A method and device for recalling lost users of online car-hailing

Abstract

The application discloses a method and device for recalling lost users of online car-hailing, wherein the method comprises the following steps: obtaining historical online car-hailing order data; analyzing user behavior data from the historical online car-hailing order data; training a first recall probability prediction model in combination with the user behavior data and historical recall data under the condition of using a recall strategy; training a second recall probability prediction model in combination with the user behavior data and historical recall data under the condition of not using the recall strategy; calculating the subsidy gain value of each lost user based on the user behavior data through the first recall probability prediction model and the second recall probability prediction model; and sorting the lost users based on the subsidy gain value to select to send a recall short message and issue corresponding coupons and subsidies. The application can selectively recall lost users to reduce the waste of server resources.

Description

Technical Field

[0001] This invention relates to the technical field of predicting and calculating the recall of lost users on ride-hailing platforms, and specifically to a method and apparatus for recalling lost users on ride-hailing platforms. Background Technology

[0002] With the development of the internet, ride-hailing platforms have also developed accordingly, and ride-hailing has become widespread in people's lives. Users typically choose from multiple ride-hailing platforms. When these platforms attract new users, they often offer coupons and ride subsidies. However, when these coupons expire or the subsidies become invalid, users may be attracted to other platforms, inevitably leading to user churn. To win back lost users and reduce the loss of existing users, ride-hailing platforms generally issue coupons and subsidies to churned users and send recall text messages in an attempt to bring them back.

[0003] However, existing ride-hailing platforms typically use a method of indiscriminately issuing coupons and subsidies to all lost users and sending recall text messages. This method is not suitable for all lost users and may even be counterproductive, as it may interfere with some lost users. Sending a large number of invalid coupons and subsidies will waste a lot of server resources, and such resource waste will not increase the probability of user recall.

[0004] Therefore, there is an urgent need to invent a method for recalling lost ride-hailing users that can accurately screen them and reduce resource waste. Summary of the Invention

[0005] In order to overcome the technical defects of the prior art that cause waste of resources, the present invention provides a method and apparatus for recalling lost users of ride-hailing services.

[0006] To solve the above problems, the present invention is implemented according to the following technical solution:

[0007] In a first aspect, the present invention discloses a method for recalling lost users of ride-hailing services, comprising the following steps:

[0008] Step S1: Obtain historical ride-hailing order data;

[0009] Step S2: Analyze user behavior data from historical ride-hailing order data;

[0010] Step S3: Combine user behavior data and historical recall data, and train the first recall probability prediction model under the recall strategy.

[0011] Step S4: Combine user behavior data and historical recall data to train a second recall probability prediction model without employing a recall strategy;

[0012] Step S5: Based on user behavior data, calculate the subsidy gain value for each churned user using the first recall probability prediction model and the second recall probability prediction model.

[0013] Step S6: Sort churned users based on subsidy gain value to select those to send recall SMS messages to and issue corresponding coupons and subsidies.

[0014] As a preferred embodiment of the first aspect, the user behavior data includes the user's historical ride-hailing order records on the platform, the user's click behavior records in the application or WeChat mini-program, the user's historical ride-hailing time preference information, the user's historical ride-hailing location preference information, the user's historical ride-hailing price information, historical coupon usage information, and the user's historical evaluation information.

[0015] As a preferred implementation of the first aspect, the step of training a first recall probability prediction model by combining user behavior data and historical recall data and employing a recall strategy specifically includes the following sub-steps:

[0016] Acquire historical recall data, which includes user behavior data and data on the relationship between recall strategies and recall success rates.

[0017] Extract historical ride-hailing order records from the user behavior data corresponding to each user;

[0018] Based on historical ride-hailing order records, if the last ride-hailing order is more than a preset period away from the current timestamp, the user will be marked as a churned user.

[0019] From the user behavior data corresponding to churned users, we extract users' click behavior records on the application or WeChat mini program, users' historical ride-hailing time preference information, users' historical ride-hailing location preference information, users' historical ride-hailing price information, historical coupon usage information, and users' historical evaluation information, and then integrate the data to form a training dataset.

