Recovery prediction system, recovery prediction method, and program

Abstract

This invention provides a return prediction system, a return prediction method, and a program that can predict specific dormant users who are likely to return to the service. [Solution] The recovery prediction system 100 includes a management device 10. The control unit 11 of the management device 10 includes various processing units such as a data acquisition unit 111, a dormancy determination unit 102, a reuse prediction unit 113, a reward determination unit 114, a timing determination unit 115, a reward granting processing unit 116, and a prediction model construction unit 150. The dormancy determination unit 102 performs a process to determine dormant users based on purchase history data 202 acquired by the data acquisition unit 111. The reuse prediction unit 113 performs a process to predict whether the dormant user will use the goods or services at the target facility again, based on the dormant user's boarding and alighting history.

Description

Technical Field

[0001] The present invention relates to a return prediction system, a return prediction method, and a program capable of predicting a specific user.

Background Art

[0002] Stores such as supermarkets and shopping centers seek to increase the number of customers by issuing coupons on websites, in advertisements and the like, or providing notifications of the holding of discount events. Further, in the aforementioned stores, management of purchasing customers is performed by registering user information of users who have purchased products or services.

[0003] By the way, rather than issuing coupons or notifying unspecified persons of discount events, it is considered that an increase in the number of customers can be expected by performing customer attraction actions such as issuing coupons and notifying so-called dormant users of discount events to attempt to bring back the aforementioned dormant users. The aforementioned dormant user refers to a customer who has used the store in the past but for whom a predetermined period (for example, several months) has elapsed since the last use, that is, a registered user who has not made a purchase at the store during the aforementioned predetermined period.

[0004] As a prior art, it is disclosed that the product interest behavior of a person is predicted by using the behavior of the person to be predicted (movement behaviors such as walking, taking a train, and riding in a car).

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of Invention

Problems to be Solved by the Invention

[0006] There are various reasons why dormant users stop using a store. For example, some of the reasons mentioned above include being unable to use the store they previously used due to moving, changing jobs, or changing schools; becoming distant from the store they previously used because a similar store opened near their place of residence or workplace; or temporarily ceasing to use the store due to changes in their living environment such as marriage, childbirth, or hospitalization.

[0007] Therefore, implementing the aforementioned customer acquisition actions for all dormant users is inefficient and ineffective. If it were possible to target specific dormant users with a high likelihood of reactivation from among the large number of dormant users and implement the aforementioned customer acquisition actions there, it would be possible to increase the number of customers at a low cost without implementing unnecessary actions. However, targeting users with a high likelihood of reactivation from among dormant users is difficult, and this has not been successfully translated into an increase in the number of customers.

[0008] The object of the present invention is to provide a return prediction system, a return prediction method, and a program that can predict specific dormant users who are likely to return to the internet. [Means for solving the problem]

[0009] A return prediction system according to one aspect of the present invention includes: a first acquisition unit that acquires usage history information including the usage history of users who have used goods or services provided at a target facility; a dormancy determination unit that determines, based on the usage history information, a dormant user whose period of non-use since the last purchase at the target facility is longer than a predetermined period; a second acquisition unit that acquires ride history information including the ride history of the dormant user who uses a vehicle operation service that operates vehicles along a route including the nearest bus stop to the target facility; and a reuse prediction unit that predicts, based on the ride history information of the dormant user, whether or not the dormant user will use goods or services at the target facility again.

[0010] Another aspect of the present invention relates to a method for predicting a return to service, in which one or more processors perform the following steps: a first acquisition step of acquiring usage history information including the usage history of a user who has used a product or service provided at a target facility; a dormancy determination step of determining, based on the usage history information, a dormant user whose period of non-use since the last purchase at the target facility is longer than a predetermined period; a second acquisition step of acquiring ride history information including the ride history of the dormant user who uses a vehicle operation service that operates vehicles along a route including the nearest bus stop to the target facility; and a reuse prediction step of predicting, based on the ride history information of the dormant user, whether or not the dormant user will use the product or service at the target facility again.

[0011] A program according to another aspect of the present invention is a program that causes one or more processors to execute: a first acquisition step of acquiring usage history information including the usage history of users who have used goods or services provided at a target facility; a dormancy determination step of determining dormant users who have not used the target facility for a predetermined period or longer since their last purchase, based on the usage history information; a second acquisition step of acquiring ride history information including the ride history of the dormant users who use a vehicle operation service that operates vehicles along a route including the nearest bus stop to the target facility; and a reuse prediction step of predicting whether the dormant users will use the goods or services at the target facility again, based on the ride history information of the dormant users.

[0012] Furthermore, the present invention can also be considered as an invention of a non-temporary, computer-readable recording medium on which the aforementioned program is stored. [Effects of the Invention]

[0013] According to the present invention, it is possible to predict dormant users who are likely to return to the service, thereby enabling the implementation of customer attraction actions such as issuing coupons and holding discount events that are effective only for specific dormant users. As a result, it is possible to increase the number of customers without incurring unnecessary costs. [Brief explanation of the drawing]

[0014] [Figure 1] Figure 1 is a block diagram showing the configuration of a recovery prediction system according to an embodiment of the present invention. [Figure 2] Figure 2 is a block diagram showing the configuration of a management device included in a recovery prediction system according to an embodiment of the present invention. [Figure 3] Figure 3 is a block diagram showing the configuration of the prediction model construction unit included in the recovery prediction system according to an embodiment of the present invention. [Figure 4] Figure 4 shows an example of customer purchase data used in a return prediction system according to an embodiment of the present invention. [Figure 5] Figure 5 shows an example of purchase history data used in the recovery prediction system according to an embodiment of the present invention. [Figure 6] Figure 6 shows an example of user registration data used in the recovery prediction system according to an embodiment of the present invention. [Figure 7] Figure 7 shows an example of boarding and alighting history data used in the return prediction system according to an embodiment of the present invention. [Figure 8] Figures 8(A) and 8(B) are schematic diagrams showing an example of a movement path of a person to be predicted in a return prediction system according to an embodiment of the present invention. [Figure 9] Figure 9 is a flowchart showing an example of the procedure for the recovery prediction process performed by the recovery prediction system according to an embodiment of the present invention. [Figure 10] Figure 10 is a flowchart showing a modified example of the procedure for the recovery prediction process performed in the recovery prediction system according to an embodiment of the present invention. [Modes for carrying out the invention]

[0015] The embodiments of the present invention will be described below with reference to the attached drawings. Note that the following embodiments are merely examples of the present invention and do not limit the technical scope of the present invention.

[0016] [Return prediction system 100] Figure 1 is a network diagram showing the configuration of a return prediction system 100 (an example of the return prediction system of the present invention) according to an embodiment of the present invention. The return prediction system 100 is a system that performs return prediction processing (see FIG. 9 and FIG. 10) for predicting users who have a high possibility of purchasing a service at a facility, among so-called dormant users registered in the facility that provides the service.

