JP7487866B2 - Loan management system, method, and program - Google Patents

Description

The present invention relates to technology that supports the provision of flexible loan services.

Financial institutions offer a variety of loan services to individuals, sole proprietors, and corporations. In many loan services, repayment plans, including interest rates, repayment periods, and monthly repayment amounts, are determined in advance, and users make repayments according to the repayment plans. The repayment plans for these loan services are determined when users begin using the service. Essentially, the repayment plans remain fixed until the user finishes using the loan service.

However, a user's economic situation and repayment ability are not constant but change daily. Therefore, convenience for users would be improved if they could flexibly select and change repayment methods, such as the repayment period and repayment amount, of the loan service.

In regards to this, while it is possible to a certain extent to obtain additional loans or change the repayment plan, the procedures are complicated. For example, the user must consider various repayment plans as candidates, and in some cases consult closely with the financial institution, and then apply to the financial institution for a change to the repayment plan that they have decided upon. The financial institution must then make a human judgment based on the user's credit risk to determine whether or not to approve the change, and the change is only implemented once the decision has been made to approve it.

In response to this, Patent Document 1 discloses a technique for determining whether or not to permit a change in repayment terms based on a user's score. Specifically, Patent Document 1 describes a technique for selecting at least one model based on the internal financial institution information about the customer from among a plurality of models each having a different relationship that represent the relationship between the financial institution's internal information about the customer and a score used to calculate the customer's default rate, calculating the customer's default rate from the selected model based on the financial institution's internal information about the customer, determining a rank based on the calculated default rate, and determining whether or not a guaranteed loan being executed by the financial institution can be changed to a proper loan based on at least the rank.

JP 2017-182284 A

According to the technology disclosed in Patent Document 1, if a user wishes, it is possible to determine whether or not to switch between a regular loan and a guaranteed loan. However, it is difficult for the user to determine which loan services are candidates and which loan service to select.

The object of the present invention is to provide a technology that enables appropriate loan services to be recommended to users.

A loan management system according to one aspect of the present disclosure includes a loan management unit that manages, for each of a plurality of users, user information that records the attributes and behavior of the user, and service content including the interest rate, repayment time, and repayment amount for the currently used loan service, which is the loan service currently being used; an input/output unit that acquires a change request for a target user among the plurality of users to change from the currently used loan service to another loan service; and a recommended loan service determination unit that determines, for the target user, one or more loan services that are estimated to generate a predetermined profit based on the change request, the user information, and the service content of the currently used loan service, as recommendable loan services, calculates an adoption expectation degree, which is the degree of ease of adoption by the target user, for the recommendable loan service, and determines one or more of the recommendable loan services as recommended loan services to be recommended to the target user based on the adoption expectation degree.

According to one aspect of the present disclosure, it is possible to recommend appropriate loan services to users.

FIG. 2 is a block diagram of a variable loan management system according to the present embodiment. 2 is a block diagram showing the hardware configuration of a learning device 11 and a management device 14. FIG. 13 is a flowchart of a loan service change process by the management device 14. FIG. 2 is a diagram for explaining the recommended loan service determination unit in detail. 13 is a flowchart of a recommended loan service selection process.

The following describes an embodiment of the present invention with reference to the drawings.

Figure 1 is a block diagram of a variable loan management system according to this embodiment.

Figure 2 is a block diagram showing the hardware configuration of the learning device 11 and the management device 14.

As shown in FIG. 1, the variable loan management system 10 can be connected to a user terminal 18 used by a user 90 via a communication network (not shown) such as the Internet. The user 90 can connect to the management device 14 using an application or browser on the user terminal 18, such as a personal computer or smartphone, and use the variable loan service managed by the management device 14.

In this embodiment, the variable loan management system 10 provides a variable loan service via a dedicated application installed on the user terminal 18. The variable loan service is a service that provides a number of loan services with different basic repayment plans and/or options that can be changed at any time.

As basic repayment plans, various repayment methods (lump sum repayment at maturity, equal principal and interest repayments, equal principal repayments, flexible principal repayments, etc.), interest rates, repayment periods, and repayment amounts (lump sums, monthly repayment amounts, etc.) can be set. Furthermore, as options, depending on the repayment method, a number of different loan services are available, such as whether or not to require a guarantor, whether or not to take out insurance (life insurance), etc.

As shown in FIG. 1, the variable loan management system 10 has a learning device 11 and a management device 14. The learning device 11 has a data collection unit 12 and a learning unit 13. The management device 14 has an input/output unit 15, a recommended loan service determination unit 16, and a loan management unit 17.

As shown in FIG. 2, both the learning device 11 and the management device 14 are configured as computers having hardware including a processor 21, a main memory 22, a storage device 23, a communication device 24, an input device 25, and a display device 26, all of which are connected to a bus 27.

