Taxi order queuing strategy determination method and system

Abstract

The embodiment of the invention provides a taxi-taking order queuing strategy determination method and system. The method comprises the following steps: acquiring a plurality of order queuing strategies and a test taxi-taking order set related to a target area; predicting a value of a target parameter corresponding to each order queuing strategy when the order queuing strategy is executed on the test taxi-taking order set; and determining at least one target order queuing strategy from the plurality of order queuing strategies based on the value of the target parameter corresponding to each order queuing strategy.

Description

Technical Field

[0001] The present invention relates to the technical field of order processing, and in particular to a method and system for determining a taxi order queuing strategy. Background Art

[0002] As cities expand and their populations grow, the number of online ride-hailing orders is usually large, and the demand for taxis by passengers often cannot be well matched with the supply of transportation capacity. At this time, it is inevitable that passengers have to queue up for orders. Especially in larger cities, there are large differences in the number of orders in different regions. At the same time, the user needs of passengers (such as vehicle model preferences, mid-way cancellation of orders, etc.) are also more complex and changeable. How to coordinate the queuing of orders and the supply of transportation capacity is a major problem faced by online ride-hailing operators. At present, relying on online experiments to verify order queuing strategies is costly and time-consuming.

[0003] Therefore, it is hoped to provide a method and system for determining a taxi order queuing strategy, which can automatically and quickly evaluate the effects of various queuing order strategies to meet the needs of actual operations, reduce testing costs, and reduce the time and effort of manual verification. Summary of the invention

[0004] One of the embodiments of the present specification provides a method for determining a taxi order queuing strategy, comprising: obtaining multiple order queuing strategies and a test taxi order set related to a target area; predicting the value of a target parameter corresponding to each order queuing strategy when the order queuing strategy is executed on the test taxi order set; and determining at least one target order queuing strategy from the multiple order queuing strategies based on the value of the target parameter corresponding to each order queuing strategy.

[0005] One of the embodiments of the present specification provides a system for determining a taxi order queuing strategy, including: an acquisition module, used to acquire multiple order queuing strategies and a test taxi order set related to a target area; a prediction module, used to predict the value of a target parameter corresponding to each order queuing strategy when the order queuing strategy is executed on the test taxi order set; a determination module, used to determine at least one target order queuing strategy from the multiple order queuing strategies based on the value of the target parameter corresponding to each order queuing strategy.

[0006] One of the embodiments of the present specification provides a device for determining a taxi order queuing strategy, including a processor, wherein the processor is used to execute the method for determining a taxi order queuing strategy as described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0008] Figure 1 It is a schematic diagram of an application scenario of a system for determining a taxi order queuing strategy according to some embodiments of this specification;

[0009] Figure 2 is an exemplary module diagram of a system for determining a taxi order queuing strategy according to some embodiments of this specification;

[0010] Figure 3 is an exemplary flow chart of a method for determining a taxi order queuing strategy according to some embodiments of this specification;

[0011] Figure 4 is an exemplary flow chart of a method for determining a region division strategy according to some embodiments of this specification;

[0012] Figure 5 is an exemplary flow chart of determining the value of a target parameter of a target area according to some embodiments of this specification;

[0013] Figure 6 is a schematic diagram of a method for determining a value of a target parameter for each test order according to some embodiments of the present specification. DETAILED DESCRIPTION

[0014] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0015] It should be understood that the "system", "device", "unit" and / or "module" used herein are a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0016] Unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0017] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0018] Figure 1 It is a schematic diagram of an application scenario of a system for determining a taxi order queuing strategy (hereinafter referred to as a taxi system) according to some embodiments of this specification.

[0019] like Figure 1 As shown, the taxi-hailing system may include a processing device 110, which is used to simulate and evaluate the effect of the order queuing strategy based on the test order set to obtain an evaluation result of the order queuing strategy. For example, the processing device 110 may predict the value of the target parameter corresponding to the order queuing strategy when the order queuing strategy is executed on the test taxi order set.

[0020] In some embodiments, the processing device 110 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), an application specific instruction set processor (ASIP), a graphics processing unit (GPU), a physical operation processing unit (PPU), a digital signal processor (DSP), a field programmable gate array (FPGA), a programmable logic device (PLD), a controller, a microcontroller unit, a reduced instruction set computer (RISC), a microprocessor, etc. or any combination thereof. In some embodiments, the processing device 110 may be a single server or a server group. The server group may be centralized or distributed (for example, the processing device 110 may be a distributed system). In some embodiments, the processing device 110 may be local or remote. In some embodiments, the processing device 110 may be implemented on a cloud platform. By way of example only, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-layer cloud, etc. or any combination thereof.

[0021] In some embodiments, the application scenario of the taxi-hailing system may also include other components, such as an online taxi-hailing platform, storage devices, networks, etc.

[0022] The online taxi-hailing platform can be used to interact with user devices on the platform, process taxi orders, etc. In some embodiments, after the processing device 110 evaluates multiple order queuing strategies, it can select at least one target order queuing strategy from them, and perform online evaluation on the online taxi-hailing platform for the at least one target order queuing strategy. Based on the result of the online evaluation, the processing device 110 can select a final order queuing strategy from the at least one target order queuing strategy for online service.

[0023] The storage device may be used to store data, instructions, and / or any other information. In some embodiments, the storage device may store data and / or information obtained from other components of the ride-hailing system (e.g., processing device 110). In some embodiments, the storage device may store data and / or instructions that processing device 110 uses to execute or use to complete the exemplary methods described in this specification. In some embodiments, the storage device may be part of processing device 110. In some embodiments, the storage device may include a mass storage device, a removable memory, etc., or any combination thereof.