[0017] Here, the dormant user refers to a user who has used a facility such as a store in the past, and among the registered users registered in the customer management data of the facility, is a user for whom a predetermined period (for example, several months) has elapsed since the last purchase at the facility. In other words, the dormant user is a registered user whose non-purchase period (non-use period) during which no product or service has been purchased at the target facility is equal to or longer than the predetermined period. The non-purchase period is a period calculated from the last purchase time or last purchase date when the dormant user last made a purchase at the target facility, and the last purchase time or the last purchase date is the date and time when the dormant user last purchased a product or service at the target facility. In addition, the predetermined period is an arbitrarily set period.

[0018] Hereinafter, in the present embodiment, that the dormant state is canceled when the dormant user purchases a product or service provided at the target facility during the dormancy period after the predetermined period has elapsed from the last purchase time is referred to as return from the dormant state, or simply return.

[0019] Examples of such facilities include retail stores that display and sell goods, such as supermarkets, apparel shops, general merchandise stores, home centers, and pet shops; service businesses such as hotels, restaurants, hospitals, massage parlors, beauty salons, gyms, and travel agencies; and commercial facilities such as department stores and shopping malls that house multiple tenant shops. If the facility is an individual store, each store manages customer information as described below. On the other hand, if the facility is a commercial facility that includes multiple tenant shops, the operator of the commercial facility manages not only the customer information of users who use the commercial facility, but also the customer information of users who use any of the tenant shops.

[0020] As shown in Figure 1, the recovery prediction system 100 includes a management device 10 and a database 30.

[0021] In the recovery prediction system 100, the management device 10 is connected to the database 30 via a network N1, which is either a wireless communication network or a wired communication network. Network N1 is, for example, a dedicated line or a public line.

[0022] The management device 10 is, for example, an information processing device managed by the operator of the facility or its related businesses, and is, for example, a server device or a cloud server. The management device 10 is installed, for example, in the central control room of the operator. The management device 10 is not limited to being composed of a single computer, but may be a computer system in which multiple computers work together, or a cloud computing system. Furthermore, the various processes performed by the management device 10 may be executed by a single processor, or they may be distributed and executed by multiple processors. Computer software for operating the recovery prediction system 100 of this embodiment is installed on the management device 10.

[0023] Furthermore, the recovery prediction system 100 cooperates with various systems, such as the existing customer management system 200 which manages customer information, and the existing boarding and alighting management system 300 which manages the entry, exit, and boarding and alighting of passengers using railways and buses, to obtain the information necessary for the recovery prediction process (see Figures 9 and 10) from each system.

[0024] In the following description, the return prediction system 100 will be described as being independent of other existing systems such as the customer management system 200 and the boarding / alighting management system 300. However, the return prediction system 100 may be incorporated into existing systems such as the customer management system 200 and the boarding / alighting management system 300. Alternatively, the return prediction system 100 may include the customer management system 200 and the boarding / alighting management system 300.

[0025] As shown in Figure 1, the management device 10 is further connected via network N2 to communicate with user terminals 40, customer management system 200, boarding / alighting management system 300, and the like.

[0026] The user terminal 40 is an information processing device or terminal device used by passengers using public transportation services (vehicle operation services) provided by public transportation operators such as railway operators and bus operators. The aforementioned public transportation operators are vehicle operation operators that provide public transportation services using vehicles such as trains and buses to transport passengers to their destination stops (stations, bus stops, etc.).

[0027] The user terminal 40 is, for example, a portable device such as a smartphone or tablet that can be carried by the passenger. The user terminal 40 has an application (software) installed that works in conjunction with the recovery prediction system 100 to display information output from the management device 10.

[0028] The customer management system 200 is a conventionally known system that manages customer data 201, which includes customer information about purchasers who have purchased goods or services at the facility, and purchase history data 202, which includes the purchase history of goods or services purchased by the purchasers. The customer management system 200 assigns a customer ID to each customer who has purchased goods or services, and manages the customer information and purchase history for each customer ID.

[0029] The purchase customer data 201 and purchase history data 202 are stored in the storage device provided by the customer management system 200. The purchase customer data 201 and purchase history data 202 are managed separately for each facility. Alternatively, the purchase customer data 201 and purchase history data 202 may also be stored in the database 30 described later.

[0030] When the customer management system 200 receives a transmission request from the management device 10, it transmits the purchase history data 202, including the purchase history, to the management device 10. It also transmits the customer purchase data 201, including the customer information, to the management device 10 in response to the transmission request. Alternatively, when the customer management system 200 receives an access request from the management device 10, it grants access to the customer purchase data 201 and the purchase history data 202. This allows the management device 10 to retrieve the customer information and the purchase history.

[0031] For example, if the facility is a single store, the purchase customer data 201 and purchase history data 202 include information about customers who have used the store (the customer information and the purchase history). Also, if the facility is a commercial facility such as a department store or shopping mall that houses multiple tenant shops, the purchase customer data 201 and purchase history data 202 include information about customers who have used one or more of the multiple tenant shops belonging to the commercial facility (the customer information and the purchase history).

[0032] In the following, examples of customer data 201 and purchase history data 202, which include customer information and purchase history of customers who have used the commercial facility, are provided.

[0033] Figure 4 shows an example of customer purchase data 201 used in the return prediction system 100. As shown in Figure 4, in addition to the customer ID, the customer purchase data 201 includes information such as the customer name, customer attribute information (age, gender, place of residence, preferred genre, occupation, etc.), registration date, membership rank, and contact information such as email address. The last purchase date and time for each customer may also be registered in the customer purchase data 201.

[0034] Figure 5 shows an example of purchase history data 202 used in the return prediction system 100. As shown in Figure 5, in addition to the customer ID, the purchase history data 202 registers information such as the store where the purchase was made, the items purchased, the date of purchase, the time of purchase, the purchase amount, and the number of points added at the time of purchase. By referring to the information in the purchase history data 202, the management device 10 can obtain the name of the store where a particular customer last made a purchase at the commercial facility, the date and time of the last purchase, etc.

[0035] The passenger boarding and alighting management system 300 is a conventionally known system that acquires and manages boarding and alighting information such as the passenger's user ID, entry station, entry time (time of entry through ticket gate), exit station, exit time (time of exit through ticket gate), and date of travel (date of use) by reading information from contactless IC transit cards using automatic ticket gates installed at the ticket gates of railway stations. In addition, the passenger boarding and alighting management system 300 acquires and manages boarding and alighting information such as the passenger's user ID, the bus stop where the passenger boarded, the time of boarding, the bus stop where the passenger alighted, the time of alighting, and the date of travel (date of use) by reading information from contactless IC transit cards using IC readers installed inside buses.

[0036] The boarding and alighting management system 300 manages the boarding and alighting information transferred from the station servers installed at each railway station, and the boarding and alighting information transferred from the IC reader installed on the bus, for each user ID to which the same number as the customer ID is assigned.

[0037] The database 30 is a storage device such as an HDD or SSD. The database 30 is configured as an external device such as another server device or other storage device that can communicate data with the management device 10 via the network N1. The database 30 may be a storage device installed in the management device 10, or it may be a storage device connected to the management device 10 by a local network.

[0038] As shown in Figure 2, the database 30 stores user registration data 31, boarding and alighting history data 32, route data 33, and operation schedule data 34. The database 30 is allocated multiple storage areas to store each type of data, and each type of data is stored in its respective storage area. Note that each type of data 31-34 may also be stored separately in multiple storage devices that are connected to the management device 10 in a communicative manner.