The data collection unit 12 shown in FIG. 1 collects and stores data 101 on the attributes and behavior of each of multiple users 90.

The learning unit 13 uses attribute and behavioral data 101 of each of the multiple users 90 as learning data to construct a predictive model that is used by the recommended loan service determination unit 16 to determine the recommended loan service, and updates it as appropriate.

The input/output unit 15 provides a variable loan service to the user 90 via an application installed on the user terminal 18.

When user 90 starts using the service, input/output unit 15 displays a start-of-use screen on user terminal 18, accepts user 90's application to start using the service from user terminal 18, and acquires data on user 90's attributes. The attribute information includes, for example, information on user 90's age and place of employment. In addition, input/output unit 15 acquires external credit information on user 90 provided by an external institution (not shown). The external credit information includes, for example, credit information and personal credit score provided by a personal credit information agency. Input/output unit 15 sends the start-of-use application, together with the user's attribute information and external credit information, to loan management unit 17, and provides the variable loan service to user 90 after screening.

The loan management unit 17 manages, for each of the multiple users 90, user information that records the attributes and behavior of each user 90, and service details including the interest rate, repayment time, and repayment amount for the currently used loan service, which is the loan service being used.

While the user 90 is using the loan service, the input/output unit 15 presents various information to the user 90 via the application, and also continuously acquires data on various actions related to the user 90. For example, the input/output unit 15 presents information on the repayment history and repayment schedule for the loan service being used. The input/output unit 15 also acquires past transaction information including the number of times the transaction was used, the payment period, and whether or not there was a delay in payment, for past transactions. Furthermore, the input/output unit 15 acquires information on the usage status of the application by the user 90. Furthermore, the input/output unit 15 acquires data acquired by a sensor (not shown) equipped in the user terminal 18.

The usage information includes the application startup time, the number of times a loan service change was considered, the details of the consideration, and the reaction to the loan service change notification. The number of times a loan service change was considered can be obtained from the number of times the change screen in the application was displayed. The details of the consideration can be obtained from the operation history on the change screen, for example, the operation history of selecting the desired loan service. The reaction to the loan service change notification can be obtained by various means, for example, from the browsing history of the site linked to in an email or short message providing information about the loan service change function.

The usage information also includes information about the use of the service by other applications. Examples of other applications include a household accounting application and a health management application.

In addition, information acquired by the sensor equipped in the user terminal 18 includes, for example, location information.

In addition, while the variable loan service is being provided, the input/output unit 15 accepts a change request from the user 90 wishing to change from the loan service being used to another loan service.

As described above, the variable loan service of this embodiment provides a large number of loan services. Therefore, it is difficult for the user 90 to select an appropriate loan service by himself/herself. Therefore, when there is a change request to change the loan service for a certain target user, the variable loan management system 10 determines multiple recommended loan services by the recommended loan service determination unit 16, and presents the recommended loan services to the user 90 via the application by the input/output unit 15.

The recommended loan service determination unit 16 determines, for the target user, one or more loan services that are estimated to generate a specified profit based on the change request, user information, and the service content of the loan service being used, as recommendable loan services, calculates the adoption expectancy, which is the degree of likelihood that the recommendable loan services will be adopted by the target user, and determines the recommendable loan services having the specified adoption expectancy as recommended loan services to be recommended to the target user. In this case, the recommended loan service determination unit 16 uses the prediction model constructed by the learning unit 13 to determine the recommendable loan services and calculate the adoption expectancy. The recommended loan service determination unit 16 will be described in detail later.

In FIG. 2, the storage device 23 stores data in a writable and readable manner, and stores data used for various processes.

The processor 21 reads the data stored in the storage device 23 into the main memory 22 and uses the main memory 22 to execute the processing of the software program. In the case of the learning device 11, the processor 21 realizes the data collection unit 12 and the learning unit 13 shown in FIG. 1. In the case of the management device 14, the processor 21 realizes the recommended loan service determination unit 16 and the loan management unit 17 shown in FIG. 1.

The communication device 24 transmits information processed by the processor 21 via a communication network that may be wired or wireless or both, and also transmits information received via the communication network to the processor 21. The received information is used by the processor 21 for software processing.

The input device 25 is a device that accepts information input by an operator using a keyboard, mouse, etc., and the input information is used for software processing by the processor 21.

The display device 26 is a device that displays image and text information on a display screen in conjunction with software processing by the processor 21.

Figure 3 is a flowchart of the loan service change process by the management device 14.

When the management device 14 receives a change request from the user terminal 18 via the input/output unit 15 (step S101), the recommended loan service determination unit 16 executes a recommended loan service selection process. The recommended loan service selection process is a process for selecting a recommended loan service, and the details of this process will be described later.

