CN111861738A - Risk control rule screening method, device, computer equipment and storage medium - Google Patents

Abstract

The present invention provides a method, device, computer equipment and storage medium for screening risk control rules. The method includes: naming the variable names of each rule of the credit evaluation model one by one to be consistent with the name of each rule; The rule variable is used as the independent variable, and the post-loan performance variable is used as the dependent variable to generate a one-dimensional vector; for each entry of the credit evaluation model, a new one-dimensional vector is generated; if one of the entries of the credit evaluation model is hit or For multiple rules, assign 1 to the hit rule variable in the new one-dimensional vector, and assign 0 to the miss variable; if the incoming loan is overdue, assign 1 to the post-loan performance variable in the new one-dimensional vector, otherwise assign 0 ; Using the logistic regression model, the weight parameters of each rule of the credit evaluation model are calculated, and the rules whose weight parameters are less than the preset value are eliminated from the credit evaluation model. The present invention can more effectively screen out the really effective risk control rules.

Description

Risk control rule screening method, device, computer equipment and storage medium

technical field

The present invention relates to a rule screening method, device, computer equipment and storage medium, in particular to a risk control rule screening method, device, computer equipment and storage medium.

Background technique

Risk control capability is the core competitiveness of institutions in the credit industry. The credit industry risk control system generally consists of a rule engine and a scorecard. The rule engine consists of dozens to hundreds of risk control rules, which are used to evaluate the repayment willingness of applicants. The scorecard outputs a score grade, which is used to assess an applicant's ability to repay. After the applicant submits the application, the rules engine will first screen the entries, and the entries that hit any rule in the engine (such as the blacklist) will be rejected directly, and the entries that do not hit the rules will be processed by the scorecard. The scorecard outputs different grades. The low grade means the repayment ability is weak, and the system directly rejects it. The high grade means the repayment ability is strong, and the system directly passes. The middle grade means the repayment ability is uncertain. Check the applicant's information to decide whether to pass or not.

In practice, risk control rules are generally formulated by risk policy personnel. In order to cover fraudulent people as much as possible (that is, to identify people who are not willing to repay), there are often more and more rules and higher maintenance costs. More importantly, the subjectively stacked rules will cause manslaughter for the incoming parts that are drained at a high cost. The risk control system urgently needs a method that can scientifically evaluate the effectiveness of the rules.

At present, there is no simple and efficient method for rule screening in the entire industry. A cumbersome and expensive approach is to decompose the rule engine into multiple subsystems, each with a different rule embedded in it. Randomly import different feeds into each subsystem, and the feeds will not be rejected if they hit the embedded rules of a certain subsystem, but will be processed by the risk control scorecard or manual review of the next process. The validity of the rules is evaluated by counting the post-credit performance of the incoming documents approved by the scorecard or manual credit review, that is, if the incoming documents of a certain subsystem show an abnormal overdue ratio, the relevant rules are determined to be valid rules, otherwise they are invalid. In this method, after the subsystem has passed enough entries, the determination of its embedded rules has statistical significance. The demand for test entries is large, and the more rules, the greater the demand for test entries. . In short, this method requires a large number of test inputs, and will release many potentially fraudulent inputs that could have been rejected by the rule engine, which will have a great impact on performance.

SUMMARY OF THE INVENTION

The technical problem to be solved by the present invention is to provide a more effective screening method, device, computer equipment and storage medium for risk control rules.

In order to solve the above-mentioned technical problems, the technical solution adopted in the present invention is: a method for screening risk control rules, comprising the following steps:

S10. Name the variable names of each rule of the credit evaluation model to be consistent with the name of each rule, and use each rule variable as an independent variable and the post-loan performance variable as a dependent variable to generate a one-dimensional vector;

S20, generating a new one-dimensional vector for each input condition of the credit evaluation model;

S30. If one or more rules are hit in the input of the credit evaluation model, assign 1 to the hit rule variable in the new one-dimensional vector, and assign 0 to the miss variable;

S40. If the incoming loan is overdue, assign 1 to the post-loan performance variable in the new one-dimensional vector, otherwise assign 0;

S50, using the logistic regression model to calculate the weight parameters of each rule of the credit evaluation model;

S60. According to the weight parameters of each rule of the credit evaluation model, remove the rules whose weight parameters are smaller than the preset value from the credit evaluation model.