[0020] The platform's recall strategy is obtained, and based on the training dataset and the recall strategy, a first recall probability prediction model is trained using the machine learning random forest algorithm.

[0021] As a preferred embodiment of the first aspect, the step of calculating the subsidy gain value for each user based on user behavior data using a first recall probability prediction model and a second recall probability prediction model specifically includes the following sub-steps:

[0022] The user behavior data of churned users are respectively input into the first recall probability prediction model and the second recall probability prediction model;

[0023] The prediction results are obtained from the first recall probability prediction model and the second recall probability prediction model, respectively;

[0024] Based on the prediction results of the first and second recall probability prediction models, the subsidy gain value for each churned user is calculated.

[0025] The specific calculation of the subsidy gain value is: P(Yi〡XiRl)-P(Yi〡XiRh);

[0026] Wherein, Y is the recall probability, X is the user profile feature, R is the subsidy strategy, P(Yi〡XiRl) represents the order probability of user i with user profile Xi under the Rl subsidy strategy, and P(Yi〡XiRh) represents the order probability of user i with user profile Xi without the subsidy strategy.

[0027] As a preferred implementation of the first aspect, the step of sorting users based on subsidy gain values ​​to select those to send recall SMS messages to and issue corresponding coupons and subsidies specifically includes the following sub-steps:

[0028] Obtain the subsidy gain value corresponding to each churned user;

[0029] Based on the subsidy gain value, churned users are sorted in descending order of subsidy gain value;

[0030] Create a subsidy sorting list and add the churned users to the subsidy sorting list in sequence;

[0031] Obtain the preset subsidy strategy, extract the information on the number or proportion of churned users recalled, and select churned users that meet the subsidy strategy.

[0032] Based on the subsidy strategy, corresponding coupons and subsidies are issued to churned users who meet the subsidy strategy.

[0033] Send recall SMS messages to each churned user who meets the subsidy policy criteria.

[0034] Secondly, the present invention also discloses a device for recalling lost ride-hailing users, comprising:

[0035] Data acquisition module M1 is used to acquire historical ride-hailing order data;

[0036] The data analysis module M2 is used to analyze user behavior data from historical ride-hailing order data;

[0037] The first training module M3 is used to train the first recall probability prediction model by combining user behavior data and historical recall data and adopting a recall strategy.

[0038] The second training module M4 is used to train the second recall probability prediction model by combining user behavior data and historical recall data without employing a recall strategy.

[0039] The prediction calculation module M5 is used to calculate the subsidy gain value of each churned user based on user behavior data through the first recall probability prediction model and the second recall probability prediction model.

[0040] The subsidy distribution module M6 is used to sort churned users based on subsidy gain values, select those to send recall SMS messages, and distribute corresponding coupons and subsidies.

[0041] As a preferred implementation of the second aspect, the user behavior data includes the user's historical ride-hailing order records on the platform, the user's click behavior records in the application or WeChat mini-program, the user's historical ride-hailing time preference information, the user's historical ride-hailing location preference information, the user's historical ride-hailing price information, historical coupon usage information, and the user's historical evaluation information.

[0042] As a preferred embodiment of the second aspect, the first training module M3, when running, specifically performs the following sub-steps:

[0043] Acquire historical recall data, which includes user behavior data and data on the relationship between recall strategies and recall success rates.

[0044] Extract historical ride-hailing order records from the user behavior data corresponding to each user;

[0045] Based on historical ride-hailing order records, if the last ride-hailing order is more than a preset period away from the current timestamp, the user will be marked as a churned user.

[0046] From the user behavior data corresponding to churned users, we extract users' click behavior records on the application or WeChat mini program, users' historical ride-hailing time preference information, users' historical ride-hailing location preference information, users' historical ride-hailing price information, historical coupon usage information, and users' historical evaluation information, and then integrate the data to form a training dataset.

[0047] The platform's recall strategy is obtained, and based on the training dataset and the recall strategy, a first recall probability prediction model is trained using the machine learning random forest algorithm.