Next, the management device 14 presents information about the recommended loan service to the user 90 via the application on the user terminal 18 through the input/output unit 15 (step S103), and then receives confirmation of the loan service to be changed from the user 90 (step S104).

Then, the management device 14 executes the change of the loan service through the input/output unit 15 (step S105). At this time, the management device 14 updates the information of the loan service being used through the loan management unit 17.

Figure 4 is a diagram for explaining the recommended loan service determination unit in detail.

Referring to FIG. 4, the recommended loan service determination unit 16 has a recommendable loan service extraction unit 31 and an adoptability determination unit 32. The recommendable loan service extraction unit 31 has a loan service enumeration unit 33, a profitability calculation unit 34, and a profitability determination unit 35.

The recommendable loan service extraction unit 31 extracts one or more (e.g., three) loan services that are estimated to generate a predetermined profit for the target user based on the change request, user information, and the service content of the loan service currently being used, as recommendable loan services. The change request may include change condition information. The change condition information is information specified by the user 90 as the conditions for the recommended loan service. For example, the conditions may include a condition that the monthly repayment amount is reduced compared to the loan service currently being used, or a condition that the repayment period is shortened compared to the loan service currently being used. When there is change condition information, the recommendable loan service extraction unit 31 may extract recommendable loan services that satisfy the conditions indicated in the change condition information. Furthermore, when there are multiple change condition information, the recommendable loan service extraction unit 31 may extract recommendable loan services that satisfy the conditions of all the change condition information, or may extract recommendable loan services that satisfy the conditions of some of the change condition information.

The adoptability determination unit 32 calculates the adoption expectancy, which is the degree of likelihood of adoption by the target user, for the recommendable loan services extracted by the recommendable loan service extraction unit 31, and determines the recommendable loan services having a predetermined adoption expectancy as recommended loan services to be recommended to the target user.

As shown in FIG. 4, the recommendable loan service extraction unit 31 has a loan service enumeration unit 33, a profitability calculation unit 34, and a profitability determination unit 35.

Figure 5 is a flowchart of the recommended loan service selection process.

The recommended loan service determination unit 16 extracts a plurality of loan services with different repayment plans including interest rates, repayment times, and repayment amounts by applying constraint conditions to the amount of income obtained when repayment is made as planned after changing from the loan service currently in use by the loan service enumeration unit 33 (step S201). The amount of income referred to here is the total amount of interest paid by simply making repayments as planned, without considering the risk of default or early repayment. As an example of the constraint condition, the difference between the amount of income obtained when the use of the loan service currently in use is continued and repayment is made as planned, and the amount of income obtained when repayment is made as planned after changing from the loan service currently in use satisfies a predetermined difference condition. The amount of income obtained when the use of the loan service currently in use is continued and repayment is made as planned is set as the first amount of income, the amount of income obtained when repayment is made as planned after changing from the loan service currently in use is set as the second amount of income, and the difference obtained by subtracting the first amount of income from the second amount of income is the increase in the amount of income when the target user changes the loan service. If this increase is large, the business entity's income will increase, but the target users may feel that they are losing out by changing the loan service. Therefore, it is preferable to impose appropriate constraints on the first income amount and limit the difference between the second income amount and the first income amount to an appropriate range.

Next, the recommended loan service determination unit 16 calculates, by the profitability calculation unit 34, an adjusted income amount that reflects the risk of the target user based on the user information for each loan service extracted in step S201 (step S202). The adjusted income amount here is the expected value of the total interest amount obtained, taking into account the risks of default and early repayment. Specifically, the adjusted income amount is the expected value of the income amount obtained from the loan service, assuming that default or early repayment occurs with a probability based on the user information of the target user.

The amount of revenue or loss (negative revenue) from servicing a loan in the event of a default or prepayment may be calculated exactly or may be approximated in some way.

For example, when a default occurs, the loss amount may be calculated by subtracting the total amount of interest earned from the start of the loan to the time the default occurs from the principal portion of the loan balance at the time of the default. Alternatively, the loss amount may be approximated by the principal portion of the loan balance at the time of the default. Alternatively, the loss amount may be approximated by the loan amount at the start of the loan when the default occurred, regardless of the time of the default.

In addition, for example, full early repayment may be taken into consideration, and if an early repayment occurs, the total amount of interest earned up to the time of occurrence may be regarded as the income amount. Partial early repayment may also be taken into consideration, and if a partial early repayment occurs, the total amount of interest earned until the repayment is completed may be regarded as the income amount.

The processing of this step S202 by the profitability calculation unit 34 will be described in detail later.