Further, in the step S50, the logistic regression model is preset with a loss function of the L1 regularization term.

Further, the step S50 specifically includes:

Using the sparse feature of the L1 regularization term of the logistic regression model, the weight parameters of each rule of the credit evaluation model are calculated.

Further, the step S60 specifically includes:

According to the weight parameters of each rule of the credit evaluation model, rules with a weight parameter less than 0.01 are eliminated from the credit evaluation model.

The present invention also provides a screening device for wind control rules, comprising:

The variable generation module is used to name the variable names of each rule of the credit evaluation model to be consistent with the name of each rule, and use each rule variable as an independent variable and the post-loan performance variable as a dependent variable to generate a one-dimensional vector;

The entry module is used to generate a new one-dimensional vector for each entry of the credit evaluation model;

The rule variable assignment module is used to assign 1 to the hit rule variable and 0 to the miss variable in the new one-dimensional vector if one or more rules are hit in the input of the credit evaluation model;

The post-loan performance variable assignment module is used to assign a value of 1 to the post-loan performance variable in a new one-dimensional vector if the incoming loan is overdue, otherwise it will be assigned a value of 0;

The logistic regression module is used to calculate the weight parameters of each rule of the credit evaluation model by using the logistic regression model;

The rule elimination module is used for eliminating the rules whose weight parameters are smaller than the preset value from the credit evaluation model according to the weight parameters of each rule of the credit evaluation model.

Further, in the logistic regression module, the logistic regression model is preset with a loss function of the L1 regularization term.

Further, the logistic regression module is specifically used for,

Using the sparse feature of the L1 regularization term of the logistic regression model, the weight parameters of each rule of the credit evaluation model are calculated.

Further, the rule elimination module is specifically used for,

According to the weight parameters of each rule of the credit evaluation model, rules with a weight parameter less than 0.01 are eliminated from the credit evaluation model.

The present invention also provides a computer device, the computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned method for screening risk control rules when executing the computer program.

The present invention also provides a storage medium, where the storage medium stores a computer program, and when the computer program is executed by the processor, the above-mentioned method for screening risk control rules can be implemented.

The beneficial effects of the present invention are: the present invention considers the evaluation and screening tasks of risk control rules as a logistic regression model, and utilizes the L1 regularization and sparseness feature of the objective function of the model to solve the weights of the rule variables. The weight of important rules is automatically iterated to infinitely close to zero, which simplifies the rule evaluation process. Rules with weight parameters less than the preset value are eliminated from the credit evaluation model, leaving the rules that really work, avoiding the unimportant hit due to incoming items. This method greatly reduces the amount of test input required and reduces the potential overdue cost.

Description of drawings

The specific structure of the present invention will be described in detail below with reference to the accompanying drawings.

1 is a flowchart of a method for screening risk control rules according to an embodiment of the present invention;

2 is a block diagram of an apparatus for screening wind control rules according to an embodiment of the present invention;

FIG. 3 is a schematic block diagram of a computer device according to an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

It is to be understood that, when used in this specification and the appended claims, the terms "comprising" and "comprising" indicate the presence of the described features, integers, steps, operations, elements and/or components, but do not exclude one or The presence or addition of a number of other features, integers, steps, operations, elements, components, and/or sets thereof.

It is also to be understood that the terminology used in this specification of the present invention is for the purpose of describing particular embodiments only and is not intended to limit the present invention. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural unless the context clearly dictates otherwise.

It should further be understood that, as used in this specification and the appended claims, the term "and/or" refers to and including any and all possible combinations of one or more of the associated listed items .