[0024] The network can facilitate or enable information and / or data exchange between various components in the taxi-hailing system. The network can be any one or more of a wired network or a wireless network. In some embodiments, the network can include one or more network access points. For example, the network can include a wired or wireless network access point. In some embodiments, the network can be a point-to-point, shared, centralized, or other topological structure or a combination of multiple topological structures.

[0025] The above description is for illustrative purposes only, and actual application scenarios may vary.

[0026] It should be noted that the application scenario 100 is provided for illustrative purposes only and is not intended to limit the scope of the present application. For those skilled in the art, various modifications or changes can be made according to the description of this specification, however, these changes and modifications will not deviate from the scope of the present application.

[0027] Figure 2 It is an exemplary module diagram of a system 200 for determining a taxi order queuing strategy (hereinafter referred to as the taxi ordering system 200 ) according to some embodiments of the present specification.

[0028] like Figure 2 As shown, the taxi-hailing system 200 may include an acquisition module 210 , a prediction module 220 , and a determination module 230 . In some embodiments, the modules in the determination system 200 may be implemented by the processing device 110 .

[0029] The acquisition module 210 can be used to acquire multiple order queuing strategies and test taxi order sets related to the target area.

[0030] The prediction module 220 can be used to predict the value of the target parameter corresponding to each order queuing strategy when the order queuing strategy is executed on the test taxi order set.

[0031] In some embodiments, for each of the order queuing strategies, the prediction module 220 can also be used to process the test taxi order set based on the order queuing strategy, determine the queuing result of each test order in the test order set; based on the queuing result of each test order, predict the value of the target parameter of each test order; based on the value of the target parameter of each test order, determine the value of the target parameter of the target area.

[0032] In some embodiments, for each of the test orders, the prediction module 220 can also be used to predict the corresponding transaction probability of the test order based on the queuing result of the test order; obtain the reference price information of the test order; and determine the value of the target parameter of the test order based on the transaction probability and the reference price information.

[0033] In some embodiments, the prediction module 220 can also be used to determine multiple user demand probabilities of users of the test order based on the queuing results of the test order using multiple user demand probability prediction models, where the multiple user probability prediction models are machine learning models; based on the multiple user demand probabilities, determine the transaction probability.

[0034] The determination module 230 may be configured to determine at least one target order queuing strategy from a plurality of order queuing strategies based on a value of a target parameter corresponding to each order queuing strategy.

[0035] In some embodiments, the determination module 230 can also be used to implement each target order queuing strategy for the online taxi-hailing platform; for each target order queuing strategy, the target value of the target parameter corresponding to the target order queuing strategy is determined based on the actual transaction data of the online taxi-hailing platform; based on the target value of the target parameter corresponding to each target order queuing strategy, the final order queuing strategy is determined from at least one target order queuing strategy.

[0036] In some embodiments, the taxi-hailing system 200 further includes an area division module 240 .

[0037] The area division module 240 can be used to divide the target area into multiple area units; cluster the multiple area units multiple times to generate multiple first candidate area division strategies; select at least one target area division strategy from the multiple first candidate area division strategies; and determine the area division strategies of multiple order matching strategies based on at least one target area division strategy.

[0038] In some embodiments, multiple clusterings are performed based on different clustering features.

[0039] In some embodiments, the region division module 240 can perform at least one round of iterative processing to determine the region division strategies for multiple order matching strategies, wherein each round of iterative processing includes: obtaining an initial population, the initial population including at least one target region division strategy; transforming the target region division strategy to generate at least one second candidate region division strategy; determining a new population based on the evaluation values ​​of at least one target region division strategy and at least one second candidate region division strategy, wherein the evaluation value is related to the target parameter; using the new population as the initial population for the next round of iterative processing until a preset termination condition is met; and determining the region division strategies for multiple order matching strategies from the new population obtained from at least one round of iterative processing.

[0040] It should be noted that the above description of the taxi-hailing system and its modules is only for the convenience of description and does not limit this specification to the scope of the embodiments. It is understandable that for those skilled in the art, after understanding the principle of the system, it is possible to arbitrarily combine the various modules, or form a subsystem to connect with other modules without deviating from this principle. For example, the acquisition module 210, the prediction module 220, and the determination module 230 can be different modules in the system, or one module can implement the functions of two or more of the above modules. For example, each module can share a storage module, or each module can have its own storage module. Such variations are all within the scope of protection of this specification.

[0041] Figure 3 It is an exemplary flowchart of a method for determining a taxi order queuing strategy according to some embodiments of this specification.

[0042] In some embodiments, process 300 may be performed by a taxi-hailing system. Figure 3 As shown, process 300 includes the following steps.

[0043] Step 310, obtaining multiple order queuing strategies and a test taxi order set related to the target area.

[0044] The target area refers to the geographical area where various types of vehicles (such as express buses, taxis, etc.) carry out operation services (such as online reservations, telephone reservations, etc.). It can be determined by the vehicle operation service provider based on actual operation needs. For example, the target area can be a city, county, town, etc. or part thereof.

[0045] In some embodiments, the taxi-hailing system may include a Geographic Information System (GIS) for presenting relevant information of the target area (such as the area, shape, and location and driving path of the vehicle of the target area and / or its sub-areas) in the form of a map or the like through a terminal device.

