图5所示，为一实施例中配送范围的确定装置所在的一种硬件结构图，除了图5所示的处理器、内存、网络接口、以及非易失性存储器之外，实施例中通常根据该配送范围的确定的实际功能，还可以包括其他硬件，对此不再赘述。
请参考图6，在一种软件实施方式中，该配送范围的确定装置可以包括：
绘制单元610，被配置为针对目标配送请求方绘制多个候选配送范围；
计算单元620，被配置为根据所述候选配送范围内的历史订单信息，预估每个候选配送范围的配送效率数据；
确定单元630，被配置为根据每个候选配送范围的配送效率数据，从所述多个候选配送范围中确定最优的配送范围。
可选的，所述绘制单元610，具体被配置为：以目标配送请求方为中心、以多个不同的导航半径绘制多个候选配送范围。
可选的，所述配送效率数据包括订单量和/或平均配送时长。
可选的，所述装置还包括：优化单元，被配置为对所述候选配送范围进行优化处理；所述计算单元620，具体被配置为：根据所述候选配送范围内的历史订单信息，预估优化后的每个候选配送范围内的订单量和/或平均配送时长。
可选的，所述优化单元，具体包括：
道路边界确定子单元，被配置为确定距离所述候选配送范围的边界最近的道路边界；
重绘子单元，被配置为根据所述道路边界绘制优化后的候选配送范围。
可选的，所述优化单元，具体包括：
数据压缩子单元，被配置为对构成所述候选配送范围的配送基准点集合进行数据压缩。
可选的，所述数据压缩子单元，具体包括：
合并子单元，被配置为遍历构成所述候选配送范围的配送基准点集合；若相邻两个配送基准点之间的距离小于阈值，将所述相邻两个配送基准点合并为一个。
可选的，将所述相邻两个配送基准点合并为一个，具体包括：将所述相邻两个配送基准点中任意一个从所述配送基准点集合中删除；或者，计算所述相邻两个配送基准点的平均配送基准点，并将所述配送基准点集合中的相邻两个配送基准点替换为所述平均配送基准点。
可选的，所述优化单元，具体包括：
热点区块获取子单元，被配置为获取所述候选配送范围外预设半径内的热点区块；
重绘子单元，被配置为将所述热点区块作为候选配送范围的一部分，形成优化后的候选配送范围。
可选的，预估每个候选配送范围内的订单量和/或平均配送时长，具体包括：基于机器学习算法，预估每个候选配送范围内的订单量和/或平均配送时长。
可选的，所述确定单元630，具体包括：
确定子单元，被配置为根据每个候选配送范围的配送效率数据，基于组合优化算法，从所述多个候选配送范围中确定最优的配送范围。
可选的，所述组合优化算法的目标为：最大化候选配送范围中目标配送请求方的收益。
可选的，所述收益包括单均配送时长的成交金额。
可选的，所述确定子单元，具体被配置为：根据每个候选配送范围的订单量和/或平均配送时长，最大化候选配送范围中单均配送时长的成交金额；将单均配送时长的成交金额最大的候选配送范围确定为最优的配送范围。
上述装置中各个单元的功能和作用的实现过程具体详见上述方法中对应步骤的实现过程，在此不再赘述。
对于装置实施例而言，由于其基本对应于方法实施例，所以相关之处参见方法实施例的部分说明即可。以上所描述的装置实施例仅仅是示意性的，其中所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块以实现目的。本领域普通技术人员在不付出创造性劳动的情况下，即可以理解并实施。
以上图6描述了配送范围的确定装置的内部功能模块和结构示意，其实质上的执行主体可以为一种电子设备，包括：处理器；用于存储处理器可执行指令的存储器；其中，所述处理器被配置为以下操作：
针对目标配送请求方绘制多个候选配送范围；
根据所述候选配送范围内的历史订单信息，预估每个候选配送范围的配送效率数据；