[0039] The user registration data 31 includes various types of registration information (user registration information) concerning multiple users of the railway. The user registration information included in the user registration data 31 is, for example, information registered when purchasing an IC transit card usable on railway lines or buses.

[0040] Furthermore, the IC transit card stores at least the following information: the user ID of the user of the IC transit card, the type of IC transit card, the expiration date of the IC transit card, and, if the IC transit card has a commuter pass function (hereinafter abbreviated as "commuter pass"), the regular usage section.

[0041] Figure 6 shows an example of user registration data 31 used in the recovery prediction system 100. As shown in Figure 6, in addition to the user ID, the user registration data 31 includes information such as the user's name, user attributes (age, gender, student, etc.), type of IC transit card, regular travel section, and contact information such as an email address. There are three types of IC transit cards, for example: commuter passes (commuter pass type), student passes (student pass type), and normal types that do not have the functionality of a commuter pass.

[0042] The boarding and alighting history data 32 includes the past boarding and alighting history of users who have used railway or bus lines (an example of the boarding history according to the present invention). The boarding and alighting history data 32 is sometimes referred to as OD (Origin to Destination) data.

[0043] Figure 7 shows an example of boarding and alighting history data 32 used in the return prediction system 100. For example, the boarding and alighting history includes, in addition to the user ID, an entry record when entering a railway station and an exit record when leaving a railway station. The boarding and alighting history also includes a boarding record (corresponding to an entry record) when boarding a bus at a bus stop and an alighting record (corresponding to an exit record) when alighting from a bus at a bus stop.

[0044] In this embodiment, the boarding and alighting history when using a train includes the user ID, the entry station (departure station) where the user entered when traveling by train, the entry date (usage date, boarding date), the entry time, the exit station which is the nearest station to the destination, the exit date, the exit time, and the type of IC transit card (usage type). Similarly, the boarding and alighting history when using a bus includes the user ID, the bus stop (boarding stop) where the user boarded when traveling by bus, the boarding date (usage date), the boarding time, the nearest bus stop to the destination, the alighting date, the alighting time, and the type of IC transit card (usage type).

[0045] The user ID of a passenger when boarding or alighting is read from the passenger's IC transit card by the automatic ticket gate installed at the station when the passenger passes through the gate. The user ID is also read from the passenger's IC transit card by an IC reader installed inside the bus when boarding or alighting from the bus. In addition, boarding and alighting information other than the user ID is acquired by the automatic ticket gate, the IC reader, or a higher-level system that manages them when the passenger boards or alights. The boarding and alighting information acquired by the automatic ticket gate, the IC reader, etc., is transferred to the boarding and alighting management system 300 and registered in the boarding and alighting history data 32 for each user ID. In other words, the boarding and alighting management system 300 aggregates the transferred boarding and alighting information into the boarding and alighting history data 32 so that it can be managed for each user ID and stores it in the database 30.

[0046] The passenger boarding and alighting management system 300 is a conventionally known system operated and managed by a railway operator or bus operator, but it may also be, for example, an interoperability system managed and operated by IC Card Interoperability Center Co., Ltd. in Japan, or the National Integrated Transport Analysis System (NITAS) managed and operated by the Ministry of Land, Infrastructure, Transport and Tourism in Japan.

[0047] [Management device 10] The specific configuration of the control device 10 will be described below with reference to Figure 2.

[0048] The management device 10 is for realizing the recovery prediction system 100 of this embodiment, and uses the prediction model 122 described later to perform recovery prediction processing to predict which of the dormant users are most likely to purchase goods or services at the target facility. Hereinafter, dormant users who are most likely to purchase goods or services at the target facility may be referred to as "prospective recovery users."

[0049] As shown in Figure 2, the management device 10 comprises a control unit 11, a storage unit 12, a communication unit 13, a display unit 14, and an operation unit 15.

[0050] The communication unit 13 is a communication interface that connects the management device 10 to networks N1 and N2 and performs data communication with each device connected to networks N1 and N2 according to a predetermined communication protocol. Specifically, the communication unit 13 performs data communication with the database 30 via network N1. The communication unit 13 also performs data communication with the user terminal 40, the customer management system 200, and the boarding / alighting management system 300 via network N2.

[0051] The memory unit 12 is a non-volatile storage medium such as an HDD or SSD that stores various types of information. The memory unit 12 stores data, thresholds, reference values, etc., used for various processes by the control unit 11. The memory unit 12 also stores control programs for executing various processes by the control unit 11, and the prediction model 122. In addition, the training dataset 123 generated by the training dataset generation unit 151 (see Figure 3), which will be described later, is stored in the memory unit 12. The prediction model 122 may be implemented as an electronic circuit equipped with a memory in which the prediction model 122 is stored. The training dataset 123 may also be stored in the database 30.

[0052] The control program 121 may be non-temporarily recorded on a computer-readable recording medium such as a CD or DVD, and may be read from the recording medium by a reading device (not shown), such as a CD drive or DVD drive, electrically connected to the management device 10, and copied to the storage unit 12. Alternatively, the control program 121 may be read from external storage connected to the network N1, input via the communication unit 13, and copied to the storage unit 12.

[0053] The control program 121 is a program for executing a recovery prediction process using the prediction model 122, or for executing a prediction model construction process by the prediction model construction unit 150 described later. Alternatively, instead of the control program 121, programs for causing the control unit 11 to execute the recovery prediction process and programs for causing the control unit 11 to execute the prediction model construction process may be stored separately in the storage unit 12.

[0054] The prediction model 122 is a trained model used in the return prediction process to predict the users likely to return, and it derives the reuse probability (an example of a reuse index) of the target facility being used again. The prediction model 122 is trained by machine learning using the training dataset 123.

[0055] The prediction model 122 is constructed by the prediction model construction unit 150, which will be described later. In this embodiment, the prediction model 122 is constructed by machine learning by the control unit 11 based on the training dataset 123 (see Figure 2) generated by the training dataset generation unit 151 (see Figure 3) and a predetermined algorithm. Furthermore, when the training dataset 123 is updated, the control unit 11 retrains the prediction model 122 to update the prediction model 122.

[0056] Suitable algorithms for constructing the prediction model 122 include, for example, logistic regression, decision trees, random forests, and gradient boosting. Alternatively, the prediction model 122 may be constructed using deep learning with multiple hidden layers in the neural network. The prediction model 122 may utilize one of the above-mentioned algorithms, or it may utilize multiple algorithms.

[0057] The display unit 14 is a display device such as a liquid crystal display or an organic EL display that displays various types of information. The operation unit 15 is an input device such as a mouse, keyboard, or touch panel that accepts input from the operator.

[0058] Incidentally, there are various reasons why the aforementioned dormant users stopped using stores and other facilities. For example, some possible reasons include being unable to use the store they previously used due to moving, changing jobs, or changing schools; becoming estranged from the store they previously used because a similar store opened near their place of residence or workplace; or temporarily ceasing to use the store due to changes in their living environment such as marriage, childbirth, or hospitalization.