Referring to FIG. 1, the first embodiment of the present invention is: a method for screening risk control rules, including the following steps:

S10. Name the variable names of each rule of the credit evaluation model to be consistent with the name of each rule, and use each rule variable as an independent variable and the post-loan performance variable as a dependent variable to generate a one-dimensional vector;

S20, generating a new one-dimensional vector for each input condition of the credit evaluation model;

S30. If one or more rules are hit in the input of the credit evaluation model, assign 1 to the hit rule variable in the new one-dimensional vector, and assign 0 to the miss variable;

S40. If the incoming loan is overdue, assign 1 to the post-loan performance variable in the new one-dimensional vector, otherwise assign 0;

S50, using the logistic regression model to calculate the weight parameters of each rule of the credit evaluation model;

S60. According to the weight parameters of each rule of the credit evaluation model, remove the rules whose weight parameters are smaller than the preset value from the credit evaluation model.

Further, in the step S50, the logistic regression model is preset with a loss function of the L1 regularization term.

Further, the step S50 specifically includes:

Using the sparse feature of the L1 regularization term of the logistic regression model, the weight parameters of each rule of the credit evaluation model are calculated.

Further, the step S60 specifically includes:

According to the weight parameters of each rule of the credit evaluation model, rules with a weight parameter less than 0.01 are eliminated from the credit evaluation model.

In this embodiment, the evaluation and screening tasks of risk control rules are considered as a logistic regression model, and the L1 regularization and sparse features of the objective function of the model are used to calculate the weights of unimportant rules in the process of solving the weights of rule variables. The automatic iteration is infinitely close to zero, which simplifies the rule evaluation process, removes the rules whose weight parameters are less than the preset value from the credit evaluation model, leaves the rules that really work, and avoids the unimportant rules that are hit by the entry. Rejected by the credit evaluation model, this method greatly reduces the amount of test input required and reduces potential overdue costs.

The above embodiment can solve the problem of evaluating the validity of various rules in the credit evaluation model in the industry. For example, a certain rule in a credit evaluation model rejects an entry. Since the entry has no post-loan performance, we There is no way to know whether the rule is valid, that is, it is impossible to judge whether the entry has been killed by mistake. The more rules a credit evaluation model has, the more entries may be rejected. How to determine which rules reject the entries that should be rejected, and which rules have a high rate of manslaughter? This problem has always been a problem for various credit evaluations in the entire industry For the problem that cannot be solved by the model, the quantitative solution proposed by the embodiment of the present invention solves the problem that there is no simple method to quantify the validity of the rules in the current credit evaluation model through the logistic regression model. Specifically, the independent variable x_i (i represents the i-th rule) of a risk control rule, and the overdue expression after loan is the dependent variable y. When the overdue performance is not overdue, the value of y is 0, and when the overdue performance is overdue, the value of y is 1. In order to quantify the rule evaluation, the rule evaluation problem is transformed into a size problem between the independent variable weights in a logistic regression model. After a period of logistic regression machine learning iteration, the weight parameters of each rule can be obtained, and the rules with a weight less than 0.01 are eliminated from the model to achieve the purpose of rule screening and optimization.

Please refer to FIG. 2, another embodiment of the present invention is: a screening device for risk control rules, comprising:

The variable generation module 10 is used to name the variable names of each rule of the credit evaluation model one by one to be consistent with the name of each rule, and use each rule variable as an independent variable and a post-loan performance variable as a dependent variable to generate a variable name. dimensional vector;

The entry module 20 is used to generate a new one-dimensional vector for each entry of the credit evaluation model;

The rule variable assignment module 30 is used to assign 1 to the hit rule variable in the new one-dimensional vector if one or more rules are hit in the entry of the credit evaluation model, and 0 to the miss variable;

The post-loan performance variable assignment module 40 is used to assign a value of 1 to the post-loan performance variable in a new one-dimensional vector if the incoming loan is overdue, otherwise, assign a value of 0;

The logistic regression module 50 is used to calculate the weight parameters of each rule of the credit evaluation model by using the logistic regression model;

The rule elimination module 60 is used for eliminating the rules whose weight parameter is smaller than the preset value from the credit evaluation model according to the weight parameters of each rule of the credit evaluation model.