Then, the recommended loan service determination unit 16 determines, from among the multiple loan services extracted in step S201 by the profitability judgment unit 35, as a recommendable loan service, a loan service whose adjusted income amount satisfies a predetermined income condition (step S203). One example of the income condition is that the adjusted income amount is equal to or greater than a predetermined threshold. The threshold can be set arbitrarily by the business entity based on the business policy, etc.

Next, the recommended loan service determination unit 16 calculates the adoption expectancy for each recommendable loan service by the adoptability determination unit 32 (step S204). The adoption expectancy is the degree of likelihood that the recommendable loan service will be adopted by the target user. At this time, the adoptability determination unit 32 calculates the adoption expectancy using the latest fit determination model 102A constructed by the learning unit 13. Details of the fit determination model 102A will be described later.

Next, the recommended loan service determination unit 16 selects a recommended loan service from among the recommendable loan services based on the expected adoptability level by the adoptability determination unit 32. For example, a recommendable loan service whose expected adoptability level exceeds a predetermined threshold may be set as the recommended loan service. Alternatively, a predetermined number of recommendable loan services with the highest expected adoptability levels may be set as the recommended loan services.

Alternatively, the adoptability determination unit 32 may calculate an index (recommendation degree) indicating the degree of recommendation based on both the adjusted income amount and the expected adoptability, and select a recommended loan service from among the recommendable loan services based on the value of the recommendation degree. In addition, the ranking of the recommended loan services when displaying a plurality of recommended loan services on the user terminal 18 and recommending them to the target user 90 may be determined based on the recommendation degree. The recommendation degree here is determined by a predetermined balance between the adjusted income amount and the expected adoptability. It is possible to arbitrarily determine which is emphasized and by what calculation formula the recommendation degree is calculated. For example, the adjusted income amount may be emphasized. The calculation formula for the recommendation degree may be changeable according to the marketing policy. In addition, in order to prevent the user from feeling that he or she is suffering a large disadvantage, the adjusted income amount may not be significantly higher than the expected income amount of the loan service being used. In addition, in consideration of the profitability of the business entity, the adjusted income amount may not be lower than the expected income amount of the loan service being used.

The processing of step S202 by the profitability calculation unit 34 is described in detail below.

As shown in FIG. 4, the profitability calculation unit 34 has a default risk calculation unit 36, an adjusted income calculation unit 37, a prepayment risk calculation unit 38, and a future income calculation unit 39.

In the processing of step S202 by the profitability calculation unit 34, the default risk calculation unit 36 predicts the probability of default occurring when the target user uses the target loan service. At this time, the default risk calculation unit 36 calculates the probability of default occurring using the latest default risk determination model constructed by the learning unit 13. The details of the default risk determination model will be described later.

The prepayment risk calculation unit 38 also predicts the probability of early repayment occurring when the target user uses the target loan service. At this time, the prepayment risk calculation unit 38 calculates the probability of early repayment occurring using the latest prepayment risk determination model 102C constructed by the learning unit 13.

The future profit calculation unit 39 predicts the amount of income (future profit) that the target user will receive from using additional services when the target user uses the target loan service. At this time, the future profit calculation unit 39 calculates the future profit using the latest future profitability evaluation model constructed by the learning unit 13.

The adjusted income calculation unit 37 calculates the adjusted income amount for each loan service listed by the loan service enumeration unit 33. The adjusted income amount is the expected total income amount obtained from the changed loan service and the additional service when a default or early repayment occurs with a probability based on the user information of the target user and the target user changes the loan service and an income amount is obtained by additionally using another service.

Explain the default risk assessment model 102B.

The default risk assessment model 102B according to this embodiment is a model that estimates the probability of a default occurring in the future. In this embodiment, the default risk assessment model 102B receives user information of the target user as input, and outputs the probability of a default occurring at each point in a predetermined future period. The period and interval for calculating the default probability are not particularly limited, but for example, the period may be a common period such as one year or three years, or may be an individual period such as the payment period of a loan service. In addition, the interval may be monthly or annual.

The learning unit 13 constructs a default risk determination model 102B by machine learning using the user information of each user previously collected by the data collection unit 12 and the occurrence and timing of default as learning data.

Explain the prepayment risk judgment model 102C.

The prepayment risk determination model 102C according to this embodiment is a model that estimates the probability of future early repayments. In this embodiment, the prepayment risk determination model 102C receives user information of the target user as input, and outputs the probability of early repayments occurring at each point in a predetermined future period. Here , early repayment refers to full early repayment. The period and interval for calculating the early repayment probability are not particularly limited, but may be the same period as the period for which the default risk determination model 102B predicts the default probability, for example.