[0059] Rather than issuing coupons or announcing discount events to an unspecified number of people, it is considered more likely to increase the number of customers by implementing customer attraction actions such as issuing coupons or announcing discount events to so-called dormant users in an attempt to reactivate them. However, implementing such customer attraction actions to all of the dormant users is inefficient and ineffective. If it is possible to target specific potential returnee users (an example of a specific dormant user in this invention) from among a large number of dormant users and implement such customer attraction actions, it is possible to increase the number of customers at low cost without implementing unnecessary actions. However, it is difficult to target such potential returnee users from among a large number of dormant users, and has not been able to successfully increase the number of customers.

[0060] In contrast, in the recovery prediction system 100 of this embodiment, the recovery prediction process described later (see Figures 9 and 10) is executed by the control unit 11, which is configured as described later. As a result, it becomes possible to predict with high accuracy specific users who are likely to recover. This makes it possible to implement customer attraction actions such as issuing coupons or holding discount events that are effective only for specific users who are likely to recover, and as a result, it becomes possible to increase the number of customers without incurring unnecessary costs.

[0061] The control unit 11 controls the operation of each part of the management device 10. The control unit 11 includes control devices such as a CPU, ROM, and RAM. The CPU is a processor that performs various arithmetic operations. The ROM is a non-volatile storage medium in which control programs such as a BIOS and OS are pre-stored to cause the CPU to perform various arithmetic operations. The RAM is a volatile or non-volatile storage medium that stores various information and is used as a temporary storage memory (work area) for the various arithmetic operations performed by the CPU. The control unit 11 controls the management device 10 and the recovery prediction system 100, etc., by having the CPU execute various control programs pre-stored in the ROM or storage unit 12.

[0062] As shown in Figure 2, the control unit 11 includes various processing units such as a data acquisition unit 111, a dormancy determination unit 112, a reuse prediction unit 113, a reward determination unit 114, a timing determination unit 115, a reward granting processing unit 116, and a prediction model construction unit 150. The data acquisition unit 111 is an example of the first and second acquisition units of the present invention. The dormancy determination unit 112 is an example of the dormancy determination unit of the present invention. The reuse prediction unit 113 is an example of the reuse prediction unit of the present invention. The timing determination unit 115 is an example of the timing determination unit of the present invention. The reward granting processing unit 116 is an example of the reward granting processing unit of the present invention.

[0063] The control unit 11 functions as one of the various processing units by having the CPU execute various arithmetic processes according to the control program 121. The control unit 11 or the CPU is an example of a computer that executes the control program. Some or all of the processing units included in the control unit 11 may be composed of electronic circuits. The control program may also be a program that causes multiple processors to function as the various processing units.

[0064] The data acquisition unit 111 performs the process of acquiring purchase history data 202 (an example of usage history information in the present invention), which includes the purchase history (an example of usage history in the present invention) of users who have used goods or services provided at the target facility (target facility). Specifically, the data acquisition unit 111 sends a transmission request to the customer management system 200. When the customer management system 200 receives the transmission request from the data acquisition unit 111, it transmits the purchase history data 202, including the purchase history, to the management device 10. In addition, in response to the transmission request, it transmits the purchased customer data 201, including the customer information, to the management device 10. Alternatively, the data acquisition unit 111 may send an access request to the customer management system 200. In this case, when the customer management system 200 receives the access request from the data acquisition unit 111, it grants access to the purchased customer data 201 and the purchase history data 202. This enables the data acquisition unit 111 to acquire the customer information and the purchase history.

[0065] The data acquisition unit 111 may include the facility name of the target facility (the name of the store where the purchase was made) in the transmission request or the access request when sending it. In this case, the customer management system 200 sends the purchase history of the user who used the store where the purchase was made that matches the facility name included in the transmission request or the access request to the management device 10, or grants access to the purchase history.

[0066] The dormancy determination unit 112 performs a process (dormancy determination process) to determine which users are dormant if the period of non-purchase (non-use period) since the last purchase at the target facility is longer than a predetermined period, based on the purchase history data 202 acquired by the data acquisition unit 111. In other words, the dormancy determination unit 112 extracts customers from among the multiple registered customers registered in the purchase history data 202 whose period of non-purchase is longer than a predetermined period as dormant users.

[0067] Furthermore, the data acquisition unit 111 performs the process of acquiring boarding and alighting history data 32 (an example of boarding history information in the present invention) which includes the boarding and alighting history of the dormant user who uses the public transportation service that operates trains and buses along a route that includes the nearest stop (train station or bus stop) to the target facility. Specifically, the data acquisition unit 111 accesses the database 30, refers to the boarding and alighting history data 32 in the database 30, and reads the boarding and alighting history corresponding to the user ID with the same number as the customer ID that the dormant user determination unit 112 has determined to be a dormant user.

[0068] The reuse prediction unit 113 performs a process to predict whether the dormant user will use the goods or services at the target facility again, based on the dormant user's boarding and alighting history and the dormant user's purchase history acquired by the data acquisition unit 111.

[0069] Specifically, the reuse prediction unit 113 determines the probability of return (an example of the reuse index of the present invention) based on the boarding and alighting history and the purchase history, and predicts whether the dormant user will use the goods or services at the target facility again based on the probability of return.

[0070] Here, the return probability is an index used in the process of determining whether a dormant user (a user likely to return) is likely to purchase goods or services at the target facility. If this return probability is equal to or greater than a predetermined threshold, the dormant user is determined to be a user likely to return and use the goods or services at the target facility again.

[0071] In this embodiment, a threshold T1 and a threshold T2 that is greater than the threshold T1 are used as the setting thresholds. If the recovery probability is greater than or equal to the threshold T1, it is determined that there is a high possibility of the dormant user recovering. If the recovery probability is greater than or equal to the threshold T2, it is determined that there is an even higher possibility of the dormant user recovering. Note that the setting thresholds are not limited to the thresholds T1 and T2, and more thresholds may be used to determine the likelihood of a user recovering. Of course, there may be only one setting threshold.

[0072] In this embodiment, the reuse prediction unit 113 predicts the return probability using a trained prediction model 122. Specifically, the reuse prediction unit 113 inputs the boarding and alighting history and purchase history of the dormant user, acquired by the data acquisition unit 111, as explanatory variables into the input unit of the prediction model 122, causing the prediction model 122 to predict the return probability, and outputs the predicted return probability as the prediction result from the output unit of the prediction model 122.

[0073] In this embodiment, an example is described in which the reuse prediction unit 113 determines the return probability based on the boarding and alighting history and the purchase history. However, the reuse prediction unit 113 may determine the return probability based on at least the boarding and alighting history. In other words, the reuse prediction unit 113 may input the boarding and alighting history of the dormant user as an explanatory variable into the input of the trained prediction model 122, and have the prediction model 122 predict the return probability.

[0074] The reward determination unit 114 determines and issues the reward to be granted by the reward granting processing unit 116. The reward may be, for example, a discount coupon or a voucher for a novelty item or gift certificate given to the user at the time of purchase. The voucher will be applicable when a product or service of a predetermined amount or more is purchased at the target store.