根据每个候选配送范围的配送效率数据，从所述多个候选配送范围中确定最优的配送范围。
可选的，针对目标配送请求方绘制多个候选配送范围，具体包括：以目标配送请求方为中心、以多个不同的导航半径绘制多个候选配送范围。
可选的，所述配送效率数据包括订单量和/或平均配送时长。
可选的，还包括：对所述候选配送范围进行优化处理；预估每个候选配送范围 的配送效率数据，具体包括：预估优化后的每个候选配送范围内的订单量和/或平均配送时长。
可选的，对所述候选配送范围进行优化处理，具体包括：确定距离所述候选配送范围的边界最近的道路边界；根据所述道路边界绘制优化后的候选配送范围。
可选的，对所述候选配送范围进行优化处理，具体包括：对构成所述候选配送范围的配送基准点集合进行数据压缩。
可选的，对构成所述候选配送范围的配送基准点集合进行数据压缩，具体包括：遍历构成所述候选配送范围的配送基准点集合；若相邻两个配送基准点之间的距离小于阈值，则将所述相邻两个配送基准点合并为一个。
可选的，将所述相邻两个配送基准点合并为一个，具体包括：将所述相邻两个配送基准点中任意一个从所述配送基准点集合中删除；或者，计算所述相邻两个配送基准点的平均配送基准点，并将所述配送基准点集合中的相邻两个配送基准点替换为所述平均配送基准点。
可选的，对所述候选配送范围进行优化处理，具体包括：获取所述候选配送范围外预设半径内的热点区块；将所述热点区块作为候选配送范围的一部分，形成优化后的候选配送范围。
可选的，预估每个候选配送范围内的订单量和/或平均配送时长，具体包括：基于机器学习算法，预估每个候选配送范围内的订单量和/或平均配送时长。
可选的，从所述多个候选配送范围中确定最优的配送范围，具体包括：基于组合优化算法，从所述多个候选配送范围中确定最优的配送范围。
可选的，所述组合优化算法的目标为：最大化候选配送范围中目标配送请求方的收益。
可选的，所述收益包括单均配送时长的成交金额。
可选的，根据每个候选配送范围的配送效率数据，基于组合优化算法，从所述多个候选配送范围中确定最优的配送范围，具体包括：根据每个候选配送范围的订单量和/或平均配送时长，最大化候选配送范围中单均配送时长的成交金额；将单均配送时长的成交金额最大的候选配送范围确定为最优的配送范围。
在上述电子设备的实施例中，应理解，该处理器可以是中央处理单元(英文： Central Processing Unit，简称：CPU)，还可以是其他通用处理器、数字信号处理器(英文：Digital Signal Processor，简称：DSP)、专用集成电路(英文：Application Specific Integrated Circuit，简称：ASIC)等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等，而前述的存储器可以是只读存储器(英文：read-only memory，缩写：ROM)、随机存取存储器(英文：random access memory，简称：RAM)、快闪存储器、硬盘或者固态硬盘。结合本发明实施例所公开的方法的步骤可以直接体现为硬件处理器执行完成，或者用处理器中的硬件及软件模块组合执行完成。
本申请中的各个实施例均采用递进的方式描述，各个实施例之间相同相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。尤其，对于电子设备实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
以上所述仅为本申请的较佳实施例而已，并不用以限制本申请，凡在本申请的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本申请保护的范围之内。