In the example of this embodiment, for simplicity, only full prepayments are considered, but this is not limiting. For partial prepayments as well, if a model is constructed that predicts the probability of prepayments occurring at each point in a specified period and the repayment amount at that time, it is possible to calculate the adjusted income amount.

In addition, the learning unit 13 constructs a prepayment risk assessment model 102C by machine learning using the user information of each user previously collected by the data collection unit 12 and the occurrence and timing of early repayment as learning data.

Explain future profitability assessment model 102D.

The future profitability evaluation model 102D according to this embodiment is a model that estimates the amount of income that a user who has changed their loan service will obtain from other services that they use thereafter. In this embodiment, the future profitability evaluation model 102D takes as input the user information of the target user, and outputs the amount of income from other services that the user will use after changing their loan service.

In the learning unit 13, a future profitability evaluation model 102D is constructed by machine learning using the user information of each user accumulated in the past by the data collection unit 12 and the amount of income from other services used by the user after changing the loan service as learning data. However, since a future profitability evaluation model 102D with sufficient accuracy cannot be constructed by machine learning until sufficient data is accumulated, the future profitability evaluation model 102D may be set by other methods. For example, based on the knowledge and experience of the business entity's staff and external people, the future use of other services by the user and/or the amount of income obtained therefrom may be estimated for each life stage of the user that can be known from the user information, and the estimated result may be set as a model.

Details of the fit determination model 102A are explained below.

The fit determination model 102A according to this embodiment is a model that estimates the adoption expectancy, which is the degree of likelihood that the recommended loan service will be adopted by the target user. The adoption expectancy is, as an example, the probability that the loan service recommended to the user will be adopted. The fit determination model 102A receives as input the user information of the target user and the service content of the target loan service, and outputs the adoption expectancy for the target loan service and the target user.

The learning unit 13 uses machine learning to construct a fit determination model 102A that predicts the probability that the loan service recommended to the user will be adopted as the expected adoption degree, using user information collected in the past by the data collection unit 12 and whether or not the loan service recommended to the user has been adopted as learning data. However, since a fit determination model 102A with sufficient accuracy cannot be constructed by machine learning until sufficient data has been accumulated, the fit determination model 102A may be set by other methods. For example, a staff member of the business entity may create a rule logic based on knowledge and experience and set it as the model.

In this embodiment, an example has been shown in which a recommended loan service is presented in response to a request from the user 90, but this is not limited to this. Recommended loan services may be presented in response to any other trigger. For example, the trigger may be a request generated based on user information, etc. within the variable loan management system 10, or a request input by an operator of the business entity to the variable loan management system 10 by specifying a target user.

In addition, in this embodiment, an example has been shown in which the adjusted income amount is calculated taking into account defaults, prepayments, and future earnings, but this is not limiting. As another example, the adjusted income amount may be calculated taking into account defaults and prepayments without taking into account future earnings, or the adjusted income amount may be calculated taking into account defaults without taking into account prepayments and future earnings.

The above-described embodiments are illustrative examples of the present invention, and are not intended to limit the scope of the present invention to these embodiments. Those skilled in the art may implement the present invention in various other forms without departing from the scope of the present invention.

The technical scope of the above-described embodiment includes the following items. However, the technical scope of the present embodiment is not limited to these items.

(Aspect 1)
a loan management unit that manages, for each of a plurality of users, user information that records the attributes and behavior of the user, and service contents including an interest rate, a repayment time, and a repayment amount for a currently used loan service that is a loan service being used;
an input/output unit for acquiring a change request for changing a target user among the plurality of users from the currently used loan service to another loan service;
a recommended loan service determination unit that determines, for the target user, one or more loan services that are estimated to generate a predetermined profit based on the change request, the user information, and the service content of the loan service being used, as recommendable loan services, calculates an adoption expectation degree, which is the degree of ease with which the recommendable loan services are adopted by the target user, and determines the recommendable loan services having the predetermined adoption expectation degree as recommended loan services to be recommended to the target user;
A loan management system having:

When a user changes loan services, it is desirable to recommend to the user a reasonable number of available loan services for the user to consider. For example, as a basic repayment plan, there are many selectable elements such as interest rate (interest rate for lump-sum repayment at maturity, interest rate for general credit transactions (unlimited), etc.), repayment amount (lump sum, principal, interest rate in repayment amount, etc.), repayment period, and, as options, whether or not to have a guarantor, whether or not to take out insurance (life insurance), etc. Even if there are loan services with a wide range of service contents, it is not desirable to recommend too many loan services to the user. However, it is not easy to select a reasonable number of recommended loan services with reasonable service contents from the countless wide range of loan services available.

According to aspect 1, from among various loan services, loan services are narrowed down based on the condition that they are estimated to generate a predetermined profit, and the degree of ease of adoption by the target user from among the narrowed-down loan services is evaluated to determine the loan services to be recommended to the target user, so that a reasonable number of recommended loan services that are easy to consider can be recommended to the user.