[0048] In a preferred embodiment of the second aspect, the prediction calculation module M5, when running, specifically performs the following sub-steps:

[0049] The user behavior data of churned users are respectively input into the first recall probability prediction model and the second recall probability prediction model;

[0050] The prediction results are obtained from the first recall probability prediction model and the second recall probability prediction model, respectively;

[0051] Based on the prediction results of the first and second recall probability prediction models, the subsidy gain value for each churned user is calculated.

[0052] The specific calculation of the subsidy gain value is: P(Yi〡XiRl)-P(Yi〡XiRh);

[0053] Wherein, Y is the recall probability, X is the user profile feature, R is the subsidy strategy, P(Yi〡XiRl) represents the order probability of user i with user profile Xi under the Rl subsidy strategy, and P(Yi〡XiRh) represents the order probability of user i with user profile Xi without the subsidy strategy.

[0054] As a preferred embodiment of the second aspect, the subsidy disbursement module M6, when running, specifically performs the following sub-steps:

[0055] Obtain the subsidy gain value corresponding to each churned user;

[0056] Based on the subsidy gain value, churned users are sorted in descending order of subsidy gain value;

[0057] Create a subsidy sorting list and add the churned users to the subsidy sorting list in sequence;

[0058] Obtain the preset subsidy strategy, extract the information on the number or proportion of churned users recalled, and select churned users that meet the subsidy strategy.

[0059] Based on the subsidy strategy, corresponding coupons and subsidies are issued to churned users who meet the subsidy strategy.

[0060] Send recall SMS messages to each churned user who meets the subsidy policy criteria.

[0061] Compared with the prior art, the beneficial effects of the present invention are:

[0062] This invention employs a differential model, designing a first recall probability prediction model and a second recall probability prediction model. The two models are trained with and without subsidy strategies, respectively, to predict data representing the degree to which each churned user is affected by coupons and subsidies. Differential operations are then performed to obtain a subsidy gain value. Churned users are ranked based on the subsidy gain value, enabling selective recall of appropriate churned users according to the subsidy strategy, thereby significantly reducing the waste of server resources. Attached Figure Description

[0063] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, wherein:

[0064] Figure 1 This is a flowchart illustrating the method for recalling lost ride-hailing users according to the present invention.

[0065] Figure 2 This is a schematic diagram of the structure of the ride-hailing user recall device of the present invention. Detailed Implementation

[0066] Preferred embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While preferred embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure will be thorough and complete, and will fully convey the scope of the present disclosure to those skilled in the art.

[0067] The term "comprising" and its variations as used herein signify open inclusion, i.e., "including but not limited to". Unless otherwise stated, the term "or" means "and / or". The term "based on" means "at least partially based on". The terms "one example embodiment" and "one embodiment" mean "at least one example embodiment". The term "another embodiment" means "at least one additional embodiment". The terms "first", "second", etc., may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0068] Access devices and servers can be connected directly or indirectly via wired or wireless communication. Access devices can be terminals or servers. The target application runs on the access device. The target application is an application capable of making data requests to the server, such as social applications, payment applications, and game applications. The server can be an application server providing services to the target application, or a proxy server distinct from the application server corresponding to the target application. The server identifies whether each access device is malicious and intercepts data packets from malicious devices. When the server acts as a proxy server, it forwards data packets not originating from malicious devices to the application server. Terminals can be desktop terminals or mobile terminals. Mobile terminals can include smartphones, tablets, laptops, desktop computers, smart speakers, smartwatches, etc., but are not limited to these. Servers can be independent physical servers, server clusters or distributed systems composed of multiple physical servers, or cloud servers providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms.

[0069] Example 1

[0070] like Figure 1 As shown, in a first aspect, the present invention discloses a method for recalling lost ride-hailing users, comprising the following steps:

[0071] Step S1: Obtain historical ride-hailing order data;

[0072] Step S2: Analyze user behavior data from historical ride-hailing order data;

[0073] Specifically, the user behavior data includes the user's historical ride-hailing order records on the platform, the user's click behavior records in the application or WeChat mini program, the user's historical ride-hailing time preference information, the user's historical ride-hailing location preference information, the user's historical ride-hailing price information, historical coupon usage information, and the user's historical review information.

