Risk assessment method and device

Abstract

The invention provides a risk assessment method and device. The method comprises the following steps: acquiring risk control thresholds of a plurality of target attribute variables associated with a target service of a financial institution; wherein the risk control threshold value is determined based on the risk control labels of the plurality of sample objects and the attribute values of the target attribute variables of the plurality of sample objects; obtaining first basic attribute data and first risk control attribute data of a target object of a to-be-executed target service; wherein the first risk control attribute data comprises external risk control data and business risk control data; and performing risk assessment on the target object based on the risk control thresholds of the plurality of target attribute variables, the first basic attribute data and the first risk control attribute data, thereby realizing dynamic adjustment of the risk control thresholds according to actual data, improving accuracy of the risk control thresholds, realizing comprehensive assessment of risk conditions of the target object from multiple angles, and improving risk assessment efficiency. Deviation possibly caused by a single data source is avoided, so that the accuracy of risk assessment is improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of risk management, and in particular to a risk assessment method and device. Background Art

[0002] With the rapid development of the financial industry, the importance of financial risk management has become increasingly prominent. Implementing accurate risk assessment for target objects of financial services such as loans can not only help financial institutions effectively identify and manage potential credit risks, but also optimize resource allocation, improve capital utilization efficiency, and enhance market competitiveness. Therefore, it is very important to accurately conduct risk assessment for target objects of financial services. Summary of the invention

[0003] The present disclosure provides a risk assessment method and device to solve one of the technical problems in the related art at least to a certain extent. The technical solution of the present disclosure is as follows:

[0004] According to a first aspect of an embodiment of the present disclosure, a risk assessment method is provided, comprising: obtaining risk control thresholds of multiple target attribute variables associated with a target business of a financial institution; wherein the risk control thresholds are determined based on risk control labels of multiple sample objects and attribute values ​​of the target attribute variables of the multiple sample objects, and the risk control labels are used to indicate the degree of risk of the sample objects in performing the target business; obtaining first basic attribute data and first risk control attribute data of a target object to perform the target business; wherein the first risk control attribute data includes external risk control data of the target object outside the financial institution and business risk control data within the financial institution; based on the risk control thresholds of the multiple target attribute variables, the first basic attribute data and the first risk control attribute data, performing a risk assessment on the target object to obtain the risk level of the target object in performing the target business.

[0005] According to a second aspect of an embodiment of the present disclosure, a risk assessment device is provided, comprising: a first acquisition module, used to acquire risk control thresholds of multiple target attribute variables associated with a target business of a financial institution; wherein the risk control thresholds are determined based on risk control labels of multiple sample objects and attribute values ​​of the target attribute variables of the multiple sample objects, and the risk control labels are used to indicate the degree of risk of the sample objects in performing the target business; a second acquisition module, used to acquire first basic attribute data and first risk control attribute data of a target object to perform the target business; wherein the first risk control attribute data includes external risk control data of the target object outside the financial institution and business risk control data within the financial institution; an assessment module, used to perform risk assessment on the target object based on the risk control thresholds of the multiple target attribute variables, the first basic attribute data and the first risk control attribute data, so as to obtain the risk level of the target object in performing the target business.

[0006] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement the risk assessment method as described in the embodiment of the first aspect of the present disclosure.

[0007] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided. When instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the risk assessment method as described in the embodiment of the first aspect of the present disclosure.

[0008] According to a fifth aspect of an embodiment of the present disclosure, a computer program product is provided, comprising: a computer program, wherein when the computer program is executed by a processor, the risk assessment method as described in the embodiment of the first aspect of the present disclosure is implemented.