(Aspect 2)
In the loan management system of aspect 1,
The recommended loan service determination unit extracts a number of loan services with different repayment plans, including interest rates, repayment times, and repayment amounts, by applying constraints to the amount of income that can be obtained if repayments are made as planned after changing from the loan service currently being used, calculates adjusted income amounts for the extracted number of loan services that reflect the risk of the target user based on the user information, and determines, among the extracted number of loan services, the loan service whose adjusted income amount satisfies specified revenue conditions as the recommendable loan service.

According to aspect 2, the loan services are narrowed down by imposing constraints on a simple income amount that can be easily calculated, and the adjusted income amount of the narrowed down loan services is evaluated, so that an appropriate number of recommended loan services can be determined with a small amount of calculation.

(Aspect 3)
In the loan management system of aspect 2,
The constraint condition includes a condition that a difference between an amount of income to be obtained if the use of the currently used loan service is continued and repayment is made as planned and an amount of income to be obtained if repayment is made as planned after changing from the currently used loan service satisfies a predetermined difference condition.

According to aspect 3, by imposing a restriction on the difference between the amount of income from the loan service currently being used and the amount of income from the new loan service, when narrowing down the recommended loan services, it is possible to prevent an inconvenient situation in which a loan service that the user feels is significantly disadvantaged or a loan service that significantly reduces the profits of the service provider is selected as the recommended loan service.

(Aspect 4)
In the loan management system of aspect 2,
The adjusted income amount is an expected value of the income amount obtained from the loan service in the event that a default occurs with a probability based on the user information of the target user.

According to aspect 4, by imposing constraints on the simple amount of income expected if repayments are made as planned, it becomes easier to narrow down the loan services, and since an evaluation is performed based on the adjusted income amount before actually recommending a loan service to a user, the service provider can recommend a loan service that has undergone a proper evaluation in terms of income.

(Aspect 5)
In the loan management system of aspect 2,
The adjusted income amount is the expected total income amount to be obtained from the changed loan service and the additional service, assuming that a default occurs with a probability based on the user information of the target user, and that the target user will obtain income by using additional services after changing the loan service.

Aspect 5 makes it possible to recommend loan services that take into account the income that a user who has changed loan services can earn by subsequently using other services.

(Aspect 6)
In the loan management system of aspect 2,
The recommended loan service determination unit determines a recommendation order for the recommended loan services based on the adjusted income amount and the expected adoption rate.

In aspect 6, the recommended loan services are ranked based on the adjusted income amount and the expected adoption rate, so that loan services that are profitable and suitable for the user can be recommended.

(Aspect 7)
In the loan management system of aspect 4,
A data collection unit that collects attribute and behavior data of a plurality of users;
a learning unit that machine-learns learning data including at least some attribute and behavior data of a plurality of users and whether or not the plurality of users have defaulted, constructs a default risk determination model for predicting the probability of default, and machine-learns learning data including at least some attribute and behavior data of the plurality of users and whether or not the plurality of users have adopted a loan service recommended to the plurality of users, constructs a fit determination model for predicting the probability that the loan service recommended to the user will be adopted as the adoption expectation degree, and appropriately updates the model;
The recommended loan service determination unit takes at least a portion of the user information of the target user as input, calculates the expected value of the income amount based on the probability of default predicted by the latest default risk assessment model, and sets it as the adjusted income amount, and takes at least a portion of the user information of the target user as input, and predicts the expected adoption rate using the latest fit assessment model.

According to aspect 7, the prediction model is appropriately updated, and the index value used to determine whether or not to recommend a loan service is calculated using the latest prediction model, which contributes to the realization of a loan service that can be flexibly changed.

(Aspect 8)
In the loan management system of aspect 7,
The default risk determination model is a prediction model for predicting the probability of a default occurring at each time point of occurrence, the model being constructed by machine learning learning data including data on the attributes and behaviors of at least a portion of the users and whether or not the users have defaulted and the time point of occurrence,
The recommended loan service determination unit receives as input at least a portion of the user information of the target user and calculates the expected income amount based on the time and probability of default occurring predicted by the latest default risk assessment model.

According to aspect 8, the adjusted income amount is predicted taking into account the probability of default at each time point, making it possible to make a reasonable decision regarding whether or not to recommend a loan service, taking into account the passage of time.

(Aspect 9)
In aspect 8,
The learning unit further performs machine learning on learning data including data on the attributes and behavior of at least a portion of the users and the presence or absence and time point of prepayment, and constructs and appropriately updates a prepayment risk determination model for predicting the occurrence probability of prepayment at each time point of occurrence;
The recommended loan service determination unit calculates the expected amount of income based on the timing and probability of default predicted by the latest default risk assessment model, and the probability and time of early repayment predicted by the latest prepayment risk assessment model.