[0074] Step S3: Combine user behavior data and historical recall data, and train the first recall probability prediction model under the recall strategy.

[0075] Specifically, it includes the following sub-steps:

[0076] Acquire historical recall data, which includes user behavior data and data on the relationship between recall strategies and recall success rates.

[0077] Extract historical ride-hailing order records from the user behavior data corresponding to each user;

[0078] Based on historical ride-hailing order records, if the last ride-hailing order is more than a preset period away from the current timestamp, the user will be marked as a churned user.

[0079] From the user behavior data corresponding to churned users, we extract users' click behavior records on the application or WeChat mini program, users' historical ride-hailing time preference information, users' historical ride-hailing location preference information, users' historical ride-hailing price information, historical coupon usage information, and users' historical evaluation information, and then integrate the data to form a training dataset.

[0080] The platform's recall strategy is obtained, and based on the training dataset and the recall strategy, a first recall probability prediction model is trained using the machine learning random forest algorithm.

[0081] Step S4: Combine user behavior data and historical recall data to train a second recall probability prediction model without employing a recall strategy;

[0082] Specifically, it includes the following sub-steps:

[0083] Obtain historical recall data;

[0084] Extract historical ride-hailing order records from the user behavior data corresponding to each user;

[0085] Based on historical ride-hailing order records, if the last ride-hailing order is more than a preset period away from the current timestamp, the user will be marked as a churned user.

[0086] From the user behavior data corresponding to churned users, we extract users' click behavior records on the application or WeChat mini program, users' historical ride-hailing time preference information, users' historical ride-hailing location preference information, users' historical ride-hailing price information, historical coupon usage information, and users' historical evaluation information, and then integrate the data to form a training dataset.

[0087] Based on the training dataset, a second recall probability prediction model is trained using the machine learning random forest algorithm.

[0088] Step S5: Based on user behavior data, calculate the subsidy gain value for each churned user using the first recall probability prediction model and the second recall probability prediction model.

[0089] Specifically, it includes the following sub-steps:

[0090] The user behavior data of churned users are respectively input into the first recall probability prediction model and the second recall probability prediction model;

[0091] The prediction results are obtained from the first recall probability prediction model and the second recall probability prediction model, respectively;

[0092] Based on the prediction results of the first and second recall probability prediction models, the subsidy gain value for each churned user is calculated.

[0093] The specific calculation of the subsidy gain value is: P(Yi〡XiRl)-P(Yi〡XiRh);

[0094] Wherein, Y is the recall probability, X is the user profile feature, R is the subsidy strategy, P(Yi〡XiRl) represents the order probability of user i with user profile Xi under the Rl subsidy strategy, and P(Yi〡XiRh) represents the order probability of user i with user profile Xi without the subsidy strategy.

[0095] Step S6: Sort churned users based on subsidy gain value to select those to send recall SMS messages to and issue corresponding coupons and subsidies.

[0096] Specifically, it includes the following sub-steps:

[0097] Obtain the subsidy gain value corresponding to each churned user;

[0098] Based on the subsidy gain value, churned users are sorted in descending order of subsidy gain value;

[0099] Create a subsidy sorting list and add the churned users to the subsidy sorting list in sequence;

[0100] Obtain the preset subsidy strategy, extract the information on the number or proportion of churned users recalled, and select churned users that meet the subsidy strategy.

[0101] Based on the subsidy strategy, corresponding coupons and subsidies are issued to churned users who meet the subsidy strategy.

[0102] Send recall SMS messages to each churned user who meets the subsidy policy criteria.

[0103] In summary, this invention creatively employs a differential model to design a first recall probability prediction model and a second recall probability prediction model. These two models are trained with and without subsidy strategies, respectively, to predict the degree to which each churned user is affected by coupons and subsidies. Differential operations are then performed to derive the subsidy gain value. Churned users are then ranked based on this subsidy gain value, enabling selective recall of appropriate churned users according to the subsidy strategy, thereby significantly reducing the waste of server resources.