[0009] The technical solution provided by the embodiments of the present disclosure brings at least the following beneficial effects:

[0010] In the present technical scheme, by obtaining risk control thresholds of multiple target attribute variables associated with the target business of the financial institution; wherein the risk control thresholds are determined based on the risk control labels of multiple sample objects and the attribute values ​​of the target attribute variables of the multiple sample objects, and the risk control labels are used to indicate the risk level of the sample objects in executing the target business; obtaining first basic attribute data and first risk control attribute data of the target object to execute the target business; wherein the first risk control attribute data includes the external risk control data of the target object outside the financial institution and the business risk control data within the financial institution; based on the risk control thresholds of multiple target attribute variables, the first basic attribute data and the first risk control attribute data, the target object is risk assessed to obtain the risk level of the target object in executing the target business, thereby determining the risk control threshold based on the risk control labels of the sample objects and the attribute values ​​of the target attribute variables, thereby realizing dynamic adjustment of the risk control threshold according to actual data, improving the accuracy of the risk control threshold, and by combining the basic attribute data of the target object, the external risk control data and the internal business The risk control data of the business is used for risk assessment, which realizes a comprehensive assessment of the risk status of the target object from multiple angles, avoids the deviation that may be caused by a single data source, and thus improves the accuracy of risk assessment. Among them, when determining the risk control threshold, based on the information value of the candidate attribute variable, the most influential target attribute variable is screened out, the attribute value of the target attribute variable is binned, and multiple target bins are obtained. The lift of each target bin is calculated, and the risk control threshold of the corresponding target attribute variable is determined based on the lift of each target bin, which improves the rationality, accuracy and reliability of risk assessment. In addition, when binning the attribute value of the target attribute variable, based on the constraints such as the first risk score mean of the positive sample, the second risk score mean of the negative sample in the target attribute variable, and the bin weight, the objective function for maximizing the risk score gap between the positive and negative sample objects in each bin is solved, and multiple target bins are obtained, which can ensure that the selection of the bin boundary is more reasonable and effectively distinguish high-risk and low-risk objects, thereby improving the risk differentiation ability.

[0011] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute improper limitations on the present disclosure.

[0013] Figure 1 is a flowchart of a risk assessment method shown in the first embodiment of the present disclosure;

[0014] Figure 2 is a flow chart of a risk assessment method according to a second embodiment of the present disclosure;

[0015] Figure 3 is a flowchart of a risk assessment method shown in the third embodiment of the present disclosure;

[0016] Figure 4 is a flowchart of a risk assessment method shown in the fourth embodiment of the present disclosure;

[0017] Figure 5 is a schematic diagram of the structure of a risk assessment device shown in the fifth embodiment of the present disclosure;

[0018] Figure 6 It is a schematic structural diagram of an electronic device shown in an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0019] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings.

[0020] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0021] It should be noted that in the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information are all carried out with the user's consent, and are in compliance with relevant laws and regulations and do not violate public order and good morals.

[0022] At present, risk control strategies mostly rely on manually set strategy rule thresholds and data classification models. For example, the results of the Logistic regression model are used to optimize the risk control strategy rule thresholds; but this model estimates the default probability of the target object (such as the borrower) to assess credit risk; however, the Logistic regression model is not good enough when processing high-dimensional data, and its linear separability assumption may lead to a decrease in prediction accuracy in multidimensional complex data. In addition, the model is sensitive to outliers and multicollinearity in the data, which may affect the stability and accuracy of the model.

[0023] Similarly, the decision tree model is also a common rule-based model that relies on recursively dividing data into different branches and using the decision results of these branches as a reference for adjusting the threshold of risk control strategy rules. Although the decision tree model is intuitive and easy to interpret, it is prone to overfitting, especially when the data is noisy. In addition, the decision tree is highly dependent on feature segmentation and has difficulty handling complex interactions between variables.

[0024] At present, with the development of artificial intelligence and big data analysis technology, more and more financial institutions have begun to use machine learning models (such as random forests, support vector machines, etc.) to optimize and enhance the thresholds of risk control strategy rules. Although these methods have improved the accuracy of predictions to a certain extent, they still have shortcomings in the transparency and explainability of the rules, and it is difficult to meet the financial industry's requirements for the intuitiveness of risk control strategy rules.

[0025] Therefore, in order to solve the above problems, the present disclosure proposes a risk assessment method and device.

[0026] The risk assessment method and device according to the embodiments of the present disclosure are described below with reference to the accompanying drawings.