[0075] The reward determination unit 114 determines the reward according to the numerical value of the recovery probability. For example, if the reward is a discount coupon, it determines the discount rate according to the recovery probability and determines a discount coupon with that discount rate as the reward. Specifically, if the recovery probability is between threshold T1 and threshold T2, the reward determination unit 114 determines a 20% discount coupon as the reward. If the recovery probability is between threshold T2 and threshold T2, the reward determination unit 114 determines a 40% discount coupon as the reward.

[0076] Furthermore, for example, if the reward is a voucher for a novelty item, the reward determination unit 114 determines a novelty item of a value corresponding to the probability of recovery and determines a voucher for that novelty item as the reward. Specifically, if the probability of recovery is greater than or equal to the threshold T1 and less than the threshold T2, the reward determination unit 114 determines a 1,000 yen gift certificate or a voucher for a novelty item worth 1,000 yen as the reward. Also, if the probability of recovery is greater than or equal to the threshold T2, the reward determination unit 114 determines a 2,000 yen gift certificate or a voucher for a novelty item worth 2,000 yen as the reward.

[0077] Furthermore, it is preferable that the reward determination unit 114 determines a reward applicable to a product or service of the same type as the product or service that the prospective user has purchased (used) most frequently, based on the prospective user's purchase history. Alternatively, it is preferable that the reward determination unit 114 determines a reward applicable to a product or service of the same type as the product or service that the prospective user has purchased (used) the largest amount, based on the prospective user's purchase history. In other words, the reward granted to the prospective user is a reward applicable to a product or service of the same type as the product or service that has the highest usage frequency or usage amount included in the usage history. It is considered that if the reward is applicable to such a product or service, the prospective user is more likely to visit the target facility and use the reward to purchase the product or service.

[0078] The timing determination unit 115 performs a process to determine the timing for granting the benefits to the prospective returning user who is eligible for the benefits granted by the benefit granting processing unit 116. Specifically, the timing determination unit 115 determines the timing at which the prospective returning user will use a specific section within a predetermined range from the nearest stop (train station or bus stop) to the target facility, and determines that timing as the granting timing.

[0079] Figure 8(A) shows the travel route of a user (a user expected to return to service) who uses the railway, where the nearest station to the user's home is Station A1, the usage section of the user's commuter pass (commuter usage section) is from Station A1 to Station A14, and the nearest station to the target facility is Station A11 within the aforementioned commuter usage section. When the user travels along the commuter usage section shown in Figure 8(A), the timing determination unit 115 determines the timing (usage timing) to be the moment when the train the user is on reaches a specific section within a predetermined range in the opposite direction of the train's movement from Station A11, as the user boards the train from Station A1 to Station A14. Whether or not the train has reached the specific section can be determined, for example, based on GPS location information obtained from the user terminal 40. Furthermore, it is possible to determine whether or not the passenger's train has reached the specified section based on information such as the entry station, entry date and time, and day of the week included in the boarding and alighting history, information such as the date and time and day of the day of the day being determined, and route information and operation schedule information included in the railway train route data 33 and operation schedule data 34.

[0080] In this case, the specified section can be, for example, the section from station A11, which is the nearest station to the target facility, to a predetermined number of stations counting in the reverse direction.

[0081] Figure 8(B) shows the travel route of a user (a user expected to return to service) who uses the railway, where the nearest station to the user's home is Station A1, the usage section of the user's commuter pass (commuter usage section) is from Station A1 to Station A9, and the nearest station to the target facility is Station A11, which is outside the aforementioned commuter usage section. When the user travels along the commuter usage section shown in Figure 8(B), the timing determination unit 115 determines the timing to be the timing at which the train the user is on reaches a specific section within a predetermined range in the opposite direction of the train's movement from Station A11, when the user boards a train from Station A1 to Station A9 (usage timing).

[0082] In this case, the specified section may be, for example, the section from station A11, which is the nearest station to the target facility, to station A9, which is the last station in the regular usage section, as well as the section from station A9 to a predetermined number of stations counting in the reverse direction.

[0083] Whether or not the aforementioned train has reached the specified section can be determined, for example, based on GPS location information obtained from the user terminal 40. Alternatively, it can be determined whether or not the aforementioned train has reached the specified section from information such as the entry station, entry date and time, and day of the week included in the boarding and alighting history, information such as the date and time and day of the week for the determination, and route information and operation schedule information included in the railway train route data 33 and operation schedule data 34.

[0084] The predetermined number can be any number of stations predetermined in advance. For example, if the predetermined number is "2", in the example shown in Figure 8(A), the specific section is the section from station A11 to station A9, which is the second station in the reverse direction, and in this case, the timing determination unit 115 determines the timing at which the passenger train arrives at station A9 as the assigned timing. Also, in the example shown in Figure 8(B), the specific section is the section from station A11 to station A9, as well as the section from station A9 to station A7, which is the second station in the reverse direction, and in this case, the timing determination unit 115 determines the timing at which the passenger train arrives at station A7 as the assigned timing.

[0085] Thus, by granting the aforementioned benefits while on board, the likelihood of users confirming the benefits increases. Furthermore, by notifying users of the benefits before arriving at the station nearest the target facility, it is considered that users are more likely to confirm the details of the benefits they have been notified about and then decide to get off at the station nearest the target facility to go to the facility.

[0086] Furthermore, the timing determination unit 115 may determine the timing at which the prospective user who is eligible for the reward uses the public transportation service on a route that includes the nearest stop (train station or bus stop) to the target facility, and based on that timing, determine the timing at which the reward granting processing unit 116 will grant the reward.

[0087] Specifically, the timing determination unit 115 determines the granting timing within the range from when the expected user returns to using the public transportation service until a predetermined time has elapsed.

[0088] For example, in the travel route shown in Figure 8(A), if a user (a user expected to return) travels from their nearest station A1 to station A14 on a regular travel route, the timing determination unit 115 determines the timing of the benefit grant to be immediately after the user enters station A1. In this case, the set value of the predetermined time is set to, for example, within the range of 0 seconds or more and less than 5 seconds. In this case, the benefit will be notified at the moment the user enters station A1.

[0089] Furthermore, the timing determination unit 115 may determine the timing of the award as the time elapsed from the time the user enters station A1 until they arrive at the platform of station A1. The elapsed time is set to any time depending on the station. In this case, the user will be notified of the benefit around the time they arrive at the platform of station A1, and the user will be able to check the details of the benefit while waiting for the train.

[0090] Alternatively, the timing determination unit may set the timing of granting the benefit to the moment immediately after entering station A14. In this case, the benefit will be notified to the user at the moment they enter station A14.

[0091] Alternatively, the timing determination unit 115 may determine the timing to be the time elapsed from the time the user enters station A14 until they arrive at the platform of station A14.

[0092] Furthermore, for example, in the travel route shown in Figure 8(B), if a user (a user expected to return) travels from their nearest station A1 to station A9 on a regular travel route, the timing determination unit 115 determines the timing to be the moment immediately after the user enters station A1. Alternatively, the timing determination unit 115 may determine the timing to be the moment after the required time has elapsed from the time the user enters station A1 until they arrive at the platform of station A1.