According to aspect 9, it is possible to predict the adjusted income amount taking into account not only defaults but also early repayments.

(Aspect 10)
In the loan management system of aspect 7,
The learning unit further performs machine learning on learning data including attribute and behavior data of the multiple users and income amounts from other services used by the multiple users after the loan service is changed, and constructs and appropriately updates a future profitability evaluation model for predicting income amounts from additional use of other services after the loan service is changed;
The recommended loan service determination unit inputs at least a portion of the user information of the target user, and calculates the expected amount of income based on the probability of default predicted by the latest default risk assessment model and the amount of income predicted by the latest future profitability evaluation model.

According to aspect 10, the adjusted revenue amount can be predicted taking into account future revenues that can be obtained by additionally using other services.

(Aspect 11)
In the loan management system of aspect 7,
The user actions include actions regarding changes to loan services.

In general, user behavior regarding changing loan services is thought to be behavior that reflects changes in the user's financial situation, repayment ability, etc. For example, when repayment ability declines, a desire to lower monthly repayment amounts may come into play.

Furthermore, if it becomes easier to change loan services as disclosed herein, it is conceivable that user behavior regarding changing loan services will begin to reflect previously unanticipated factors, such as the degree of interest and involvement that users have in making a living by being conscious of future repayments on the loan service they are currently using.

According to aspect 11, by including user behavior regarding changes made by the user in the learning data, it is possible to calculate a score that takes into account the user's changes in situation and/or the degree of interest and involvement. As a result, it becomes possible to appropriately determine whether or not to allow a change in the repayment method.

10... Variable loan management system, 11... Learning device, 12... Data collection unit, 13... Learning unit, 14... Management device, 15... Input/output unit, 16... Recommended loan service determination unit, 17... Loan management unit, 18... User terminal, 21... Processor, 22... Main memory, 23... Storage device, 24... Communication device, 25... Input device, 26... Display device, 27... Bus, 31... Recommended loan service extraction unit, 32... Adoption judgment unit, 33... Loan service enumeration unit, 34... Profitability calculation unit, 35... Profitability judgment unit, 36... Default risk calculation unit, 37... Adjusted income calculation unit, 38... Prepayment risk calculation unit, 39... Future income calculation unit, 90... User, 101... Data, 102A... Fit judgment model, 102B... Default risk judgment model, 102C... Prepayment risk judgment model, 102D... Future profitability evaluation model

Claims (13)