[0027] Figure 1 It is a flowchart of the risk assessment method shown in the first embodiment of the present disclosure.

[0028] like Figure 1 As shown, the risk assessment method includes the following steps:

[0029] Step 101: Obtain risk control thresholds of multiple target attribute variables associated with a target business of a financial institution.

[0030] Among them, the risk control threshold is determined based on the risk control labels of multiple sample objects and the attribute values ​​of the target attribute variables of the multiple sample objects. The risk control labels are used to indicate the risk level of the sample objects in executing the target business.

[0031] It is important to understand that risk control thresholds are used to divide standards or boundaries into different risk levels. Risk control thresholds can help financial institutions quickly determine the risk level of a potential customer or transaction counterparty, and make corresponding decisions.

[0032] Therefore, in order to improve the accuracy of the risk control threshold, as an example, a large amount of data of sample objects is collected, for example, including the target attribute variables and risk control labels of the sample objects, and the risk control threshold is determined based on the attribute values ​​of the target attribute variables and the risk control labels; wherein the target attribute variables are the attribute variables with greater influence in the data of the sample objects; and the risk control labels are used to indicate the risk level of the sample objects in executing the target business. For example, the value of a customer's risk control label may be 1 or 0, wherein 1 indicates that the customer is a "high-risk customer" and 0 indicates that the customer is a "low-risk" customer.

[0033] Step 102: Acquire first basic attribute data and first risk control attribute data of a target object of a target business to be executed.

[0034] Among them, the first risk control attribute data includes the external risk control data of the target object outside the financial institution and the business risk control data within the financial institution.

[0035] It is important to understand that when financial institutions conduct risk management, they need to conduct risk assessments on potential customers or transaction objects (i.e. target objects) to decide whether to approve loans, credit card applications, investment transactions and other businesses.

[0036] Therefore, in order to conduct risk assessment accurately and comprehensively, as a possible implementation method, it is necessary to obtain data of multiple dimensions (i.e., target attribute variables) associated with the target business of the financial institution, such as the customer's personal information, financial status, historical transaction records, etc., and then conduct risk assessment based on multiple target attribute variables.

[0037] As an example, the first basic attribute data and the first risk control attribute data of the target object for the target business to be executed are obtained; wherein the first basic attribute data include: personal information, financial information, occupational information, etc., and the first risk control attribute data include external risk control data of the target object outside the financial institution and business risk control data within the financial institution; the external risk control data include: the credit record of the target object in other financial institutions or third-party data providers, such as personal credit data, debt repayment ability, operator data, etc.; the business risk control data include: the historical transaction records, repayment records, overdue situations, number of defaults, etc. of the target object in the financial institution.

[0038] Step 103: Based on the risk control thresholds of the multiple target attribute variables, the first basic attribute data and the first risk control attribute data, a risk assessment is performed on the target object to obtain a risk level of the target object in executing the target business.

[0039] In order to accurately conduct risk assessment, as a possible implementation method, by comprehensively analyzing the first basic attribute data of the target object (such as age, income, credit score, etc.) and the first risk control attribute data (such as historical transaction records, external credit reports, etc.), combined with the risk control threshold, the risk level of the target object is evaluated and finally divided into different risk levels (such as low risk, medium risk, high risk).

[0040] As an example, for any target attribute variable, the attribute value of any target attribute variable is obtained from the first basic attribute data and the first risk control attribute data; in response to the attribute value of any target attribute variable being greater than or equal to the risk control threshold of any target attribute variable, it is determined that the risk level of the target object executing the target business is greater than the set risk level.

[0041] That is to say, by obtaining the attribute value of any target attribute variable from the first basic attribute data and the first risk control attribute data, and judging whether the attribute value of any target attribute variable is greater than or equal to the corresponding risk control threshold, when the attribute value of any target attribute variable is greater than or equal to the corresponding risk control threshold, it is determined that the risk level of the target object performing the target business is greater than the set risk level, that is, it is determined that the risk level of the target object performing the target business is high risk.

