TWI709923B - Behavioral model credit assessment system - Google Patents

Abstract

A behavioral model credit assessment system has a communication module, a microprocessor, a scoring generation module, and a verification module. The scoring generation module is configured to generate a total credit score according to a first parameter corresponding to at least one sub-factor parameter of the at least one factor parameter of a corresponding evaluated object received by the communication module and a weight corresponding to the at least one factor parameter. The verification module is configured to obtain, according to the plurality of overdue data corresponding to the plurality of evaluated objects, updated first parameters of the at least one sub-factor parameters corresponding to overdue data. The verification module further obtains an update weight and a high-risk interval corresponding to at least one factor parameter of the over-discharge data according to the updated first parameter of the sub-factor parameter corresponding to overdue data, wherein when the total credit score is located in the high-risk interval, the evaluated object is a high-risk object.

Description

Behavior Model Credit Rating System

The present invention relates to a credit rating system, in particular to a behavior model credit rating system that can perform credit evaluation through behavior models.

When ordinary people buy goods with a high unit price such as houses or cars, they usually cannot easily pay off their burdens in a single payment. Therefore, in order to facilitate the people to purchase similar products, financial institutions have proposed loan services to meet related purchase needs.

When people want to use financial institutions to make loans to purchase high-priced goods, in order to minimize the occurrence of overpayment (not repaying on time) by the people who become customers, conventional financial institutions will first grant credit 5P (borrower, use of funds, The principle of repayment source, debt protection and credit outlook) is used to verify the loan, determine whether to approve the loan application of the customer or determine the final loan amount, so as to reduce the probability of overpayment by the customer.

It is still difficult for conventional financial institutions to reduce the probability of a customer's overpayment by the 5P principle of credit granting. Therefore, the present invention proposes a behavior model credit evaluation system, which updates the correspondence between the customer’s overrun situation and its behavior pattern in real time. Effectively use a variety of factor parameters associated with the customer's behavior to determine the probability of a customer's overpayment, thereby effectively reducing the probability of an overpayment after the loan is approved.

In order to achieve the above and other purposes, this authoring department provides a behavior model credit rating system, which includes a communication module, a microprocessor, a score generation module, and a verification module. The score generation module is electrically connected to the microprocessor for generating a total credit score according to the first parameter and the weight, where the first parameter corresponds to at least one sub-factor parameter of the at least one factor parameter associated with the assessed object and is determined by The weight is received by the communication module and corresponds to at least one factor parameter. The verification module is electrically connected to the microprocessor to obtain at least one updated first parameter based on a plurality of overrun data, and obtain at least one update weight with at least one updated first parameter, and then obtain a high risk with at least one update weight The interval, wherein each of the overrun data individually corresponds to one of the multiple evaluated objects and is received by the communication module, and at least one updated first parameter corresponds to at least one of the multiple overrun data The factor parameter, at least one update weight corresponding to the at least one factor parameter associated with the plurality of overrelease data, wherein when the total credit score is within the high-risk interval, the assessed object is a high-risk object.

In an embodiment, the verification module obtains a plurality of update weights according to a plurality of updated first parameters, and obtains a plurality of main parameter values individually by the plurality of update weights and the plurality of updated first parameters, wherein the The multiple updated first parameters and the multiple main parameter values individually correspond to one of the multiple evaluated objects.

In one embodiment, the over-disbursement data is the monthly outstanding information corresponding to a loan item.

In one embodiment, the first parameter represents the probability that the sub-factor parameter corresponds to the occurrence of the overshoot situation.

In one embodiment, the high-risk interval is between 0.4 and 1.6 times a maximum main parameter value, where the maximum main parameter value is the maximum value of the main parameter values.

In one embodiment, the behavior model credit rating system further includes a weight setting module, which is electrically connected to the microprocessor, for adjusting the weight according to the update weight, and for adjusting the first parameter according to the update first parameter , Wherein the update weight and the weight correspond to the same factor parameter, and the update first parameter and the first parameter correspond to the same sub-factor parameter.

In one embodiment, the weight setting module is used to set the initial weight, the first parameter, and the maximum main parameter value.

In one embodiment, the behavior model credit rating system further includes: a factor parameter determination module electrically connected to the microprocessor, and the factor parameter determination module is used to set at least one factor parameter used for calculating the total credit score.

In an embodiment, the at least one sub-factor parameter is at least one factor parameter selected by the subject.