Claims (17)

  1. 一种配送范围的确定方法，包括：
    针对目标配送请求方绘制多个候选配送范围；
    根据所述候选配送范围内的历史订单信息，预估每个候选配送范围的配送效率数据；
    根据每个所述候选配送范围的所述配送效率数据，从所述多个候选配送范围中确定最优的配送范围。
  2. 根据权利要求1所述的方法，其特征在于，针对所述目标配送请求方绘制多个候选配送范围，包括：
    以所述目标配送请求方为中心、以多个不同的导航半径绘制多个候选配送范围。
  3. 根据权利要求1所述的方法，其特征在于，所述配送效率数据包括订单量和/或平均配送时长。
  4. 根据权利要求3所述的方法，还包括：
    对所述候选配送范围进行优化处理，得到优化后的候选配送范围；
    预估每个所述候选配送范围的配送效率数据，包括：
    预估每个所述优化后的候选配送范围内的订单量和/或平均配送时长。
  5. 根据权利要求4所述的方法，其特征在于，对所述候选配送范围进行优化处理，包括：
    确定距离所述候选配送范围的边界最近的道路边界；
    根据所述道路边界绘制所述优化后的候选配送范围。
  6. 根据权利要求4所述的方法，其特征在于，对所述候选配送范围进行优化处理，包括：
    对构成所述候选配送范围的配送基准点集合进行数据压缩。
  7. 根据权利要求6所述的方法，其特征在于，对构成所述候选配送范围的配送基准点集合进行数据压缩，包括：
    遍历构成所述候选配送范围的所述配送基准点集合；
    若相邻两个所述配送基准点之间的距离小于阈值，则将所述相邻两个配送基准点合并为一个。
  8. 根据权利要求7所述的方法，其特征在于，将所述相邻两个配送基准点合并为一个，包括以下之一：
    将所述相邻两个配送基准点中任意一个从所述配送基准点集合中删除；或者，
    计算所述相邻两个配送基准点的平均配送基准点，并将所述配送基准点集合中的所述相邻两个配送基准点替换为所述平均配送基准点。
  9. 根据权利要求4所述的方法，其特征在于，对所述候选配送范围进行优化处理，包括：
    获取所述候选配送范围外预设半径内的热点区块；
    将所述热点区块作为所述候选配送范围的一部分，形成所述优化后的候选配送范围。
  10. 根据权利要求3所述的方法，其特征在于，预估每个所述候选配送范围内的订单量和/或平均配送时长，包括：
    基于机器学习算法，预估每个所述候选配送范围内的所述订单量和/或所述平均配送时长。
  11. 根据权利要求1所述的方法，其特征在于，从所述多个候选配送范围中确定最优的配送范围，包括：
    基于组合优化算法，从所述多个候选配送范围中确定最优的配送范围。
  12. 根据权利要求11述的方法，其特征在于，所述组合优化算法的目标为：最大化所述候选配送范围中所述目标配送请求方的收益。
  13. 根据权利要求12所述的方法，其特征在于，所述收益包括单均配送时长的成交金额。
  14. 根据权利要求13所述的方法，其特征在于，根据每个所述候选配送范围的所述配送效率数据，基于所述组合优化算法，从所述多个候选配送范围中确定最优的配送范围，包括：
    根据每个所述候选配送范围的订单量和/或平均配送时长，最大化所述候选配送范围中所述单均配送时长的成交金额；
    将所述单均配送时长的成交金额最大的所述候选配送范围确定为最优的配送范围。
  15. 一种配送范围的确定装置，包括：
    绘制单元，被配置为针对目标配送请求方绘制多个候选配送范围；
    计算单元，被配置为根据所述候选配送范围内的历史订单信息，预估每个候选配送范围的配送效率数据；
    确定单元，被配置为根据每个所述候选配送范围的配送效率数据，从所述多个候选配送范围中确定最优的配送范围。
  16. 一种计算机可读存储介质，存储有计算机程序，当执行所述计算机程序时，处 理器用于执行上述权利要求1-14中任一项所述的方法。
  17. 一种电子设备，包括：
    处理器；
    用于存储所述处理器可执行指令的存储器；
    所述处理器被配置为执行上述权利要求1-14中任一项所述的方法。
PCT/CN2019/108572 2018-09-28 2019-09-27 配送范围的确定 Ceased WO2020063875A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811143347.6A CN110969382A (zh) 2018-09-28 2018-09-28 一种配送范围的确定方法及装置
CN201811143347.6 2018-09-28

Publications (1)

Publication Number Publication Date
WO2020063875A1 true WO2020063875A1 (zh) 2020-04-02

Family

ID=69953392

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/108572 Ceased WO2020063875A1 (zh) 2018-09-28 2019-09-27 配送范围的确定

Country Status (2)