[0093] Furthermore, the timing determination unit 115 may determine the timing of the reward grant to be the moment immediately after the user enters station A9 in order to return home, as shown in the travel route in Figure 8(B). In this case, the reward will be notified at the moment the user enters station A9. This increases the likelihood that the user, upon seeing the reward immediately after entering, will proceed to the platform going in the opposite direction from their intended destination, move to station A11, the nearest station to the target facility, and then proceed to the target facility.

[0094] Furthermore, whether a user has entered or exited a station, and the timing of entry and exit, can be determined based on the boarding and alighting information obtained from the station's automatic ticket gates. Of course, this can also be determined using GPS location information obtained from the user terminal 40, station location information, and the user's boarding and alighting history.

[0095] The reward granting processing unit 116 grants the reward to the prospective returning user who has been determined by the reuse prediction unit 113 to be likely to use the service again. Specifically, the reward granting processing unit 116 notifies the user terminal 40 used by the prospective returning user of the reward information. The notification method may include, for example, sending the information to an email address, or using the notification function of an application that operates on the user terminal 40 in cooperation with the return prediction system 100 to notify the user of the reward information.

[0096] In this embodiment, when the reward granting timing arrives, the reward information is notified to the user terminal 40 used by the prospective user.

[0097] The prediction model construction unit 150 generates a training dataset 123 (see Figure 2), and then constructs (generates) a trained prediction model 122 by training the prediction model 122 using machine learning based on the generated training dataset 123 and a predetermined algorithm. Specifically, the training dataset 123 is input to the prediction model 122, and the objective function corresponding to the algorithm is trained.

[0098] Figure 3 is a block diagram showing the configuration of the prediction model construction unit 150. As shown in Figure 3, the prediction model construction unit 150 includes various processing units such as a training dataset generation unit 151 and an evaluation and judgment unit 152.

[0099] The training dataset generation unit 151 performs the process of generating a training dataset 123 for training the prediction model 122. Specifically, the training dataset generation unit 151 collects the purchase history of a large number of inactive users who were previously inactive but have since resumed their inactive state, the boarding and alighting history of the inactive users, and the user registration information of the inactive users from the purchase customer data 201, purchase history data 202, user registration data 31, boarding and alighting history data 32, etc., and generates a training dataset 123 that includes this information.

[0100] The training dataset 123 generated by the training dataset generation unit 151 is used as training data for machine learning the prediction model 122. That is, when the prediction model construction unit 150 needs to machine learn the prediction model 122, it inputs the training dataset 123 into the prediction model 122 to perform machine learning.

[0101] The information input to the predictive model 122 for machine learning may include, at a minimum, the purchase history of the returning inactive user and the boarding and alighting history of the returning inactive user. Of course, other information that is effective for machine learning may also be input. For example, the average number of purchases per month before the returning inactive user became inactive, the average purchase amount, the genre of purchased items, whether or not the user used public transportation services while inactive, the number of such trips, the number of times the user used the nearest station to the target facility, whether or not the user possessed a commuter pass, whether or not the user used public transportation services before becoming inactive, and the number of such trips may also be input.

[0102] For example, if the user frequently uses the nearest station to the facility during the dormant period, or frequently uses public transportation on the line including the said nearest station, the likelihood of reactivating the dormant state is high. Conversely, if the user has a low frequency of using either of these services, the likelihood of reactivation tends to be low.

[0103] Furthermore, the longer the time elapsed from alighting at the nearest station to the facility until re-entering, that is, the longer the user stays around the station, the higher the likelihood of the facility waking up from its dormant state. Conversely, if the elapsed time is short, the likelihood of waking up tends to be lower.

[0104] Furthermore, if the user frequently uses stations close to the target facility on the aforementioned public transportation routes (for example, stations within a 10-minute travel time), the likelihood of reactivating the system from its dormant state is higher, while a low frequency of use tends to decrease the likelihood of reactivation.

[0105] Furthermore, users who use many stations tend to be more likely to reactivate their accounts than users who use fewer stations. The latter are more proactive and have a wider range of activity than the former who always use the same station, so it is thought that they are more likely to move to the station nearest the target facility as a result of receiving notification of the aforementioned benefits.

[0106] Furthermore, users with commuter passes are more likely to reactivate their accounts from inactivity than users without them. This is because, for the former, a fare is required to travel to the nearest station to the facility in question, while for the latter, the fare is either not required or only a small amount.

[0107] Furthermore, regardless of whether they own a commuter pass or not, users whose frequently traveled routes or regularly used routes include the nearest station to the target facility tend to be more likely to reactivate their account from a dormant state compared to users whose routes or regularly used routes do not include the nearest station.

[0108] Furthermore, users who used the aforementioned public transportation before their account was inactive but whose frequency of use of such transportation decreased significantly during the inactive period tend to be less likely to return to their inactive state.

[0109] In the machine learning of the predictive model 122 using the training dataset 123, these various trends are taken into consideration when performing the machine learning of the predictive model 122.

[0110] The evaluation and determination unit 152 receives information such as the user ID of the prospective user whose return probability has been determined by the reuse prediction unit 113 to be equal to or greater than the set threshold, and refers to the purchase history of the prospective user to determine whether the prospective user has purchased goods or services at the target facility within the validity period of the benefit. If a purchase by the prospective user is determined, the evaluation and determination unit 152 recognizes the prospective user as a dormant returning user, collects the prospective user's purchase history, boarding and alighting history, and user registration information, and newly registers this information in the training dataset 123. For example, if the number of new registrations exceeds a predetermined set value, it sends an update request to the training dataset generation unit 151. Upon receiving this update request, the training dataset generation unit 151 retrains the prediction model 122 with the training dataset 123 and updates the prediction model 122.

[0111] [Recovery prediction processing] The recovery prediction method of this disclosure will be described below with reference to the flowchart in Figure 9, along with an example of the procedure for the recovery prediction process performed in the recovery prediction system 100. Note that one or more steps included in the recovery prediction process described below may be omitted as appropriate, and the execution order of each step may differ to the extent that similar effects are produced. Furthermore, in the following description, the execution entity for each step will be described as the control unit 11, but the execution entity for each step may be one or more processors.

[0112] <Step S11> First, in step S11, the control unit 11 acquires the user's purchase history. This process is performed by the data acquisition unit 111. Step S11 is an example of the first acquisition step of the recovery prediction method of the present invention.

[0113] <Step S12> In the next step S12, the control unit 11 performs a process (dormancy determination process) to determine whether a user is a dormant user whose period of non-use since the last purchase at the target facility is longer than a predetermined period, based on the purchase history data 202 acquired in step S11. This process is performed by the dormancy determination unit 112. Step S12 is an example of the dormancy determination step of the recovery prediction method of the present invention.

[0114] <Step S13> In the next step S13, the control unit 11 acquires the boarding and alighting history of the dormant user determined in step S12. This process is performed by the data acquisition unit 111. Step S13 is an example of the second acquisition step of the recovery prediction method of the present invention.

[0115] <Step S14> In the next step S14, the control unit 11 performs a process to determine the recovery probability. Specifically, the control unit 11 predicts the recovery probability using the prediction model 122. This process is performed by the reuse prediction unit 113. Step S14 is an example of the reuse prediction step of the recovery prediction method of the present invention.