In an embodiment, the factor parameter is at least one of color, random question, operating habits, consumer electronic product acceptance, community activity, financial relationship, or digital footprint.

As a result, with the behavioral pattern credit rating system proposed by the present invention, the total credit score of the rated object can be quickly obtained according to the factor parameters associated with or corresponding to the rated object and its sub-factor parameters, so that the rated object can be quickly determined Whether it is a high-risk object (the probability of over-release is relatively high), it can be used to determine whether to approve the subject’s loan application or adjust its loan amount. In addition, the accuracy of credit scoring can be optimized by the overpayment data generated after the loan is approved, thereby effectively reducing the occurrence of overpayment.

10: Behavioral Model Credit Rating System Server

20: Internet

30, 40: External server

110: Microprocessor

120: Communication module

130: Factor parameter determination module

140: Score generation module

150: verification module

160: Weight setting module

170: database

FIG. 1 is a schematic diagram of a use environment of a behavioral model credit rating system according to an embodiment of the present invention; FIG. 2 is a schematic diagram of a behavioral model credit rating system server according to an embodiment of the present invention; and FIG. 3 is a schematic diagram of over-released data according to an embodiment of the invention.

In order to fully understand the purpose, features and effects of the present invention, the following specific embodiments are used in conjunction with the accompanying drawings to give a detailed description of the present invention. The description is as follows: please refer to Figure 1, which is A schematic diagram of an embodiment of the use environment of the behavior mode credit rating system of the invention, wherein the behavior mode credit rating system can be implemented by the behavior mode credit rating system server, and the present invention is not limited thereto. FIG. 1 includes a behavior model credit rating system server 10, a network 20, and external servers 30 and 40. The external server 30 is used to provide information on factor parameters including at least one sub-factor parameter, and the factor parameter is associated with a subject to be evaluated. The external server 30 can be analyzed by an online banking system server or a financial giant At least one of the application platform server and the data storage server is implemented, and the present invention is not limited to this. In one embodiment, the subject of evaluation is, for example, a customer who requests a loan from a financial institution, and the external server 30 may be a server to which the financial institution belongs, and the present invention is not limited thereto.

The factor parameter is, for example, color, and its sub-factor parameters can be different colors such as red, orange, yellow, green, blue, black, etc. The personality of the assessed object can be determined by selecting different sub-factor parameters . For example, a subject who chooses red can be judged to have a more lively and enthusiastic personality than a subject who chooses blue. The factor parameter is, for example, a random question, and its sub-factor parameter can be used to express the degree of lies of the assessed object, and is represented by high, medium, and low. For example, the factor parameter that is a random question can be a question with objective facts, such as the age of the subject. Therefore, the degree of lies of the subject can be judged by the answers of the subject. The factor parameters are, for example, operating habits, and its sub-factor parameters can be used to express the state of the assessed object during operation (for example, operating a mouse to click on the corresponding webpage), and are expressed in terms of slow, steady and urgent. By selecting different sub-factor parameters, The personality of the assessed object can be determined. For example, when the operating state of the assessed object is judged as a "slow" sub-factor parameter, it can mean that the personality of the assessed object is relatively mild or slow in response. The factor parameter is, for example, the acceptance of consumer electronic products, and its sub-factor parameter can be represented by a numerical value (for example, 0-10). The factor parameter is, for example, the activity of the community, which can be obtained by the number of texts posted or the number of uses of the social software by the subject, and the sub-factor parameter can be represented by a numerical value (for example, 0-10). The factor parameter is, for example, financial relationship (for example, the number of transactions or assets of the assessed object), and its sub-factor parameter can be represented by a value (for example, 0~10). The higher the value, the higher the number of transactions or the higher the assets. The financial relationship between institutions is closer. The factor parameter is, for example, a digital footprint (the number of times an electronic device is used to conduct transactions or the number of online banking operations), and its sub-factor parameter can be represented by a numerical value (for example, 0~10). The higher the numerical value, the electronic device The higher the number of transactions or operations of online banking. In other words, the factor parameter can be at least one of color, random question, operating habits, consumer electronic product acceptance, community activity, financial relationship, or digital footprint.