[0046] The order queuing strategy refers to a mechanism that can be used to determine whether an order needs to be queued and its queuing order. For example, for multiple taxi orders in the target area, the order queuing strategy can be used to determine whether the multiple taxi orders need to be queued and their ranking order when queuing. The orders that need to be queued will form an order queue. The orders in the queue are arranged according to specific rules (such as the order of order issuance), and the taxi system will allocate the orders in the order queue in sequence. For orders in the queue state, they need to wait for the orders in front of the queue to be allocated before they are allocated to the service provider (such as the driver). When an order does not need to be queued, unanswered historical orders and newly issued orders will be allocated at the same time, and there is no order of allocation.

[0047] In some embodiments, the target area may include multiple sub-areas, and different sub-areas may have the same or different shapes, areas, etc. A sub-area may be regarded as a queuing fence, and orders in the same sub-area / queuing fence will be queued in the same order queuing queue. In some embodiments, each order queuing strategy includes a region division strategy for dividing the target area into multiple sub-areas. At least two order queuing strategies have different region division strategies, so as to determine the impact of different region division strategies in subsequent strategy evaluations, thereby selecting a suitable region division strategy.

[0048] In some embodiments, the region division strategy may be randomly generated. In some embodiments, the region division strategy may be generated by clustering multiple region units of the target region. For more information about the region division strategy, see Figure 4 and its description.

[0049] In some embodiments, each order queuing strategy also includes a queuing triggering strategy.

[0050] The queue triggering strategy refers to the conditions under which multiple orders need to be queued. It can be determined based on operational requirements (such as the supply and demand of vehicle services). For example, the queue triggering strategy can be determined based on the ratio between the number of orders initiated by current passengers and the number of idle vehicles.

[0051] In some embodiments, the queue triggering strategy is related to the threshold of the number of unanswered orders. Unanswered orders refer to orders that are not answered by vehicles (drivers) after passengers initiate orders, that is, taxi orders that are not assigned to vehicle services. Exemplarily, when the number of unanswered orders in a certain area (such as within a single queue fence) reaches the threshold of the number of unanswered orders, these unanswered orders need to be queued.

[0052] In some embodiments, the queuing triggering strategy is also related to the unanswered time length threshold. The unanswered time length refers to the waiting time of the unanswered order from the order placement to the current time (such as 5 minutes, 10 minutes). Exemplarily, when the number of unanswered orders in a certain area reaches the unanswered order number threshold, and the average value of the unanswered time length reaches the unanswered time length threshold, these unanswered orders need to be queued.

[0053] The queue triggering strategy can be determined in various ways. Taking the threshold of the number of unanswered orders as an example, the number threshold can be set by the user based on experience, or obtained by statistics on historical orders on the taxi system. There are at least two order queueing strategies with different queue triggering strategies, so as to determine the impact of different queue triggering strategies in subsequent strategy evaluations, and thus select the appropriate queue triggering strategy. Continuing with the example of the threshold of the number of unanswered orders, the reference range of the number threshold can be determined, and random values ​​can be taken within the reference range to generate different queue triggering strategies.

[0054] The test taxi order set refers to a set of unanswered orders used to simulate and test the order queuing strategy. The test taxi order set may include real historical orders or virtual orders. In some embodiments, the test taxi order set may include multiple unanswered taxi orders in the target area within a preset test time period. The preset test time period may be any historical moment or historical time period (such as the past hour, day, year, half a year, etc.).

[0055] In some embodiments, the taxi-hailing system may determine a test taxi-hailing order set based on historical data according to test requirements (such as the number and scale of orders). For example, the taxi-hailing system may obtain a taxi-hailing order set containing a certain number and scale (such as millions, tens of millions, etc.) in the target area within the most recent preset test time period.

[0056] Step 320 , predicting the value of the target parameter corresponding to each order queuing strategy when the order queuing strategy is executed on the test taxi order set.

[0057] The target parameters can be used to evaluate the effect of the order queuing strategy. They can be determined according to operational needs. For example, the target parameters may include but are not limited to the growth (or loss) ratio of order volume after the order queuing strategy is implemented, the number of complaints (or satisfaction) from passengers, the response rate of drivers, the cancellation rate of passengers, etc. The values ​​of the target parameters are the values ​​corresponding to the various target parameters.

[0058] In some embodiments, the target parameter includes Gross Merchandise Volume (GMV).

[0059] The commodity transaction amount may reflect the transaction amount corresponding to each order queuing strategy when the order queuing strategy is executed. It can be understood that the larger the commodity transaction amount, the better the effect of the order queuing strategy. In some embodiments, the total transaction amount corresponding to the order queuing strategy when the order queuing strategy is executed within a preset time period may be determined. The preset time period may be a time period pre-set according to actual needs, for example, one month, one quarter, etc.

[0060] In some embodiments, the taxi-hailing system can execute each order queuing strategy on the test taxi order set to determine the value of its corresponding target parameter. In this specification, this process can be referred to as offline rough arrangement (offline simulation test / operation) of the order queuing strategy.

[0061] For more information about offline coarse sorting of order queuing strategies, see Figure 5 and its description.

[0062] Step 330 , based on the value of the target parameter corresponding to each order queuing strategy, determine at least one target order queuing strategy from the multiple order queuing strategies.

[0063] The target order queuing strategy refers to the order queuing strategy used for online use or online trials.

[0064] Among them, online use refers to applying the target order queuing strategy to the online taxi platform to process actual taxi orders. Online testing is different from the above-mentioned offline simulation test. It refers to implementing the target order queuing strategy on the actual taxi order set before online use, and evaluating the target order queuing strategy based on actual transaction data.