[0104] Other steps in the method for recalling lost ride-hailing users described in this embodiment are found in existing technologies.

[0105] Example 2

[0106] like Figure 2 As shown, in a second aspect, the present invention also discloses a device for recalling lost ride-hailing users, comprising:

[0107] Data acquisition module M1 is used to acquire historical ride-hailing order data;

[0108] The data analysis module M2 is used to analyze user behavior data from historical ride-hailing order data;

[0109] The first training module M3 is used to train the first recall probability prediction model by combining user behavior data and historical recall data and adopting a recall strategy.

[0110] The second training module M4 is used to train the second recall probability prediction model by combining user behavior data and historical recall data without employing a recall strategy.

[0111] The prediction calculation module M5 is used to calculate the subsidy gain value of each churned user based on user behavior data through the first recall probability prediction model and the second recall probability prediction model.

[0112] The subsidy distribution module M6 is used to sort churned users based on subsidy gain values, select those to send recall SMS messages, and distribute corresponding coupons and subsidies.

[0113] As a preferred implementation of the second aspect, the user behavior data includes the user's historical ride-hailing order records on the platform, the user's click behavior records in the application or WeChat mini-program, the user's historical ride-hailing time preference information, the user's historical ride-hailing location preference information, the user's historical ride-hailing price information, historical coupon usage information, and the user's historical evaluation information.

[0114] As a preferred embodiment of the second aspect, the first training module M3, when running, specifically performs the following sub-steps:

[0115] Acquire historical recall data, which includes user behavior data and data on the relationship between recall strategies and recall success rates.

[0116] Extract historical ride-hailing order records from the user behavior data corresponding to each user;

[0117] Based on historical ride-hailing order records, if the last ride-hailing order is more than a preset period away from the current timestamp, the user will be marked as a churned user.

[0118] From the user behavior data corresponding to churned users, we extract users' click behavior records on the application or WeChat mini program, users' historical ride-hailing time preference information, users' historical ride-hailing location preference information, users' historical ride-hailing price information, historical coupon usage information, and users' historical evaluation information, and then integrate the data to form a training dataset.

[0119] The platform's recall strategy is obtained, and based on the training dataset and the recall strategy, a first recall probability prediction model is trained using the machine learning random forest algorithm.

[0120] In a preferred embodiment of the second aspect, the prediction calculation module M5, when running, specifically performs the following sub-steps:

[0121] The user behavior data of churned users are respectively input into the first recall probability prediction model and the second recall probability prediction model;

[0122] The prediction results are obtained from the first recall probability prediction model and the second recall probability prediction model, respectively;

[0123] Based on the prediction results of the first and second recall probability prediction models, the subsidy gain value for each churned user is calculated.

[0124] The specific calculation of the subsidy gain value is: P(Yi〡XiRl)-P(Yi〡XiRh);

[0125] Wherein, Y is the recall probability, X is the user profile feature, R is the subsidy strategy, P(Yi〡XiRl) represents the order probability of user i with user profile Xi under the Rl subsidy strategy, and P(Yi〡XiRh) represents the order probability of user i with user profile Xi without the subsidy strategy.

[0126] As a preferred embodiment of the second aspect, the subsidy disbursement module M6, when running, specifically performs the following sub-steps:

[0127] Obtain the subsidy gain value corresponding to each churned user;

[0128] Based on the subsidy gain value, churned users are sorted in descending order of subsidy gain value;

[0129] Create a subsidy sorting list and add the churned users to the subsidy sorting list in sequence;

[0130] Obtain the preset subsidy strategy, extract the information on the number or proportion of churned users recalled, and select churned users that meet the subsidy strategy.

[0131] Based on the subsidy strategy, corresponding coupons and subsidies are issued to churned users who meet the subsidy strategy.

[0132] Send recall SMS messages to each churned user who meets the subsidy policy criteria.

[0133] In summary, when the device described in this embodiment of the invention is running, it can execute all the steps of the method for recalling lost ride-hailing users described in Embodiment 1.

[0134] Other structures of the ride-hailing user recall device described in this embodiment are referenced in the prior art.