For example, when the external server 30 is implemented as an online banking system server, the external server 30 can collect sub-factors corresponding to different factors by operating a sub-factor information collection interface (such as a web page) Factor parameter. Furthermore, the assessed object can operate the sub-factor information collection interface to respond to a sub-factor collection questionnaire (for example, a questionnaire designed with different sub-factor parameters), so that the online banking system server can target the sub-factors by the assessed object The questionnaire answers provided by the factor collection questionnaire collect different sub-factor parameters. For example, the red sub-factor parameters are collected, the sub-factor parameters corresponding to the degree of lies and are "high" are collected, and the collected sub-factor parameters are not limited by this. In addition, the information of the sub-factor parameters corresponding to the operating habit factor parameters can be collected through the state of the operating sub-factor information collection interface of the assessed object (such as operating speed). At the same time, the sub-factor parameters corresponding to the digital footprint can be collected by collecting the number of times the subject has operated the online banking, and the present invention is not limited by this.

For example, when the external server 30 is an embodiment of the financial massive analysis and application platform server, the external server 30 may generate and provide the factor parameter as the sub-factor parameter of the acceptance of consumer electronic products, and the factor parameter is the social The sub-factor parameter of the group activity brightness, and the present invention is not limited thereto.

For example, when the external server 30 is an embodiment of a data storage server, the external server 30 may generate and provide the factor parameter as the sub-factor parameter of the financial relationship, and the present invention is not limited thereto.

In the embodiment of FIG. 1, only one external server 30 is taken as an example, but the present invention is not limited thereto. In other embodiments, there may be multiple external servers 30 according to the requirements of the sub-factor parameters.

In one embodiment, the external server 40 can be implemented by an overrun operation management system, and the invention is not limited thereto. In this embodiment, the external server 40 can be used to provide at least one piece of overpayment data corresponding to the assessed object, wherein the overrun data is the monthly outstanding information of the assessed object corresponding to a loan item. For example, the sub-factor parameters of the over-release data can include the first-month over-release information, the second-month over-release information and other information, where the first-month over-release information and the second-month over-release information are Information about the overdue release of the subject in the first few months after the loan was approved. In other embodiments, the overrun data can also be used as factor parameters, and the present invention is not limited thereto.

Therefore, the behavioral model credit rating system server 10 can be used to communicate with the external servers 30 and 40 through the network 20 (intranet or Internet), so as to communicate with the assessed object through the sub-factor parameter information of the factor parameter. Perform scoring and generate a corresponding total credit score, and when the total credit score is within a high-risk range, the subject can be judged as a high-risk subject. At the same time, the behavioral model credit rating system server 10 can also pass The data optimizes the accuracy of the total credit score.

Please refer to FIG. 2, which is a schematic diagram of an embodiment of the behavioral pattern credit rating system server 10. The behavior model credit rating system server 10 includes a communication module 120, a microprocessor 110, and a factor parameter determination model. Group 130, score generation module 140, verification module 150, weight setting module 160, and database 170, wherein the microprocessor 110 and the communication module 120, the factor parameter determination module 130, the score generation module 140, the verification module The group 150, the weight setting module 160, and the database 170 are electrically connected for data exchange and processing related operations.

The communication module 120 can be connected to the network 20 in a wired or wireless manner to receive the aforementioned at least one sub-factor parameter corresponding to a factor parameter and the aforementioned overrun data. The communication module 120 can be an Ethernet network. , Optical fiber network or wireless local area network communication circuit, and the present invention is not limited by this.

The factor parameter determination module 130 is used to generate a factor parameter determination interface, so an operator (for example, a loan professional) can determine and set the score according to the factor parameters provided by the external server 30 and multiple external servers 30. The generating module 140 generates at least one factor parameter referred to or used by the total credit score. In other words, the factor parameter can be set to one or more (for example, two or five) according to requirements, and the present invention is not limited thereto.

The score generating module 140 is used for generating a total credit according to a first parameter corresponding to at least one sub-factor parameter of at least one factor parameter corresponding to the at least one factor parameter of the subject and a weight corresponding to the at least one factor parameter received by the communication module 120 score.

For example, taking the factor parameter as color, which includes sub-factor parameters such as red, orange, yellow, green, and blue, as an example, when the sub-factor parameter corresponding to the assessed object is red, the first parameter corresponding to red is 120. And the factor parameter is that the weight corresponding to the color is 0.51, so the score generating module 140 can obtain a total credit score of 61.2 (120*0.51=61.2). In other embodiments, when the score generating module 140 uses more than one factor parameter, the score generating module 140 obtains the credit score corresponding to each factor parameter individually, and then adds the multiple credit scores to obtain the corresponding credit score. The total credit score of the subject.