[0065] In some embodiments, the taxi-hailing system can obtain the value of the target parameter (such as the commodity transaction amount) corresponding to each order queuing strategy through offline simulation operations of the order queuing strategy, and can sort the commodity transaction amounts (such as in descending order) to obtain multiple (such as 5, 10, etc.) commodity transaction amounts with the largest values ​​and / or the top rankings, and use the multiple order queuing strategies corresponding to the aforementioned multiple commodity transaction amounts as the target order queuing strategies.

[0066] In some embodiments, the order system may implement each target order queuing strategy for the online taxi-hailing platform; for each target order queuing strategy, based on the actual transaction data of the online taxi-hailing platform, determine the target value of the target parameter corresponding to the target order queuing strategy; based on the target value of the target parameter corresponding to each target order queuing strategy, determine the final order queuing strategy from at least one target order queuing strategy. In this specification, this process may be referred to as online fine sorting (online test / operation) of the order queuing strategy.

[0067] When implementing the target order queuing strategy on the online taxi-hailing platform, the target area can be divided into multiple sub-areas using the area division strategy in the target order queuing strategy; for the real-time orders in each sub-area, it is determined whether the real-time orders in each sub-area need to be queued based on the queue triggering strategy in the target order queuing strategy. When queuing is required in a sub-area, each queued order can be allocated in turn. Furthermore, based on the actual transaction data (such as order transaction volume and order transaction amount) of the target area during the implementation of the target order queuing strategy, the target value (also referred to as the true value of the target parameter) of the target area can be determined.

[0068] In some embodiments, when there are multiple target order queuing strategies, different target order queuing strategies can be implemented on the target area in different test time periods. For example, target order queuing strategy A can be implemented on the target area for one week, and the effect of strategy A can be evaluated based on the actual transaction data within this week; then target order queuing strategy B can be implemented on the target area for one week, and the effect of strategy B can be evaluated based on the actual transaction data within this week. In some embodiments, when there are multiple target order queuing strategies, different target order queuing strategies can be implemented on different test areas, and these target order queuing strategies can be evaluated based on the actual transaction data of different test areas. The test area can have similar characteristics (such as area, order size, etc.) to the target area.

[0069] The final order queuing strategy refers to the order queuing strategy used for actual online use.

[0070] In some embodiments, the taxi-hailing system may sort the target values ​​of the target parameters corresponding to each target order queuing strategy respectively (such as in descending order) to determine the target order queuing strategy corresponding to the largest target value as the final order queuing strategy.

[0071] In some embodiments of this specification, by performing offline rough sorting of order queuing strategies for multiple scheduled order strategies, determining a target order queuing strategy, and then performing online fine sorting of the target order queuing strategy to determine the final order queuing strategy, it is possible to perform offline evaluation of the effects of a large number of queuing strategies, thereby reducing the cost and time of online experiments. At the same time, the multiple target order queuing strategies whose target parameter values ​​meet the conditions can also obtain a final order queuing strategy with better operating effect when performing subsequent online fine sorting.

[0072] Figure 4 It is an exemplary flow chart of a method for determining a region division strategy according to some embodiments of this specification.

[0073] In some embodiments, process 400 may be performed by a taxi-hailing system. Figure 4As shown, process 400 includes the following steps.

[0074] Step 410: Divide the target area into a plurality of area units.

[0075] The area unit may refer to a unit area generated by evenly dividing the target area, and different area units may have the same shape, outline (boundary), and area. For example, the target area may be divided into a plurality of grids at a preset distance interval, and each grid corresponds to an area unit. The grid may be in the form of a rectangle, a hexagon, or other polygons.

[0076] In some embodiments, a region unit may have multiple region characteristics.

[0077] The regional characteristics may include geographical characteristics, for example, the location of the regional unit (eg, urban area, suburban area, commercial area, residential area, etc.).

[0078] The regional characteristics may include order characteristics related to taxi orders, such as the number of taxi orders within the regional unit, the frequency of order initiation, and the distribution of the number of order initiation in each time period (such as morning, noon, afternoon, 8:00-9:00, 9:00-10:00, 17:00-18:00, etc.).

[0079] The regional characteristics may include user demand characteristics related to passengers in the regional unit, for example, the number and probability of orders that are canceled before or after a response after a passenger initiates an order.

[0080] In some embodiments, the taxi-hailing system may collect statistics on historical data of taxi-hailing orders in each area unit to obtain order characteristics and user demand characteristics.

[0081] Step 420: clustering the multiple area units multiple times to generate multiple first candidate area division strategies.

[0082] The first candidate region division strategy is similar to the region division strategy described in step 310 and can be used to divide the target region into multiple sub-regions. In this specification, determining the first candidate region division strategy may also refer to determining the region division result (ie, the corresponding sub-region) corresponding to the division strategy.

[0083] In some embodiments, the taxi-hailing system may randomly cluster (also referred to as combine) multiple area units. Exemplarily, for a certain area unit, the taxi-hailing system may randomly merge the area unit with one or more of its adjacent area units. After the area units are randomly combined, multiple sub-areas of the target area may be generated, and these sub-areas may correspond to a first candidate area division strategy. By performing multiple random clusterings, different groups of sub-areas may be generated, and each group of sub-areas may correspond to a first candidate area division strategy.