Further, in the logistic regression module 50, the logistic regression model is preset with a loss function of the L1 regularization term.

Further, the logistic regression module 50 is specifically used to:

Using the sparse feature of the L1 regularization term of the logistic regression model, the weight parameters of each rule of the credit evaluation model are calculated.

Further, the rule elimination module 60 is specifically used to:

According to the weight parameters of each rule of the credit evaluation model, rules with a weight parameter less than 0.01 are eliminated from the credit evaluation model.

It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned risk control rule screening device may refer to the corresponding description in the foregoing method embodiments. For the convenience and brevity of the description, it will not be repeated here. .

The above-mentioned wind control rule screening apparatus can be implemented in the form of a computer program, and the computer program can be executed on the computer equipment as shown in FIG. 3 .

Please refer to FIG. 3 , which is a schematic block diagram of a computer device provided by an embodiment of the present application. The computer device 500 may be a terminal or a server, wherein the terminal may be an electronic device with communication functions, such as a smart phone, a tablet computer, a notebook computer, a desktop computer, a personal digital assistant, and a wearable device. The server can be an independent server or a server cluster composed of multiple servers.

Referring to FIG. 3 , the computer device 500 includes a processor 502 , a memory and a network interface 505 connected by a system bus 501 , wherein the memory may include a non-volatile storage medium 503 and an internal memory 504 .

The nonvolatile storage medium 503 can store an operating system 5031 and a computer program 5032 . The computer program 5032 includes program instructions, which, when executed, can cause the processor 502 to execute a method for screening wind control rules.

The processor 502 is used to provide computing and control capabilities to support the operation of the entire computer device 500 .

The internal memory 504 provides an environment for running the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a risk control rule screening method.

The network interface 505 is used for network communication with other devices. Those skilled in the art can understand that the structure shown in FIG. 3 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device 500 to which the solution of the present application is applied. The specific computer device 500 may include more or fewer components than shown, or combine certain components, or have a different arrangement of components.

Wherein, the processor 502 is configured to run the computer program 5032 stored in the memory, so as to realize the above method for screening the risk control rules.

It should be understood that, in this embodiment of the present application, the processor 502 may be a central processing unit (Central Processing Unit, CPU), and the processor 502 may also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), dedicated integrated Circuit (Application Specific Integrated Circuit, ASIC), off-the-shelf Programmable Gate Array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. Wherein, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor or the like.

It can be understood by those skilled in the art that all or part of the processes in the methods for implementing the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program includes program instructions, and the computer program can be stored in a storage medium, and the storage medium is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the flow steps of the above-described method embodiments.

Therefore, the present invention also provides a storage medium. The storage medium may be a computer-readable storage medium. The storage medium stores a computer program, wherein the computer program includes program instructions. When the program instructions are executed by the processor, the processor executes the above-mentioned screening method for risk control rules.

The storage medium may be various computer-readable storage media that can store program codes, such as a U disk, a removable hard disk, a read-only memory (Read-Only Memory, ROM), a magnetic disk, or an optical disk.

Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two, in order to clearly illustrate the differences between hardware and software Interchangeability, the above description has generally described the components and steps of each example in terms of function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Skilled artisans may implement the described functionality using different methods for each particular application, but such implementations should not be considered beyond the scope of the present invention.

In the several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is only a logical function division, and other division methods may be used in actual implementation. For example, multiple units or components may be combined or may be integrated into another system, or some features may be omitted, or not implemented.