The verification module 150 is used for obtaining an updated first parameter corresponding to at least one sub-factor parameter of the overriding data according to a plurality of overriding data individually corresponding to a plurality of evaluated objects. The verification module 150 updates the first parameter of at least one sub-factor parameter corresponding to the multiple overrun data to obtain an update weight and high-risk interval corresponding to at least one factor parameter of the multiple overrun data. The updated first parameter may represent the relative probability of the corresponding sub-factor parameter corresponding to the occurrence of the overshoot situation.

The following takes the factor parameter as the color, which includes sub-factor parameters such as red, orange, yellow, green, and blue, and ten overrun data as examples to illustrate the embodiment of the first parameter, where the overrun data is the first month The overrun information is taken as an example, and the present invention is not limited thereto. The verification module 150 judges that among the ten overrun data, six overrun data correspond to red, two overrun data correspond to green, two overrun data correspond to orange, and the probability that red corresponds to overrun data is 60. The probability that green corresponds to overrun data is 20%, the probability that orange corresponds to overrun data is 20%, and the probability that yellow and blue correspond to overrun data is 0%. Therefore, when the factor parameter is the color corresponding to the total score of 200, the verification module 150 generates the updated first parameter (200*0.6=120) corresponding to the red sub-factor parameter, which also represents the occurrence of the corresponding overshoot situation. Relatively high probability), corresponding to the updated first parameter of the green and orange sub-factor parameters (200*0.2=40, which also means that the probability of the corresponding overshoot is relatively middle), corresponding to the yellow and blue sub-factor parameters Update the first parameter of (200*0=0, which also means that the probability of corresponding overshoot is relatively low).

The verification module 150 can further obtain an update weight corresponding to each factor parameter and a plurality of main parameter values corresponding to the plurality of evaluated objects according to the updated first parameters individually corresponding to the plurality of evaluated objects. In one embodiment, the verification module 150 can obtain the update weight corresponding to each factor parameter and the main parameter value corresponding to at least one parameter combination by the principal component analysis method. The main parameter value can represent the relative probability corresponding to the parameter combination (specific overshoot situation), in other words, the largest main parameter with the maximum value The value can represent the highest relative probability of an overrelease. Therefore, if the credit score of the evaluated object is closer to the maximum main parameter value, the evaluated object can be judged as a high-risk object. In other embodiments, if the credit score of the assessed object falls within the high-risk range, it is determined that the assessed object is a high-risk object, where the high-risk range is approximately 0.4 to 1.6 times the maximum main parameter value, and the Creation is not limited by this.

The following takes FIG. 3 as an example to further illustrate the operation method of the verification module 150, and the present invention is not limited thereto. Please refer to FIG. 3, which is an example of over-disbursement data (for example, the first-month over-disbursement information) generated by the assessed objects A to E after lending. The verification module 150 can individually obtain the score of each sub-factor parameter according to the multiple sub-factor parameters corresponding to the assessed object and the total score corresponding to each factor parameter, and this score corresponds to the aforementioned updated first parameter. Then, the verification module 150 generates multiple main parameter values U _A , U _B , U _C , U _D , and U _E corresponding to the assessed object, where U _A is 120a+60b+80c, and U _B is 40a+60b+ 80c, U _C is 40a+180b+120c, U _D is 120a+180b+120c, U _E is 120a+180b+120c, a, b, and c respectively represent the update of color, degree of lies and acceptance of consumer electronic products Weights. In other words, the main parameter values U _A , U _B , U _C , U _D , and U _{E respectively} corresponding to the evaluated object are the product combination of the updated first parameter and the corresponding update weight corresponding to the sub-factor parameter corresponding to the evaluated object . The verification module 150 can perform calculations on the main parameter values U _A , U _B , U _C , U _D , and U _E by the principal component analysis method to obtain the values of a, b, and c individually, and then obtain the main parameter values U _A individually , U _B , U _C , U _D , and U _E values, and get the maximum main parameter value and the corresponding high-risk interval.

The weight setting module 160 is used to update (or adjust) the weight corresponding to the same factor parameter according to the update weight corresponding to one of the factor parameters and generated by the verification module 150, and use the update weight generated by the verification module 150 One parameter is used to update (or adjust) the first parameter corresponding to the same sub-factor parameter. In one embodiment, the weight setting module 160 can further be used to set the initial weight, the first parameter, and the maximum main parameter value.

The database 170 is used to store information or data required by the behavioral model credit rating system, such as the aforementioned at least one sub-factor parameter and overrun data corresponding to the evaluated object, and the present invention is not limited thereto.