[0042] As another example, in response to the fact that the attribute values ​​of multiple target attribute variables in the first basic attribute data and the first risk control attribute data are all less than the risk control thresholds of the multiple target attribute variables, it is determined that the risk level of the target object performing the target business is not greater than the set risk level, that is, the risk level of the target object performing the target business is determined to be low risk.

[0043] That is to say, when the attribute values ​​of multiple target attribute variables in the first basic attribute data and the first risk control attribute data are all less than the corresponding risk control thresholds, it is determined that the risk level of the target object executing the target business is not greater than the set risk level.

[0044] In summary, by determining the risk control threshold based on the risk control label of the sample object and the attribute value of the target attribute variable, the risk control threshold can be dynamically adjusted according to the actual data, thereby improving the accuracy of the risk control threshold. In addition, by combining the basic attribute data of the target object, external risk control data and internal business risk control data for risk assessment, a comprehensive assessment of the risk status of the target object from multiple angles can be achieved, avoiding the deviation that may be caused by a single data source, thereby improving the accuracy of risk assessment.

[0045] In order to clearly explain how to obtain the risk control thresholds of multiple target attribute variables associated with the target business of the financial institution in the above embodiment, the present disclosure proposes another risk assessment method.

[0046] Figure 2 It is a flowchart of the risk assessment method shown in the second embodiment of the present disclosure.

[0047] like Figure 2 As shown, the risk assessment method includes the following steps:

[0048] Step 201: Obtain second basic attribute data, second risk control attribute data, and risk control labels of multiple sample objects that execute the target business.

[0049] Among them, the second risk control attribute data includes external risk control data of the corresponding sample object outside the financial institution and business risk control data within the financial institution.

[0050] In order to improve the accuracy of risk assessment, as an example, based on the second basic attribute data, second risk control attribute data and risk control labels of a large number of sample objects, the risk control threshold of the target attribute variable in the second basic attribute data and the second risk control attribute data is determined, so as to perform risk assessment on the target object based on the risk control threshold of the target attribute variable.

[0051] Therefore, the second basic attribute data, second risk control attribute data and risk control labels of multiple sample objects are obtained from the business database. Among them, the second basic attribute data includes the basic information of the sample object, such as age, gender, occupation, etc., and the second risk control attribute data includes the external risk control data of the corresponding sample object outside the financial institution and the business risk control data within the financial institution. The external risk control data of the sample object outside the financial institution includes: the credit record of the sample object in other financial institutions or third-party data providers; the business risk control data of the sample object in the financial institution includes: the historical transaction record, repayment record, overdue situation, number of defaults, etc. of the sample object in the financial institution.

[0052] Step 202: Calculate information values ​​of multiple candidate attribute variables based on the risk control tags.

[0053] Among them, multiple candidate attribute variables are located in the second basic attribute data and second risk control attribute data of multiple sample objects.

[0054] In order to improve the targetedness of risk assessment, as a possible implementation method, target attribute variables with greater influence are screened out from multiple candidate attribute variables in the second basic attribute data and the second risk control attribute data, and risk assessment is performed based on the risk control threshold of the target attribute variable.

[0055] As an example, based on the risk control label, the information values ​​of multiple candidate attribute variables are calculated, so as to screen out the target attribute variable from the multiple candidate attribute variables according to the information values ​​of the multiple candidate attribute variables. The multiple candidate attribute variables are located in the second basic attribute data and the second risk control attribute data of the multiple sample objects, and the multiple candidate attribute variables include, for example, age, credit rating, income, debt, etc.

[0056] The calculation steps of the information values ​​of multiple candidate attribute variables are as follows:

[0057] 1. For any candidate attribute variable, obtain multiple candidate bins of the attribute value of any candidate attribute variable;

[0058] As an example, the attribute value of any candidate attribute variable is binned according to the set binning strategy to obtain multiple candidate bins for the attribute value of any candidate attribute variable. For example, if the candidate attribute variable is age, the attribute value under age is divided into a candidate bin according to the set age interval (5 years old), such as 20-25 years old is one candidate bin, 26-30 years old is another candidate bin, 30-35 years old is another candidate bin, etc.; for another example, if the candidate attribute variable is occupation, it is divided according to the type of occupation, and the occupation in each candidate bin is the same.