[0084] In some embodiments, the taxi-hailing system can obtain the regional features of each regional unit respectively. The regional features can be represented in the form of regional feature vectors. The taxi-hailing system can cluster the regional units according to the regional feature vectors of the regional units to determine the first candidate regional division strategy. After clustering the regional units, multiple cluster centers can be determined. The regional units corresponding to each cluster center can be determined as a cluster. The regional units corresponding to each cluster can be merged into a sub-region. Multiple sub-regions corresponding to multiple cluster centers can correspond to a candidate regional division strategy. Clustering of regional units can be based on algorithms such as K-Means clustering algorithm (K-means algorithm) and density-based clustering algorithm (DBSCAN algorithm).

[0085] In some embodiments, the taxi-hailing system may determine a plurality of different first candidate area division strategies based on multiple clusterings.

[0086] In some embodiments, multiple clusterings may be performed based on different clustering features.

[0087] The clustering feature can be a combination of multiple regional features, which can be pre-configured according to demand. For example, the taxi-hailing system can combine geographic features, one or more order features, and one or more user demand features to generate multiple different clustering features, and cluster the regional units based on each clustering feature. It can be understood that different clustering features can obtain different first candidate region division strategies.

[0088] In some embodiments of the present specification, different clustering results are obtained by generating different clustering features and performing clustering multiple times, so as to obtain a plurality of different candidate region division results of the target region.

[0089] Step 430: Select at least one target area division strategy from a plurality of first candidate area division strategies.

[0090] The target area division strategy may be a strategy selected from the first candidate area division strategy and having a better area division effect. For example, when performing area division based on the target area division strategy, the value of the target indicator of the target area may be made the highest or higher than a threshold. As an example only, each first candidate area division strategy may be combined with the same queuing trigger strategy to determine a candidate order queuing strategy. The taxi-hailing system may determine the value of the target parameter corresponding to each candidate order queuing strategy in a manner similar to that described in step 320. The taxi-hailing system may select the candidate order queuing strategy with the highest or top few values ​​of the target parameter, and use the first candidate area division strategy among these candidate order queuing strategies as the target area division strategy. It should be understood that the taxi-hailing system may also evaluate the area division effect of the first candidate area division strategy based on other indicators in addition to the target parameter.

[0091] Step 440: Determine a region division strategy for a plurality of order queuing strategies based on at least one target region division strategy.

[0092] The taxi-hailing system may transform at least one target area division strategy to generate area division strategies for multiple order queuing strategies. For example, the division method of one or more sub-areas in the target area division strategy may be changed to generate a new area division strategy for generating multiple order queuing strategies. Multiple area division strategies may be generated by performing multiple different transformations on the target area division strategy. It should be understood that the target area division strategy itself may also be directly used as an area division strategy for generating multiple order queuing strategies.

[0093] In some embodiments, the taxi-hailing system may transform the target area division strategy based on a genetic algorithm to generate an area division strategy. Specifically, the taxi-hailing system may perform at least one round of iterative processing to determine the area division strategies of multiple order queuing strategies, wherein each round of iterative processing includes steps S1 to S4 as described below.

[0094] Step S1, obtaining an initial population, where the initial population includes at least one target area division strategy.

[0095] The initial population may be a collection of target region partition strategies, wherein the target region partition strategy in the initial population may also be referred to as the first individual or seed individual.

[0096] In the first round of iteration, the initial population is the target region partition strategy determined in step 430. In the second and subsequent rounds of iteration, the initial population is the new population determined in the previous round of iteration.

[0097] Step S2: transform at least one target region division strategy to generate at least one second candidate region division strategy.

[0098] Transformation processing refers to a method of adjusting the sub-regions in the target region division strategy. In some embodiments, the taxi-hailing system can adjust the boundary points (such as coordinate values) of one or more sub-regions in the first body through transformation processing, merge multiple sub-regions, etc., to change the region division method and obtain a new region division strategy (i.e., the second candidate region division strategy).

[0099] The second candidate region division strategy refers to a region division strategy generated after transforming the target region division strategy, which may also be referred to as a second individual.

[0100] In some embodiments, the taxi-hailing system can select a target area division strategy from the initial population based on a preset number of transformations (e.g., 2, 4) or a preset transformation ratio (e.g., 10%, 20%) to perform transformation processing. For each target area division strategy, different forms of transformation processing can be performed to generate different second candidate area division strategies.

[0101] In some embodiments, the taxi-hailing system can select the target area division strategy through a preset selection function based on the evaluation values ​​of individuals in the initial population. The selection function can be various preset selection operators. For example, a roulette selection operator, an expected value selection operator, a uniform sorting operator, etc. The larger the evaluation value of an individual, the greater the probability that the individual is selected as the first individual by the selection function.

[0102] Step S3, determining a new population based on evaluation values ​​of at least one target region division strategy and at least one second candidate region division strategy, wherein the evaluation value is related to the target parameter.

[0103] The evaluation value can be used to evaluate the spatial division effect of the first individual and / or the second individual. The evaluation value is related to the target parameter. For example, the evaluation value can be determined based on the commodity transaction amount. For example, when the order queuing strategy corresponding to the first individual and / or the second individual is implemented, the larger the commodity transaction amount obtained, the better the effect of the spatial division method of the first individual and / or the second individual, and the higher the evaluation value accordingly.

[0104] In some embodiments, the taxi-hailing system can generate a corresponding order queuing strategy based on each first individual and / or second individual, and perform operations such as steps 310 to 330 to determine the commodity transaction amount corresponding to each order queuing strategy. The larger the commodity transaction amount, the greater the evaluation value of the first individual and / or the second individual.