a loan management unit that manages, for each of a plurality of users, user information that records the attributes and behavior of the user, and service contents including an interest rate, a repayment time, and a repayment amount for a currently used loan service that is a loan service being used;
an input/output unit for acquiring a change request for changing a target user among the plurality of users from the currently used loan service to another loan service;
a recommended loan service determination unit that determines, for the target user, one or more loan services that are estimated to generate a predetermined profit based on the change request, the user information, and the service content of the loan service being used, as recommendable loan services, calculates an adoption expectation degree, which is the degree of ease with which the recommendable loan services are adopted by the target user, and determines one or more of the recommendable loan services as recommended loan services to be recommended to the target user based on the adoption expectation degree;
a learning unit that performs machine learning on learning data including data on the attributes and behaviors of at least a portion of a plurality of users and information on whether the loan service recommended to the plurality of users has been adopted, and constructs a fit determination model for predicting the probability that the loan service recommended to a user will be adopted as the adoption expectation degree;
having
the recommended loan service determination unit receives as input at least a portion of the user information of the target user and the service content of the recommendable loan service, and predicts the adoption expectancy of the recommendable loan service using the fit determination model;
Loan management system. The recommended loan service determination unit extracts a plurality of loan services with different repayment plans including interest rates, repayment times, and repayment amounts by applying constraints to the amount of income that can be obtained if repayments are made as planned after changing from the loan service currently being used, calculates an adjusted income amount for the extracted plurality of loan services that reflects the risk of the target user based on the user information, and determines, among the extracted plurality of loan services, a loan service whose adjusted income amount satisfies a predetermined revenue condition as the recommendable loan service.
The loan management system according to claim 1 . The constraint condition includes a condition that a difference between an amount of income to be obtained if the use of the currently used loan service is continued and repayment is made as planned and an amount of income to be obtained if repayment is made as planned after changing from the currently used loan service satisfies a predetermined difference condition.
The loan management system according to claim 2 . The adjusted income amount is an expected value of the income amount obtained from the loan service in the case where a default occurs with an occurrence probability based on the user information of the target user.
The loan management system according to claim 2 . The adjusted income amount is an expected value of the total income amount to be obtained from the changed loan service and the additional service, assuming that a default occurs with a probability based on the user information of the target user, and assuming that the target user changes the loan service and then uses another service as an additional income amount.
The loan management system according to claim 2 . The recommended loan service determination unit determines a recommendation order for the recommended loan services based on the adjusted income amount and the expected adoption rate.
The loan management system according to claim 2 . The learning unit further performs machine learning on learning data including data on at least some of the attributes and behaviors of a plurality of users and whether or not the plurality of users have defaulted, constructs a default risk determination model for predicting a probability of default , and appropriately updates the default risk determination model and the fit determination model .
The recommended loan service determination unit receives as input at least a portion of the user information of the target user, calculates the expected value of the income amount based on the probability of default predicted by a latest default risk determination model, and sets it as the adjusted income amount; and receives as input at least a portion of the user information of the target user and the service content of the recommendable loan service , and predicts the adoption expectation degree by a latest fit determination model.
The loan management system according to claim 4. The default risk determination model is a prediction model for predicting the occurrence probability of a default for each time point of occurrence, the prediction model being constructed by machine learning learning data including data on the attributes and behaviors of at least a portion of the users and the presence or absence and time point of default of the users,
the recommended loan service determination unit inputs at least a portion of the user information of the target user and calculates the expected income amount based on the time and probability of default predicted by a latest default risk assessment model;
The loan management system according to claim 7. The learning unit further performs machine learning on learning data including data on the attributes and behavior of at least a portion of the users and the presence or absence and time point of prepayment, and constructs and appropriately updates a prepayment risk determination model for predicting the occurrence probability of prepayment at each time point of occurrence;
The recommended loan service determination unit inputs at least a portion of the user information of the target user, and calculates the expected value of the income amount based on the timing and probability of default predicted by the latest default risk assessment model, and the probability and time of early repayment predicted by the latest prepayment risk assessment model.
The loan management system according to claim 8. The learning unit further performs machine learning on learning data including attribute and behavior data of the multiple users and income amounts from other services used by the multiple users after the loan service is changed, and constructs and appropriately updates a future profitability evaluation model for predicting the amount of income from additionally using other services after the loan service is changed;
The recommended loan service determination unit inputs at least a portion of the user information of the target user, and calculates an expected value of the income amount based on the probability of default predicted by the latest default risk determination model and the income amount predicted by the latest future profitability evaluation model.
The loan management system according to claim 7. The user's actions include actions regarding changes to loan services.
The loan management system according to claim 7. For each of a plurality of users, user information that records the attributes and behavior of the user and service contents including an interest rate, a repayment time, and a repayment amount for a currently used loan service that is a loan service being used are managed;
acquiring a change request for a target user among the plurality of users to change the currently used loan service to another loan service;
For the target user, one or more loan services that are estimated to generate a predetermined profit based on the change request, the user information, and the service content of the loan service being used are set as recommendable loan services, an adoption expectation degree, which is a degree of likelihood that the recommendable loan services will be adopted by the target user, is calculated for the recommendable loan services, and one or more of the recommendable loan services are set as recommended loan services to be recommended to the target user based on the adoption expectation degree;
machine learning learning data including data on the attributes and behavior of at least a portion of a plurality of users and information on whether the loan service recommended to the plurality of users has been adopted, and constructing a fit determination model for predicting the probability that the loan service recommended to the user will be adopted as the adoption expectation degree;
predicting the expected adoption rate of the recommendable loan service using the fit determination model by inputting at least a part of the user information of the target user and the service content of the recommendable loan service;
A loan management method implemented by a computer. For each of a plurality of users, user information that records the attributes and behavior of the user and service contents including an interest rate, a repayment time, and a repayment amount for a currently used loan service that is a loan service being used are managed;
acquiring a change request for changing a target user among the plurality of users from the currently used loan service to another loan service;
For the target user, one or more loan services that are estimated to generate a predetermined profit based on the change request, the user information, and the service content of the loan service being used are set as recommendable loan services, an adoption expectation degree, which is a degree of likelihood that the recommendable loan services will be adopted by the target user, is calculated for the recommendable loan services, and one or more of the recommendable loan services are set as recommended loan services to be recommended to the target user based on the adoption expectation degree;
machine learning learning data including data on the attributes and behavior of at least a portion of a plurality of users and information on whether the loan service recommended to the plurality of users has been adopted, and constructing a fit determination model for predicting the probability that the loan service recommended to the user will be adopted as the adoption expectation degree;
predicting the expected adoption rate of the recommendable loan service using the fit determination model by inputting at least a part of the user information of the target user and the service content of the recommendable loan service;
A loan management program that causes a computer to do this.

Images