The steps in the method of the embodiment of the present invention may be adjusted, combined and deleted in sequence according to actual needs. Units in the apparatus of the embodiment of the present invention may be combined, divided, and deleted according to actual needs. In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as an independent product, may be stored in a storage medium. Based on this understanding, the technical solution of the present invention is essentially or a part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , including several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited to this. Any person skilled in the art can easily think of various equivalents within the technical scope disclosed by the present invention. Modifications or substitutions should be included within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims (10)

1. A wind control rule screening method is characterized by comprising the following steps: comprises the following steps of (a) carrying out,

s10, naming the variable names of each rule of the credit evaluation model one by one to be consistent with the names of the rules, taking each rule variable as an independent variable, taking the credit expression variable as a dependent variable, and generating a one-dimensional vector;

s20, generating a new one-dimensional vector for each entry of the credit evaluation model;

s30, if one or more rules are hit in the incoming condition of the credit evaluation model, assigning 1 to the hit rule variable and assigning 0 to the miss variable in the new one-dimensional vector;

s40, if the loan is overdue, assigning 1 to the post-loan expression variable in the new one-dimensional vector, and otherwise, assigning 0;

s50, calculating the weight parameters of each rule of the credit evaluation model by using a logistic regression model;

and S60, according to the weight parameters of each rule of the credit evaluation model, eliminating the rule with the weight parameter smaller than the preset value from the credit evaluation model.

2. The method for screening wind control rules according to claim 1, wherein: in step S50, the logistic regression model is preset with a loss function of the L1 regularization term.

3. The wind control rule screening method according to claim 2, wherein: the step S50 specifically includes the steps of,

and calculating the weight parameters of each rule of the credit evaluation model by using the sparse characteristics of the L1 regularization term of the logistic regression model.

4. The wind control rule screening method according to claim 3, wherein: the step S60 specifically includes the steps of,

and according to the weight parameters of all the rules of the credit evaluation model, removing the rules with the weight parameters less than 0.01 from the credit evaluation model.

5. The utility model provides a wind accuse rule sieving mechanism which characterized in that: comprises the steps of (a) preparing a mixture of a plurality of raw materials,

the variable generation module is used for naming the variable names of the rules of the credit evaluation model one by one to be consistent with the names of the rules, taking the rule variables as independent variables, taking the credited expression variables as dependent variables and generating a one-dimensional vector;

the system comprises a component feeding module, a component extracting module and a component analyzing module, wherein the component feeding module is used for generating a new one-dimensional vector for each component fed by a credit evaluation model;

the rule variable assignment module is used for assigning 1 to a hit rule variable and assigning 0 to a missed rule variable in a new one-dimensional vector if one or more rules are hit in the incoming condition of the credit evaluation model;

the post-loan performance variable assignment module is used for assigning 1 to the post-loan performance variable in a new one-dimensional vector if the post-loan is overdue, and assigning 0 if the post-loan is overdue;

the logistic regression module is used for calculating the weight parameters of all rules of the credit evaluation model by using the logistic regression model;

and the rule removing module is used for removing the rules with the weight parameters smaller than the preset value from the credit evaluation model according to the weight parameters of all the rules of the credit evaluation model.

6. The wind control rule screening apparatus of claim 5, wherein: in the logistic regression module, a logistic regression model is preset with a loss function of an L1 regularization term.

7. The wind control rule screening apparatus of claim 6, wherein: the logistic regression module is specifically configured to,

and calculating the weight parameters of each rule of the credit evaluation model by using the sparse characteristics of the L1 regularization term of the logistic regression model.

8. The wind control rule screening apparatus of claim 7, wherein: the rule culling module is specifically configured to,

and according to the weight parameters of all the rules of the credit evaluation model, removing the rules with the weight parameters less than 0.01 from the credit evaluation model.

9. A computer device, characterized by: the computer device comprises a memory having a computer program stored thereon and a processor that, when executing the computer program, implements the method of wind control rule screening according to any one of claims 1 to 4.

10. A storage medium, characterized by: the storage medium stores a computer program which, when executed by a processor, implements the method of wind control rule screening according to any one of claims 1 to 4.

Images