Therefore, when a subject wants to make a loan or a loan, the lending institution (such as a financial institution) can collect at least one sub-factor parameter corresponding to the subject through the above-mentioned behavior pattern credit rating system server 10, and then At least one sub-factor parameter corresponding to the assessed object generates a total credit score corresponding to the assessed object, and judges whether the assessed object is a high-risk object by judging whether the total credit score falls within the high-risk range (the probability of over-release is relatively high) Object).

In summary, with the behavioral pattern credit rating system of the present invention, the total credit score of the rated object can be quickly obtained based on the factor parameters associated with or corresponding to the rated object and its sub-factor parameters, and the rating can be quickly determined Whether the object is a high-risk object (with a relatively high probability of over-release), it can be used to determine whether to approve the subject’s loan application or adjust its loan amount. In addition, the verification module 150 can update at any time and adjust the weight corresponding to the factor parameter and the first parameter corresponding to each sub-factor parameter by using the overpayment data generated after the loan is verified to optimize the accuracy of the total credit score. , Thereby effectively reducing the occurrence of over-release.

The present invention has been disclosed in a preferred embodiment above, but those skilled in the art should understand that the embodiment is only used to describe the present invention and should not be construed as limiting the scope of the present invention. It should be noted that all changes and substitutions equivalent to this embodiment should be included in the scope of the present invention. Therefore, the protection scope of the present invention should be defined by the scope of the patent application.

10: Behavioral Model Credit Rating System Server

110: Microprocessor

120: Communication module

130: Factor parameter determination module

140: Score generation module

150: verification module

160: Weight setting module

170: database

Claims (9)

A behavior model credit rating system, which includes: a communication module; a microprocessor electrically connected to the communication module; a score generating module electrically connected to the microprocessor for according to a first parameter and A weight generates a total credit score, wherein the first parameter corresponds to at least one sub-factor parameter of at least one factor parameter associated with a subject and is received by the communication module, and the weight corresponds to the at least one factor parameter. A factor parameter; and a verification module electrically connected to the microprocessor for obtaining at least one updated first parameter based on a plurality of overrun data, and obtaining at least one update weight based on the at least one updated first parameter, and A high-risk interval is obtained by using the at least one update weight, wherein each of the overrun data individually corresponds to one of the multiple evaluated objects and is received by the communication module, and the at least one update first parameter corresponds to the Wait for at least one sub-factor parameter associated with a plurality of overdraft data, and the at least one update weight corresponds to at least one factor parameter associated with the plurality of overdraft data; wherein when the total credit score is within the high risk interval When the evaluation object is a high-risk object, the verification module obtains multiple update weights according to multiple update first parameters, and obtains multiple update weights and multiple update first parameters individually A plurality of main parameter values, wherein the plurality of updated first parameters and the plurality of main parameter values individually correspond to one of the plurality of evaluated objects. The behavioral model credit rating system according to claim 1, wherein the high-risk interval is between 0.4 and 1.6 times a maximum main parameter value, and the maximum main parameter value is the maximum value of the plurality of main parameter values . For example, in the behavioral model credit rating system described in claim 1, each of the overdue data is the monthly outstanding information of a loan item corresponding to each of the assessed objects. For the behavioral model credit rating system described in claim 1, the first parameter represents the probability that the sub-factor parameter corresponds to the occurrence of the overrun situation. For the behavioral model credit evaluation system of claim 1, the behavioral model credit evaluation system further includes: a weight setting module, electrically connected to the microprocessor, for adjusting the weight according to the update weight, and for adjusting the weight according to the update weight Update the first parameter to adjust the first parameter, wherein the update weight and the weight correspond to the same factor parameter, and the update first parameter and the first parameter correspond to the same sub-factor parameter. The behavioral model credit rating system according to claim 5, wherein the weight setting module is used to set the initial weight, the first parameter, and the maximum main parameter value. For the behavioral model credit rating system of claim 1, the behavioral model credit rating system further includes: a factor parameter determination module electrically connected to the microprocessor, and the factor parameter determination module is configured to calculate the total credit At least one factor parameter used for scoring. The behavioral pattern credit evaluation system according to claim 1, wherein the at least one sub-factor parameter is at least one factor parameter selected by the subject. The behavioral model credit rating system according to claim 1, wherein the factor parameter is a color, a random question, an operating habit, a consumer electronic product acceptance, a community activity, a financial relationship, or a digital footprint At least one of them.

Images