[0059] 2. Determine a first number of positive sample objects and a second number of negative sample objects in any candidate sub-bin according to the risk control label of each sample object in any candidate sub-bin;

[0060] That is, based on the risk control label of each sample object in any candidate bin, the first number of positive sample objects and the second number of negative sample objects in any candidate bin are counted. The positive sample object is a sample object with a positive risk control label, and the negative sample object is a sample object with a negative risk control label. The positive label is used to indicate that the risk level is lower than the set threshold, and the negative label is used to indicate that the risk level is not lower than the set threshold.

[0061] 3. Calculate the information sub-value of any candidate bin according to the first quantity and the second quantity;

[0062] As an example, the information sub-value of any candidate bin is determined using the following formula:

[0063]

[0064] Where i is the i-th candidate bin of the attribute value of any candidate attribute variable, good i is the first number of positive sample objects in the i-th candidate bin, bad i is the second number of negative sample objects in the i-th candidate bin, good is the number of positive sample objects in the multiple sample objects, and bad is the number of negative sample objects in the multiple sample objects.

[0065] 4. Determine the information value of any candidate attribute variable based on the information sub-value of each candidate bin.

[0066] As an example, the information sub-values ​​of multiple candidate bins of the attribute value of any candidate attribute variable are added together to obtain the information value of the candidate attribute variable. The specific formula is as follows:

[0067]

[0068] Where n is the number of bins for the attribute values ​​of any candidate attribute variable.

[0069] Step 203: Determine the risk control threshold of the target attribute variable among the multiple candidate attribute variables based on the information values ​​of the multiple candidate attribute variables.

[0070] In order to accurately determine the risk control threshold of each target attribute variable, as a possible implementation method, based on the information values ​​of multiple candidate attribute variables, the target attribute variable is screened out from multiple candidate attribute variables, and the attribute values ​​of the target attribute variable are binned to obtain multiple target bins. Then, the lift of each target bin corresponding to each target attribute variable is calculated. Therefore, based on the lift of each target bin corresponding to each target attribute variable, the risk control threshold of the target attribute variable is determined.

[0071] Step 204: Acquire first basic attribute data and first risk control attribute data of the target object of the target business to be executed.

[0072] Among them, the first risk control attribute data includes the external risk control data of the target object outside the financial institution and the business risk control data within the financial institution.

[0073] Step 205: Based on the risk control thresholds of the multiple target attribute variables, the first basic attribute data and the first risk control attribute data, a risk assessment is performed on the target object to obtain a risk level of the target object in executing the target business.

[0074] It should be noted that the execution process of steps 204 to 205 can be implemented in any manner in the embodiments of the present disclosure, and the embodiments of the present disclosure do not limit this and will not be described in detail.

[0075] In summary, by obtaining the second basic attribute data, second risk control attribute data and risk control labels of multiple sample objects that execute the target business; based on the risk control labels, calculating the information values ​​of multiple candidate attribute variables; based on the information values ​​of multiple candidate attribute variables, determining the risk control threshold of the target attribute variable among the multiple candidate attribute variables, thereby, by calculating the information values ​​of the candidate attribute variables, the importance of each attribute variable to the risk assessment can be quantified, and the most influential target attribute variable can be screened out, thereby determining the risk control threshold of the target attribute variable, and improving the rationality, accuracy and reliability of the risk assessment.

[0076] In any embodiment of the present disclosure, based on the lift of each target bin of the attribute value of the target attribute variable in the multiple candidate attribute variables, the risk control threshold of the target attribute variable is determined. Figure 3 The embodiments are described in detail.

[0077] Figure 3 It is a flowchart of the risk assessment method shown in the third embodiment of the present disclosure.

[0078] like Figure 3As shown, the risk control threshold of the target attribute variable can be determined by following steps 301 to 304:

[0079] Step 301 : screening multiple candidate attribute variables according to their information values ​​to obtain at least one target attribute variable.