[0105] In some embodiments, the taxi-hailing system can perform screening processing on the population composed of the first individual and the second individual to generate a new population. The new population includes the first individual (target area division strategy) and the second individual (second candidate area division strategy) remaining after the screening process. For example, the taxi-hailing system can sort (such as in descending order) based on the evaluation value of each first individual and the second individual, and screen out the first individual and the second individual with higher evaluation values ​​to determine as the new population. Exemplarily, the taxi-hailing system can filter out a preset number of individuals with evaluation values ​​at the back, and / or filter out individuals with evaluation values ​​lower than a preset threshold (such as a certain commodity transaction amount).

[0106] Step S4, using the new population as the initial population for the next round of iterative processing until a preset termination condition is met.

[0107] In some embodiments, the taxi-hailing system may use the new population as the initial population for the next round of iterative processing, and repeat the operations of steps S2 to S3 until the preset termination condition is met. The individuals (region division strategies) in the new population may be designated as the target region division strategies in the next round of iteration.

[0108] The preset termination conditions may include but are not limited to reaching a preset number of iterations, the number of individuals in the new population reaching a preset number of individuals, and the difference between the maximum evaluation values ​​of multiple adjacent iterations being less than a threshold.

[0109] After the iteration is completed, the taxi-hailing system can determine the regional division strategies of multiple order queuing strategies from the new population obtained from at least one round of iterative processing.

[0110] In some embodiments, the taxi-hailing system can determine a preset number of individuals from the new population obtained by at least one round of iterative processing according to the evaluation values ​​of the individuals in the new population as the regional division strategy for multiple order queuing strategies. The preset number can be determined according to actual needs (such as the test cycle of subsequent online fine sorting). For example, the preset number can be 5 or 10.

[0111] In some embodiments of the present specification, multiple first candidate area division strategies combining different regional characteristics can be obtained by performing multiple clustering according to the regional units of the target area. A better target area division strategy is selected from the first candidate area division strategies, and transformation processing is performed based on this, so that more regional division strategies that may meet the expected operational goals can be discovered. In other words, through process 400, multiple high-quality regional division strategies can be generated for subsequent order queuing strategy evaluation (i.e., process 300) so that the target order queuing strategy finally obtained has better results. Compared with randomly generating a large number of regional division strategies, process 400 can reduce the amount of data processing and improve the efficiency of formulating order queuing strategies.

[0112] Figure 5 This is an exemplary flow chart for determining the value of a target parameter corresponding to an order queuing strategy according to some embodiments of this specification.

[0113] In some embodiments, process 500 may be executed by a taxi-hailing system. Process 500 may be executed for each order queuing strategy in step 310 to predict the value of the target parameter when each order queuing strategy is executed. Figure 5 As shown, process 500 includes the following steps.

[0114] Step 510: Process the test taxi order set based on the order queuing strategy to determine the queuing result of each test order in the test order set.

[0115] The queuing result of the test order can show whether the test order needs to be queued, and the queueing order of the test order in the order queue when the test order needs to be queued.

[0116] As described in step 310, the order queuing strategy may include a regional division strategy and a queuing triggering strategy. The taxi-hailing system may divide the target area into a plurality of sub-areas based on the regional division strategy. For each sub-area, order queuing simulation may be performed separately. Specifically, for each sub-area, the test orders located in the sub-area at the test time may be determined from the test order set, and it may be determined whether the sub-area satisfies the queuing triggering strategy at the test time. For example, if the number of test orders in the sub-area exceeds the threshold of the number of unanswered orders, it may be determined that the test orders in the sub-area need to be queued. Furthermore, the test orders may be arranged in the order of order issuance based on the order issuance time to generate an order queuing queue. When it is necessary to predict the effect of the order queuing strategy in a certain test time period, it may be determined whether each sub-area needs to queue at multiple test times in the test time period, as well as the order queuing queues at multiple test times.

[0117] Step 520 , predicting the value of the target parameter of each test order based on the queuing result of each test order.

[0118] In some embodiments, for each test order, the taxi system can predict the transaction probability corresponding to the test order based on the queuing result of the test order, and obtain the reference price information of the test order, and then determine the value of the target parameter of the test order based on the transaction probability and the reference price information.

[0119] Order completion means that the order can be finally completed and the order revenue amount is generated. The order completion probability can reflect the possibility of order completion. The completion probability can be a value in the interval [0, 1] (such as 0.5, 0.9). The larger the value, the greater the probability of completion.

[0120] It can be understood that if the order is accepted by the driver and the driver and the passenger do not cancel the order, the order is considered a completed order. In other words, the taxi-hailing system can determine the probability of the driver accepting the order, the probability of the passenger canceling the order, the probability of the driver canceling the order, etc., to obtain the probability of a completed order.

[0121] The taxi-hailing system can determine the transaction probability in a variety of ways. For example, for a test order, the taxi-hailing system can determine the transaction probability based on the sub-area where the test order is located and its ranking order, as well as the order journey length, weather conditions, etc. Exemplarily, if there are a large number of orders in the sub-area where a test order is located, when its ranking in the queue is relatively low (such as the ranking order is 7th, 10th, etc.), and at the same time, when the order has a longer journey, the probability that the user corresponding to the queued order cancels the order will be greater, and the transaction probability of the order will be smaller. In some embodiments, the transaction probability determination model can be used to process the feature information of the test order (including whether queuing is required, queue order and other feature information) to determine the transaction probability. The transaction probability determination model can be a pre-trained machine learning model.

[0122] In some embodiments, the taxi-hailing system may determine the transaction probability based on the user demand probability.

[0123] User needs may include various user needs in the taxi-hailing scenario. User needs may include the passenger's bubbling behavior, order-issuing behavior, check-box behavior, cancel behavior before answering, cancel behavior after answering, etc. User needs may also include the driver's answering behavior.