Country Link
CN (1) CN110969382A (zh)
WO (1) WO2020063875A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113706066B (zh) * 2020-05-22 2025-03-07 北京三快在线科技有限公司 配送范围的调整方法、装置、存储介质及电子设备
CN116010466A (zh) * 2021-10-21 2023-04-25 北京三快在线科技有限公司 配送范围的确定方法、装置、设备及存储介质
CN115994667B (zh) * 2023-01-17 2025-03-18 拉扎斯网络科技（上海）有限公司 配送区域确定方法、装置、电子设备及存储设备

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2002005153A2 (en) * 2000-07-07 2002-01-17 2020Me Holdings Ltd. System, method and medium for facilitating transactions over a network
CN105825360A (zh) * 2016-03-31 2016-08-03 北京小度信息科技有限公司 商户配送范围的调整方法和装置
CN107590242A (zh) * 2017-09-14 2018-01-16 北京三快在线科技有限公司 一种地址信息处理方法及装置

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2014075108A (ja) * 2012-10-04 2014-04-24 Millet Co Ltd 配送温度帯自動最適化ショッピングカートシステム
CN107437122B (zh) * 2016-05-26 2019-03-26 北京三快在线科技有限公司 确定商家配送范围的方法及装置
CN107180324A (zh) * 2017-05-05 2017-09-19 百度在线网络技术（北京）有限公司 一种拼单方法和装置
CN108364085B (zh) * 2018-01-02 2020-12-15 拉扎斯网络科技（上海）有限公司 一种外卖配送时间预测方法和装置

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2002005153A2 (en) * 2000-07-07 2002-01-17 2020Me Holdings Ltd. System, method and medium for facilitating transactions over a network
CN105825360A (zh) * 2016-03-31 2016-08-03 北京小度信息科技有限公司 商户配送范围的调整方法和装置
CN107590242A (zh) * 2017-09-14 2018-01-16 北京三快在线科技有限公司 一种地址信息处理方法及装置

Also Published As

Publication number Publication date
CN110969382A (zh) 2020-04-07

Similar Documents

Publication Publication Date Title
US11379891B2 (en) Method, apparatus, and computer program product for forecasting demand
KR101636603B1 (ko) 검색 효율이 증가된 예비-쇼핑 예약 시스템을 위한 방법 및 시스템
US20180315059A1 (en) Method and system of managing item assortment based on demand transfer
US20140052750A1 (en) Updating cached database query results
WO2019105226A1 (zh) 一种用于营销活动效果预测的方法、装置及电子设备
US20210035196A1 (en) Method and system for identifying, tracking, and predicting the location of moving merchants
CN106062811A (zh) 收益管理系统及收益管理方法
CN112215546B (zh) 一种对象页面生成方法、装置、设备及存储介质
CN108615129B (zh) 一种运力监测方法、装置及电子设备
WO2020063875A1 (zh) 配送范围的确定
US12387228B2 (en) Method, apparatus, and computer program product for forecasting demand using real time demand
TW201508672A (zh) 利用大量資料來源以定義行動使用者之即時背景，然後基於該背景決定並遞送高度相關行動訊息
CN107688955A (zh) 一种基于自适应dbscan密度聚类的城市商业圈群簇划分方法
CN114626660A (zh) 用于激增调整预报的方法和装置
CN111259272A (zh) 搜索结果排序方法及装置
CN111464337A (zh) 资源配置方法、装置、电子设备
US20220086594A1 (en) Apparatus and method for distance-based option data object filtering and modification
CN113111251A (zh) 项目推荐方法、装置及系统
US8626788B2 (en) Method for determining relative ranking data in a broker mediated geospatial information service environment
CN115271884A (zh) 一种基于多源数据的商品选品方法、装置及电子设备
US11861579B1 (en) Intelligent inventory system
US20250181593A1 (en) Server and method for processing request for search for on-demand service
Shih et al. An expected win rate-based real-time bidding strategy for branding campaigns on display advertising
US20220366461A1 (en) Method, apparatus, and computer program product for forecasting demand
US20210256509A1 (en) Planning currency exchange transactions based on international travel budgeting information

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19865459

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19865459

Country of ref document: EP

Kind code of ref document: A1