[0080] In order to improve the screening efficiency of the target attribute variables, as an example, multiple candidate attribute variables are sorted according to their information values ​​to obtain a candidate sequence; from the candidate sequence, the candidate attribute variables whose information values ​​are lower than the set information threshold are deleted to obtain the target sequence; and the candidate attribute variables in the target sequence are used as the target attribute variables.

[0081] That is to say, according to the calculated information value, these candidate attribute variables are sorted to form an ordered candidate sequence; in the candidate sequence, the candidate attribute variables with higher information values ​​are arranged in front, and those with lower information values ​​are arranged in the back, and a pre-set information threshold is obtained, and the set information threshold is used as a standard for screening candidate attribute variables. Then, the candidate sequence is traversed, and the candidate attribute variables with information values ​​lower than the set threshold are removed from the sequence, so as to obtain a target sequence containing only high information value variables.

[0082] Step 302: for any target attribute variable, the attribute value of any target attribute variable is binned to obtain a plurality of target bins.

[0083] In order to improve the accuracy and rationality of risk assessment, as a possible implementation method, the attribute values ​​of the target attribute variable are binned.

[0084] As an example, based on the first risk score mean of positive samples, the second risk score mean of negative samples, and constraints such as bin weights in the target attribute variable, an objective function for maximizing the risk score gap between positive and negative sample objects in each bin is solved to obtain multiple target bins.

[0085] Step 303: Calculate the lift of the negative sample objects to which the attribute values ​​in each target bin belong according to the third number of positive sample objects and the fourth number of negative sample objects in each target bin.

[0086] Among them, the positive sample object is the sample object with the risk control label as the positive label, and the negative sample object is the sample object with the risk control label as the negative label. The positive label is used to indicate that the risk level is lower than the set threshold, and the negative label is used to indicate that the risk level is not lower than the set threshold.

[0087] In order to quantify the improvement of each target bin in distinguishing positive and negative sample objects relative to the overall sample set, as a possible implementation method, the lift of the negative sample objects to which the attribute values ​​in each target bin belong is calculated.

[0088] As an example, for any target bin, determine the first sum of the fourth number of negative sample objects and the third number of positive sample objects in any target bin; determine the first ratio based on the third number and the first sum; determine the second ratio between the fifth number of negative sample objects in multiple sample objects and the total number of multiple sample objects; determine the lift of any target bin based on the first ratio and the second ratio. The specific formula is as follows:

[0089]

[0090] Among them, bad i is the fourth number of negative sample objects in any target bin, good i is the third number of positive sample objects in any target bin, bad represents the fifth number of negative sample objects in the multiple sample objects, and good+bad represents the total number of the multiple sample objects.

[0091] Step 304: Determine the risk control threshold of any target attribute variable based on the lift of each target bin.

[0092] As an example, the lift of each target bin is compared with the set lift, and the threshold of the bin corresponding to the lift greater than the set lift is used as the risk control threshold.

[0093] For example, as shown in Table 1, Table 1 is a mapping relationship table, which is used to indicate the mapping relationship between variable names, bins, positive sample objects, negative sample objects, negative sample object ratios and lift values. The threshold of the bin corresponding to the lift greater than 2 in Table 1 is used as the risk control threshold of the corresponding target attribute variable. For example, the risk control threshold of the credit rating in Table 1 is D, the risk control threshold of the current credit card limit is 0.85, and the risk control threshold of the number of loan approval inquiries in the last three months is 6.

[0094] Table 1 Mapping relationship table

[0095]

[0096] In summary, multiple candidate attribute variables are screened according to their information values ​​to obtain at least one target attribute variable; for any target attribute variable, the attribute value of any target attribute variable is binned to obtain multiple target bins; according to the third number of positive sample objects and the fourth number of negative sample objects in each target bin, the lift of the negative sample objects to which the attribute values ​​in each target bin belong is calculated, thereby screening out the target attribute variables that have a greater impact on risk assessment from the candidate attribute variables, and by binning the attribute values ​​of the target attribute variables, the distinguishing ability of risk assessment and the accuracy of risk assessment are improved.