[0135] Example 3

[0136] This invention also discloses an electronic device, including at least one processor and a memory communicatively connected to the at least one processor. The memory stores instructions executable by the at least one processor. When the at least one processor executes the instructions, it specifically implements the following steps: acquiring historical ride-hailing order data; analyzing user behavior data from the historical ride-hailing order data; combining the user behavior data and historical recall data, training a first recall probability prediction model with a recall strategy; combining the user behavior data and historical recall data, training a second recall probability prediction model without a recall strategy; calculating the subsidy gain value for each churned user based on the user behavior data, using the first and second recall probability prediction models; and ranking the churned users based on the subsidy gain value to select which users to send recall SMS messages to and issue corresponding coupons and subsidies.

[0137] Example 4

[0138] This invention also discloses a storage medium storing a computer program. When the computer program is executed by a processor, it specifically implements the following steps: acquiring historical ride-hailing order data; analyzing user behavior data from the historical ride-hailing order data; combining user behavior data and historical recall data, and training a first recall probability prediction model with a recall strategy; combining user behavior data and historical recall data, and training a second recall probability prediction model without a recall strategy; calculating the subsidy gain value for each churned user based on the user behavior data, using the first and second recall probability prediction models; and ranking churned users based on the subsidy gain value to select which users to send recall SMS messages to and issue corresponding coupons and subsidies.

[0139] This disclosure can be a method, apparatus, system, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of this disclosure.

[0140] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination thereof. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0141] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0142] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, Java, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0143] Various aspects of this disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0144] These computer-readable program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processing unit of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner. Thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0145] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0146] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction, which includes one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0147] Various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical applications, or improvements to the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for recalling lost ride-hailing users, characterized in that, Includes the following steps: Acquire historical ride-hailing order data and historical recall data; the historical recall data includes user behavior data, recall strategies, and the relationship between recall success rates. User behavior data is analyzed from historical ride-hailing order data. This user behavior data includes users' historical ride-hailing order records on the platform, users' click behavior records on the application or WeChat mini-program, users' historical ride-hailing time preference information, users' historical ride-hailing location preference information, users' historical ride-hailing price information, historical coupon usage information, and users' historical review information. By combining user behavior data and historical recall data, and employing a recall strategy, a first recall probability prediction model is trained. The step of training a first recall probability prediction model by combining user behavior data and historical recall data, under the condition of adopting a recall strategy, includes: Extract historical ride-hailing order records from the user behavior data corresponding to each user; Based on historical ride-hailing order records, if the last ride-hailing order is more than a preset period away from the current timestamp, the user will be marked as a churned user. From the user behavior data corresponding to churned users, we extract users' click behavior records on the application or WeChat mini program, users' historical ride-hailing time preference information, users' historical ride-hailing location preference information, users' historical ride-hailing price information, historical coupon usage information, and users' historical evaluation information, and then integrate the data to form a training dataset. The platform's recall strategy is obtained, and a first recall probability prediction model is trained based on the training dataset and the recall strategy. By combining user behavior data and historical recall data, a second recall probability prediction model is trained without employing a recall strategy. Based on user behavior data, the subsidy gain value for each churned user is calculated using the first and second recall probability prediction models. The calculation of the subsidy gain value is specifically as follows: P(Y) i 〡X i R l )-P(Y i 〡X i R h ); Wherein, Y is the recall probability, X is the user profile feature, R is the subsidy strategy, and P(Y) is the user profile feature. i 〡X i R l ) indicates that the user profile is X i User i in R l The order placement probability under the subsidy strategy, namely P(Y) i 〡X i R h ) indicates that the user profile is X i The probability of user i placing an order under the no-subsidy strategy; Users who have lost interest are sorted based on their subsidy gain value in order to select those who will receive a recall SMS and receive corresponding coupons and subsidies.

2. The method for recalling lost ride-hailing users according to claim 1, characterized in that, The recall strategy of the acquisition platform, based on the training dataset and combined with the recall strategy, trains a first recall probability prediction model, including: The platform's recall strategy is obtained, and based on the training dataset and the recall strategy, a first recall probability prediction model is trained using the machine learning random forest algorithm.