[0124] Among them, bubbling behavior refers to the behavior of not confirming the order after the passenger inputs the starting and ending positions of the order; ordering behavior refers to the behavior of sending an order request after the passenger inputs the starting and ending positions of the order; checking behavior refers to the behavior of passengers checking the options corresponding to various types of vehicles (such as express, taxi, etc.); canceling before answering behavior refers to the behavior of passengers canceling their orders before the passenger's order is assigned a vehicle; canceling after answering behavior refers to the behavior of passengers canceling their orders before getting on the vehicle after the passenger's order is assigned a vehicle. The driver's answering behavior refers to the driver's behavior of accepting the order. In some embodiments, the taxi-hailing system can be divided into multiple different types of vehicles according to the vehicles corresponding to the drivers, and the answering probability corresponding to the drivers of each type of vehicle can also be determined.

[0125] In some embodiments, the taxi-hailing system may determine multiple user demand probabilities of users of the test order based on the queuing results of the test order using multiple user demand probability prediction models, wherein the multiple user probability prediction models are machine learning models.

[0126] The user demand probability prediction model can be used to determine the probabilities corresponding to various user demands of users (such as passengers and drivers), which can be a trained machine learning model, such as a deep learning model or other customized neural network model. In some embodiments, different user demands can correspond to different user demand probability prediction models.

[0127] In some embodiments, Figure 6 As shown, the multiple user demand probability models may include a passenger bubble order probability model 621, a passenger check model 622, a pre-answer cancellation rate model 623, and a competitive answer model 624.

[0128] Among them, Figure 6 As shown, by executing the order queuing strategy 612 on the test taxi order set 611, a queuing result can be obtained, which can show whether each test order is queued 613 and the queuing order 614 of each test order.

[0129] For each test order, its queuing result (whether to queue 613) and queuing order 614 can be input into the passenger bubble order probability model 621, and the order probability 631 of the test order can be output.

[0130] The inputs of the passenger selection model 622, the pre-answer cancellation probability model 623, and the competitive answer model 624 may be the same as the inputs of the passenger bubbling order probability model 621. In some embodiments, the inputs of the user demand probability prediction model may also include other features of the test order, such as the trip length, passenger features, order time, location and weather in the area, etc.

[0131] Among them, the output of the passenger selection model 622 includes the category selection probability 632, which includes the probability of each car model being selected by the passenger; the output of the pre-answer cancellation rate model 623 includes the pre-answer cancellation probability 633; ​​the output of the competitive response model 624 includes the car model response probability 634, which includes the probability that the order is answered by the driver of each car model.

[0132] In some embodiments, the user demand probability prediction model can be obtained based on training. Taking the passenger bubble order probability model as an example, its training samples can be multiple sample taxi orders in the corresponding queue state in the sample area (such as the target area or its sub-area), and the sample area represents a sample queue queue. The total number of sample queue taxi orders represents the length of the sample queue queue, and each sample queue order has a corresponding sample queue order in the sample queue queue. Among them, multiple sample queue taxi orders can be obtained based on the historical taxi order data corresponding to the sample area; the training label can be determined according to the proportion of the actual number of orders issued in the multiple sample queue taxi orders (such as 60%, 85%), and the value of the proportion can be expressed as the probability of issuing an order. The training label can be manually labeled or labeled in other ways.

[0133] It can be understood that through the user demand probability prediction model of the artificial intelligence algorithm, deep learning is carried out on different areas, different queuing situations (such as queue length, ranking order) and other influencing characteristics (such as weather, itinerary), in order to obtain the commonalities or rules of user needs corresponding to the order users, and then quickly and accurately predict the probability of various types of user needs, reducing the time and energy required for manual judgment.

[0134] In some embodiments, the probability of cancellation after answering 635 can be determined based on preset rules 625. The preset rules 625 can be determined based on experience. In some embodiments, the preset rules 625 can be related to the length of the order journey. The longer the length of the order journey, the greater the probability of cancellation after answering. Exemplarily, the preset rules 625 may include the probability of cancellation after answering 635 corresponding to different journey lengths. The probability of cancellation after answering 635 is determined based on the preset rules 625 rather than the machine learning model because the probability of cancellation after answering 635 is usually related to the distance from the driver assigned to the order to the starting point of the order, but the driver assigned to the order cannot be accurately determined in the queuing simulation, and thus the complete input of the machine learning model cannot be obtained. Therefore, the probability of cancellation after answering 635 can be determined more accurately based on the preset rules 625.

[0135] The reference price information may reflect the unit price information of the taxi order. In some embodiments, the reference price information includes the unit price corresponding to each vehicle type. In some embodiments, the taxi system may estimate the reference price information based on the vehicle type (model) selected by the user in each taxi order and other order information (such as the starting point, the end point, the trip length, weather information, etc.).

[0136] In some embodiments, the taxi-hailing system may determine the value of the target parameter of the test order based on the following formula (1): G= E cr ×(1-Cancel p)×Price×Choose×Response×(1- Cancel e ) Formula (1)

[0137] Among them, G represents the gross merchandise volume (GMV), E represents cr Indicates the probability of placing an order, Cancel p represents the probability of cancellation before answering, Price represents the reference price, Choose represents the probability of selecting a category, Response represents the probability of responding to a model, and Cancel represents the probability of selecting a model. e Indicates the probability of cancellation after response. It should be noted that when there are multiple models, Price includes the unit price corresponding to each model, and Choose and Response include the probability corresponding to each model. When calculating GMV, it is necessary to determine the GMV corresponding to each model and add them up.