[0116] <Step S15> In the next step S15, the control unit 11 determines whether the recovery probability is greater than or equal to the threshold T1. If it is determined that the recovery probability is greater than or equal to the threshold T1, the process proceeds to step S16. On the other hand, if it is determined that the recovery probability is less than the threshold T1, the recovery prediction process ends without the issuance of any benefits or notification.

[0117] <Step S16> In the next step S16, the control unit 11 determines whether the recovery probability is greater than or equal to the threshold T2. If it is determined that the recovery probability is greater than or equal to the threshold T2, the process proceeds to step S17. On the other hand, if it is determined that the recovery probability is less than the threshold T2, the process proceeds to step S18.

[0118] <Step S17> In step S17, the control unit 11 determines and issues a benefit with a high discount rate to the user who is expected to recover, for whom the recovery probability is determined to be equal to or greater than the threshold T2.

[0119] <Step S18> In step S18, the control unit 11 determines and issues a benefit with a lower discount rate to the prospective user whose recovery probability is determined to be greater than or equal to the threshold T1 and less than the threshold T2.

[0120] <Step S19> Once the processing in steps S17 and S18 is complete, in the next step S19, the control unit 11 determines the notification timing. This processing is performed by the timing determination unit 115.

[0121] <Step S20> Subsequently, in step S20, a determination is made as to whether or not the notification timing has arrived.

[0122] <Step S21> Then, when it is determined in step S20 that the notification timing has arrived, the control unit 11 notifies the user terminal 40 used by the prospective user of the benefit information at that time. This process is carried out by the benefit granting processing unit 116.

[0123] [Variations of recovery prediction processing] As shown in Figure 10, after the processing in step S13 is completed, the control unit 11 may, for example, determine whether or not the dormant user used public transportation (rail or bus) during the dormant period based on the user's boarding and alighting history (S131). If it is determined that the user used public transportation, the control unit 11 may determine whether or not the frequency of use (number of uses) during the dormant period is equal to or greater than a predetermined threshold T3 (S132). Here, the frequency of use is the number of uses during the dormant period, or the average number of uses during the dormant period. In this case, if no positive determination is made in either step S131 or S132, the series of recovery prediction processes will end without issuing the reward or sending a notification. If there is no use of public transportation, or if there is use but the frequency of use is low and less than the threshold T3, it is considered unlikely that the user will take a train or bus to travel to the target facility. In this case, issuing the reward and sending a notification would likely be pointless, so the series of recovery prediction processes will end without calculating the recovery probability.

[0124] As described above, the return prediction system 100 according to this embodiment performs the return prediction process described above, making it possible to predict with high accuracy specific users who are likely to return from among a large number of dormant users. This makes it possible to implement customer attraction actions such as issuing coupons or announcing discount events that are effective only to specific users who are likely to return, and as a result, it becomes possible to increase the number of customers at the target facility without incurring unnecessary costs.

[0125] Furthermore, the prediction model 122 can be used to predict the probability of a user returning to the service, thereby improving the accuracy of determining which users are likely to return.

[0126] Furthermore, since the timing determination unit 115 determines the timing of the granting of the benefits, the benefit information is notified at a time when potential returning users are likely to take action to visit the target facility. This further increases the likelihood that potential returning users will return from their dormant state.

[0127] In the above-described embodiment, a recovery prediction system 100 including a management device 10 and a database 30 was illustrated, but the recovery prediction system of the present invention may consist only of the management device 10.

[0128] [Notes on the invention] The following is an overview of the invention extracted from the above-described embodiments. Note that each configuration and processing function described below can be selected and combined as desired.

[0129] <Note 1> A first acquisition unit acquires usage history information, including the usage history of users who have used products or services provided at the target facility, A dormancy determination unit determines, based on the aforementioned usage history information, which determines dormant users whose period of non-use since the last purchase at the target facility is longer than a predetermined period. A second acquisition unit acquires ride history information, including the ride history of the dormant user, who uses a vehicle operation service that operates vehicles along a route that includes the nearest bus stop to the aforementioned target facility. A reuse prediction unit predicts whether the dormant user will use the goods or services at the target facility again, based on the ride history information of the dormant user. A recovery prediction system equipped with this feature.

[0130] <Note 2> The aforementioned ride history includes the entry and exit records of the aforementioned bus stops, The return prediction system according to Appendix 1, wherein the reuse prediction unit determines a reuse index based on the ride history of the dormant user and predicts whether the dormant user will use the goods or services at the target facility again based on the reuse index.

[0131] <Note 3> The return prediction system described in Appendix 2, wherein the reuse prediction unit predicts that the dormant user will use the goods or services at the target facility again when the reuse index is above a predetermined threshold.

[0132] <Note 4> The reuse prediction unit is, The return prediction system according to Appendix 2 or 3, wherein the ride history of a target dormant user is input into a predictive model trained using a dataset including the usage history and ride history of multiple dormant return users who have used goods or services at the target facility from a dormant state where the period of non-use was longer than the predetermined period, thereby causing the predictive model to predict the reuse index.

[0133] <Note 5> The reuse prediction unit predicts whether the dormant user will use the goods or services at the target facility again, based on the dormant user's ride history information and usage history information, as described in Appendix 1 of the return prediction system.

[0134] <Note 6> The aforementioned ride history includes the entry and exit records of the aforementioned bus stops, The return prediction system according to Appendix 5, wherein the reuse prediction unit determines a reuse index based on the dormant user's ride history and user information, and predicts whether the dormant user will use the goods or services at the target facility again based on the reuse index.

[0135] <Note 7> The reuse prediction unit predicts that the dormant user will use the goods or services at the target facility again when the reuse index is above a predetermined threshold, as described in Appendix 6 of the return prediction system.

[0136] <Note 8> The reuse prediction unit is, The return prediction system according to Appendix 6 or 7, wherein the system causes the predictive model to predict the reuse index by inputting the usage history and ride history of the dormant user to be predicted into a predictive model that has been trained using a dataset including the usage history and ride history of multiple dormant users who have used goods or services at the target facility from a dormant state in which the period of non-use has been longer than the predetermined period.

[0137] <Note 9> The aforementioned ride history is, A return prediction system as described in any of Appendix 1 to 8, which includes one or more of the following: information on the bus stop used, the date of use of the bus stop, the time of use of the bus stop, and the frequency of use of the vehicle operation service.

[0138] <Note 10> The recovery prediction system according to any one of the appendices 1 to 9, further comprising a benefit granting processing unit that grants predetermined benefits to specific dormant users who are determined by the reuse prediction unit to be likely to use the service again.

[0139] <Note 11> The aforementioned benefit granting processing unit further comprises a timing determination unit that determines the timing for granting the predetermined benefit, The recovery prediction system according to Appendix 10, wherein the timing determination unit determines the timing at which a specific dormant user will use a specific section within a predetermined range from the nearest bus stop to the target facility, and sets that timing as the assigned timing.

[0140] <Note 12> The recovery prediction system according to Appendix 10, further comprising a timing determination unit that determines the timing at which the reward granting processing unit grants the predetermined reward based on the timing at which the specific dormant user uses the vehicle operation service.