3. The method for recalling lost ride-hailing users according to claim 1, characterized in that, The process of sorting users based on subsidy gain values ​​to select those to send recall SMS messages to and distribute corresponding coupons and subsidies includes the following sub-steps: Obtain the subsidy gain value corresponding to each churned user; Based on the subsidy gain value, churned users are sorted in descending order of subsidy gain value; Create a subsidy sorting list and add the churned users to the subsidy sorting list in sequence; Obtain the preset subsidy strategy, extract the information on the number or proportion of churned users recalled, and select churned users that meet the subsidy strategy. Based on the subsidy strategy, corresponding coupons and subsidies are issued to churned users who meet the subsidy strategy. Send recall SMS messages to each churned user who meets the subsidy policy criteria.

4. A device for recalling lost users of ride-hailing services, characterized in that, include: The data acquisition module is used to acquire historical ride-hailing order data and historical recall data; the historical recall data includes user behavior data and data on the relationship between recall strategies and recall success rates. The data analysis module is used to analyze user behavior data from historical ride-hailing order data. The user behavior data includes the user's historical ride-hailing order records on the platform, the user's click behavior records on the application or WeChat mini program, the user's historical ride-hailing time preference information, the user's historical ride-hailing location preference information, the user's historical ride-hailing price information, historical coupon usage information, and the user's historical review information. The first training module is used to train the first recall probability prediction model by combining user behavior data and historical recall data, under the condition of adopting a recall strategy. The step of training a first recall probability prediction model by combining user behavior data and historical recall data, under the condition of adopting a recall strategy, includes: Extract historical ride-hailing order records from the user behavior data corresponding to each user; Based on historical ride-hailing order records, if the last ride-hailing order is more than a preset period away from the current timestamp, the user will be marked as a churned user. From the user behavior data corresponding to churned users, we extract users' click behavior records on the application or WeChat mini program, users' historical ride-hailing time preference information, users' historical ride-hailing location preference information, users' historical ride-hailing price information, historical coupon usage information, and users' historical evaluation information, and then integrate the data to form a training dataset. The platform's recall strategy is obtained, and a first recall probability prediction model is trained based on the training dataset and the recall strategy. The second training module is used to train a second recall probability prediction model by combining user behavior data and historical recall data without employing a recall strategy. The prediction calculation module is used to calculate the subsidy gain value for each churned user based on user behavior data, through the first recall probability prediction model and the second recall probability prediction model. The calculation of the subsidy gain value is specifically as follows: P(Y) i 〡X i R l )-P(Y i 〡X i R h ); Wherein, Y is the recall probability, X is the user profile feature, R is the subsidy strategy, and P(Y) is the user profile feature. i 〡X i R l ) indicates that the user profile is X i User i in R l The order placement probability under the subsidy strategy, namely P(Y) i 〡X i R h ) indicates that the user profile is X i The probability of user i placing an order under the no-subsidy strategy; The subsidy distribution module is used to sort churned users based on subsidy gain values, select those to send recall SMS messages, and distribute corresponding coupons and subsidies.

5. The ride-hailing user recall device according to claim 4, characterized in that, Obtain the platform's recall strategy, and based on the training dataset and the recall strategy, train the first recall probability prediction model. Specifically, perform the following sub-steps: The platform's recall strategy is obtained, and based on the training dataset and the recall strategy, a first recall probability prediction model is trained using the machine learning random forest algorithm.

6. The ride-hailing user recall device according to claim 4, characterized in that, When the subsidy disbursement module is running, it specifically performs the following sub-steps: Obtain the subsidy gain value corresponding to each churned user; Based on the subsidy gain value, churned users are sorted in descending order of subsidy gain value; Create a subsidy sorting list and add the churned users to the subsidy sorting list in sequence; Obtain the preset subsidy strategy, extract the information on the number or proportion of churned users recalled, and select churned users that meet the subsidy strategy. Based on the subsidy strategy, corresponding coupons and subsidies are issued to churned users who meet the subsidy strategy. Send recall SMS messages to each churned user who meets the subsidy policy criteria.

Images