[0138] Step 530 , based on the value of the target parameter of each test order, determine the value of the target parameter of the target area.

[0139] In some embodiments, the taxi-hailing system may obtain the GMV corresponding to multiple test orders in each sub-area, and determine the GMV of the target area based on the sum of the GMV of all sub-areas.

[0140] In some embodiments of this specification, by implementing an order queuing strategy on a test order set and predicting the order transaction probability and target parameter value based on the queuing results, the impact of the order queuing strategy on the target parameters of the entire target area can be accurately evaluated. In addition, by introducing a probability prediction model for various user demands, the probability of various user demands and the final order transaction rate can be automatically determined, reducing the time and energy loss caused by manual analysis and improving the efficiency of the entire simulation test process.

[0141] It should be noted that the above description of the relevant process is only for example and explanation, and does not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to the process under the guidance of this specification. However, these modifications and changes are still within the scope of this specification.

[0142] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

[0143] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.

[0144] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.

[0145] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0146] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values ​​is as accurate as possible within the feasible range.

[0147] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, etc., cited in this specification are hereby incorporated by reference in their entirety. Except for application history documents that are inconsistent with or conflicting with the contents of this specification, documents that limit the broadest scope of the claims of this specification (currently or later attached to this specification) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the materials attached to this specification are inconsistent or conflicting with the contents described in this specification, the descriptions, definitions, and / or use of terms in this specification shall prevail.

[0148] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A method for determining a taxi order queuing strategy, characterized in that: include: Obtain multiple order queuing strategies and test taxi order sets related to the target area; Predicting the value of the target parameter corresponding to each order queuing strategy when the order queuing strategy is executed on the test taxi order set; At least one target order queuing strategy is determined from the plurality of order queuing strategies based on the value of the target parameter corresponding to each of the order queuing strategies.

2. The method according to claim 1, characterized in that Each of the order queuing strategies includes a region division strategy for dividing the target region into a plurality of sub-regions; The regional division strategies of the multiple order queuing strategies are generated based on the following steps: Dividing the target area into a plurality of area units; Performing multiple clustering on the multiple area units to generate multiple first candidate area division strategies; Selecting at least one target area division strategy from the plurality of first candidate area division strategies; Based on the at least one target area division strategy, a regional division strategy of the multiple order matching strategies is determined.

3. The method according to claim 2, characterized in that: The multiple clusterings are performed based on different clustering features.

4. The method according to claim 3, characterized in that The determining of the regional division strategy of the plurality of order matching strategies based on the at least one target regional division strategy includes: Perform at least one round of iterative processing, each round of iterative processing includes: Acquire an initial population, wherein the initial population includes the at least one target area division strategy; Transforming the target region division strategy to generate at least one second candidate region division strategy; determining a new population based on evaluation values ​​of the at least one target region partitioning strategy and the at least one second candidate region partitioning strategy, wherein the evaluation value is related to the target parameter; Using the new population as the initial population for the next round of iterative processing until a preset termination condition is met; From the new population obtained by the at least one round of iterative processing, a regional division strategy for the multiple order pairing strategies is determined.

5. The method according to claim 1, characterized in that Each of the order queuing strategies includes a queuing triggering strategy, and the queuing triggering strategy is related to a threshold value of the number of unanswered orders.

6. The method according to claim 1, characterized in that For each of the order queuing strategies, when the prediction implements the order queuing strategy on the test taxi order set, the target parameter value of the target area includes: For each of the order queuing strategies, Processing the test taxi order set based on the order queuing strategy to determine a queuing result for each test order in the test order set; Based on the queuing result of each of the test orders, predict the value of the target parameter of each of the test orders; Based on the value of the target parameter of each of the test orders, a value of the target parameter of the target area is determined.

7. The method according to claim 6, characterized in that The predicting the value of the target parameter of each test order based on the queuing result of each test order includes: For each of said test orders, Based on the queuing result of the test order, predict the transaction probability corresponding to the test order; Obtaining reference price information of the test order; Based on the transaction probability and the reference price information, a value of a target parameter of the test order is determined.

8. The method according to claim 7, characterized in that The predicting of the transaction probability corresponding to the test order includes: Based on the queuing result of the test order, multiple user demand probabilities of users of the test order are determined using multiple user demand probability prediction models, wherein the multiple user probability prediction models are machine learning models; The transaction probability is determined based on the multiple user demand probabilities.

9. The method according to claim 1, characterized in that: The method further comprises: Implementing each of the target order queuing strategies on the online taxi-hailing platform; For each of the target order queuing strategies, based on actual transaction data of the online taxi-hailing platform, determine a target value of a target parameter corresponding to the target order queuing strategy; Based on the target value of the target parameter corresponding to each of the target order queuing strategies, a final order queuing strategy is determined from the at least one target order queuing strategy.

10. The method according to claim 1, characterized in that The target parameter includes commodity transaction amount.

11. A system for determining a taxi order queuing strategy, characterized in that: include: An acquisition module is used to acquire multiple order queuing strategies and test taxi order sets related to a target area; A prediction module, used for predicting the value of the target parameter corresponding to each order queuing strategy when the order queuing strategy is executed on the test taxi order set; A determination module is used to determine at least one target order queuing strategy from the plurality of order queuing strategies based on the value of the target parameter corresponding to each of the order queuing strategies.

12. A device for determining a taxi order queuing strategy, characterized in that: It includes a processor, and the processor is used to execute the method for determining the taxi order queuing strategy according to any one of claims 1 to 10.

Images