[0141] <Note 13> The timing determination unit determines the granting timing within a range from the start of use of the vehicle operation service by a specific dormant user until a predetermined time has elapsed, as described in Appendix 12 of the recovery prediction system.

[0142] <Note 14> The aforementioned predetermined benefit is a benefit applicable to the same type of product or service as the product or service with the highest usage frequency or usage amount included in the usage history, as described in any of Appendix 10 to 13 of the recovery prediction system.

[0143] <Note 15> The first acquisition step involves obtaining usage history information, including the usage history of users who have used products or services provided at the target facility. A dormancy determination step that determines dormant users whose period of non-use since the last purchase at the target facility is longer than a predetermined period, based on the aforementioned usage history information, A second acquisition step involves acquiring ride history information, including the ride history of the dormant user, who uses a vehicle operation service that operates vehicles along a route that includes the nearest bus stop to the aforementioned target facility. A reuse prediction step that predicts whether the dormant user will use the goods or services at the target facility again, based on the ride history information of the dormant user, A recovery prediction method performed by one or more processors.

[0144] <Note 16> The first acquisition step involves obtaining usage history information, including the usage history of users who have used products or services provided at the target facility. A dormancy determination step that determines dormant users whose period of non-use since the last purchase at the target facility is longer than a predetermined period, based on the aforementioned usage history information, A second acquisition step involves acquiring ride history information, including the ride history of the dormant user, who uses a vehicle operation service that operates vehicles along a route that includes the nearest bus stop to the aforementioned target facility. A reuse prediction step that predicts whether the dormant user will use the goods or services at the target facility again, based on the ride history information of the dormant user, A program that causes one or more processors to run. [Explanation of Symbols]

[0145] 10: Management device 11: Control Unit 12: Storage section 13: Communications Department 14:Display section 15:Operation unit 30: Database 31: User registration data 32: Boarding and alighting history data 33: Route Data 34: Train schedule data 40: User terminal 100: Recovery prediction system 111: Data acquisition unit 112: Dormancy determination section 113: Reuse prediction unit 114: Special Offers Department 115: Timing determination unit 116: Reward Granting Processing Section 121: Control Program 122: Predictive Models 123: Training dataset 150: Predictive Model Construction Department 151: Training Dataset Generation Unit 152: Evaluation and Judgment Department 200: Customer Management System 201: Purchase Customer Data 202: Purchase history data 300: Passenger boarding and alighting management system

Claims

1. A first acquisition unit acquires usage history information, including the usage history of users who have used products or services provided at the target facility. A dormancy determination unit determines, based on the aforementioned usage history information, which determines dormant users whose period of non-use since the last purchase at the target facility is longer than a predetermined period. A second acquisition unit acquires ride history information, including the ride history of the dormant user, who uses a vehicle operation service that operates vehicles along a route that includes the nearest bus stop to the aforementioned target facility. A reuse prediction unit predicts whether the dormant user will use the goods or services at the target facility again, based on the ride history information of the dormant user. A recovery prediction system equipped with this feature.

2. The aforementioned ride history includes the entry and exit records of the aforementioned bus stops, The return prediction system according to claim 1, wherein the reuse prediction unit determines a reuse index based on the ride history of the dormant user and predicts whether the dormant user will use the goods or services at the target facility again based on the reuse index.

3. The return prediction system according to claim 2, wherein the reuse prediction unit predicts that the dormant user will use the goods or services at the target facility again when the reuse index is above a predetermined threshold.

4. The reuse prediction unit is, The return prediction system according to claim 2, wherein the ride history of a dormant user to be predicted is input into a predictive model trained using a dataset including the usage history and ride history of multiple dormant users who have used goods or services at the target facility from a dormant state in which the period of non-use has been longer than the predetermined period, thereby causing the predictive model to predict the reuse index.

5. The return prediction system according to claim 1, wherein the reuse prediction unit predicts whether the dormant user will use the goods or services at the target facility again, based on the ride history information and usage history information of the dormant user.

6. The aforementioned ride history includes the entry and exit records of the aforementioned bus stops, The return prediction system according to claim 5, wherein the reuse prediction unit determines a reuse index based on the dormant user's ride history and usage history information, and predicts whether the dormant user will use the goods or services at the target facility again based on the reuse index.

7. The return prediction system according to claim 6, wherein the reuse prediction unit predicts that the dormant user will use the goods or services at the target facility again when the reuse index is above a predetermined threshold.

8. The reuse prediction unit is, The return prediction system according to claim 6, wherein the user inputs the usage history and ride history of a dormant user to be predicted into a predictive model trained using a dataset including the usage history and ride history of multiple dormant users who have used goods or services at the target facility from a dormant state in which the period of non-use has been longer than the predetermined period, thereby causing the predictive model to predict the reuse index.

9. The aforementioned ride history is, The return prediction system according to claim 1 or 2, comprising one or more of the following: information on the bus stop used, the date of use of the bus stop, the time of use of the bus stop, and the frequency of use of the vehicle operation service.

10. The recovery prediction system according to claim 1 or 2, further comprising a reward granting processing unit that grants predetermined rewards to specific dormant users who are determined by the reuse prediction unit to be likely to use the service again.

11. The aforementioned benefit granting processing unit further comprises a timing determination unit that determines the timing for granting the predetermined benefit, The recovery prediction system according to claim 10, wherein the timing determination unit determines the timing at which a specific dormant user will use a specific section within a predetermined range from the nearest bus stop to the target facility, and determines that timing at which the user will use the designated timing.

12. The recovery prediction system according to claim 10, further comprising a timing determination unit that determines the timing at which the reward granting processing unit grants the predetermined reward based on the timing at which the specific dormant user uses the vehicle operation service.

13. The recovery prediction system according to claim 12, wherein the timing determination unit determines the granting timing within a range from the start of use of the vehicle operation service by the specific dormant user until a predetermined time has elapsed.

14. The recovery prediction system according to claim 10, wherein the predetermined benefit is a benefit applicable to the same type of product or service as the product or service with the highest usage frequency or usage amount included in the usage history.

15. The first acquisition step involves obtaining usage history information, including the usage history of users who have used products or services provided at the target facility. A dormancy determination step that determines dormant users whose period of non-use since the last purchase at the target facility is longer than a predetermined period, based on the aforementioned usage history information, A second acquisition step involves acquiring ride history information, including the ride history of the dormant user, who uses a vehicle operation service that operates vehicles along a route that includes the nearest bus stop to the aforementioned target facility. A reuse prediction step that predicts whether the dormant user will use the goods or services at the target facility again, based on the ride history information of the dormant user, A recovery prediction method performed by one or more processors.

16. The first acquisition step involves obtaining usage history information, including the usage history of users who have used products or services provided at the target facility. A dormancy determination step that determines dormant users whose period of non-use since the last purchase at the target facility is longer than a predetermined period, based on the aforementioned usage history information, A second acquisition step involves acquiring ride history information, including the ride history of the dormant user, who uses a vehicle operation service that operates vehicles along a route that includes the nearest bus stop to the aforementioned target facility. A reuse prediction step that predicts whether the dormant user will use the goods or services at the target facility again, based on the ride history information of the dormant user, A program that causes one or more processors to run.

Images