[0097] In any embodiment of the present disclosure, based on the constraint conditions, the objective function for maximizing the risk score difference between the positive sample objects and the negative sample objects in each bin is solved to obtain multiple target bins. Figure 4 The embodiments are described in detail.

[0098] Figure 4 It is a flowchart of the risk assessment method shown in the fourth embodiment of the present disclosure.

[0099] like Figure 4 As shown, multiple target bins can be determined by following steps 401-402:

[0100] Step 401, obtaining multiple constraints.

[0101] The constraint condition is determined based on at least one of the mean of the first risk score of the positive sample objects, the mean of the second risk score of the negative sample objects, and the bin weight in any target attribute variable.

[0102] In the implementation of the present disclosure, based on the second basic attribute data, second risk control attribute data and risk control labels of multiple sample objects, the first risk score of the positive sample object and the second risk score of the negative sample object in any target attribute variable are determined.

[0103] As an example, based on the scoring model, the first risk score of each positive sample object and the second risk score of each negative sample object in the multiple sample objects are determined according to the second basic attribute data, the second risk control attribute data and the risk control labels of the multiple sample objects.

[0104] It should be noted that the multiple constraints include: a linear relationship constraint of scores, a constraint on the sum of scores, and a constraint on the weight of scores.

[0105] Among them, the linear relationship constraint of the score is expressed as the following formula:

[0106]

[0107] Among them, S g represents the mean risk score of the positive sample objects (the mean of the first risk score of each positive sample object), S b represents the mean risk score of the negative sample objects (the mean of the second risk scores of each negative sample object), represents the risk score variance of the positive sample object (the variance of the first risk score of each sample object), represents the risk score variance of the negative sample objects (the variance of the second risk score of each negative sample object);

[0108] The constraint of the score and is expressed as follows:

[0109] S g +S b =0;

[0110] Among them, S g represents the mean risk score of the positive sample objects, S b Represents the mean risk score of negative sample objects.

[0111] The scoring weight constraint is expressed as the following formula:

[0112]

[0113] Among them, T is the total number of bins in each target attribute variable, K is the weight threshold, and w ij is the binning weight for each target attribute variable, p is the number of input variables, q i Indicates the number of binning groups for the i-th target attribute variable.

[0114] Step 402 : Under the constraints of multiple constraints, solve the objective function for maximizing the risk score difference between the positive sample objects and the negative sample objects in each bin to obtain multiple target bins.

[0115] In the disclosed embodiment, the objective function is solved under the constraints of multiple constraints to ensure that the risk score gap between the positive sample objects and the negative sample objects in each bin is maximized, and multiple target bins are obtained to improve the accuracy of risk differentiation. The objective function is expressed as follows:

[0116]

[0117] in, represents the first risk score of the rth positive sample object, represents the second risk score of the rth negative sample object, m represents the number of positive sample objects, and n represents the number of negative sample objects.

[0118] In addition, it should be noted that, by solving the objective function under multiple constraints, the number of positive sample objects and the number of negative sample objects in each target bin can also be obtained.

[0119] In summary, by obtaining multiple constraints; under the constraints of multiple constraints, solving the objective function used to maximize the risk score gap between the positive sample objects and the negative sample objects in each bin, so as to obtain multiple target bins, thereby solving the objective function based on the constraints can ensure that the selection of the bin boundary is more reasonable, so that the data in each bin has similar risk characteristics, and by maximizing the risk score gap between the positive sample objects and the negative sample objects in each bin, it is possible to effectively distinguish high-risk and low-risk objects, thereby improving the risk differentiation ability of the model.

[0120] On the basis of any of the above embodiments, taking the target business as loan as an example, the risk assessment method of the embodiment of the present disclosure is implemented based on the following steps.

[0121] Step 1: Data collection and preprocessing

[0122] Collect basic information about the borrower, such as age, income, debt, credit rating, etc., and obtain their historical repayment behavior and credit records, such as the number of loan approval inquiries in the last three months (cnt_loan_audit_query_3m), current credit card limit usage (rate_credit_card_limit_usage_cur), and other external data. For example, the customer data table (700 data) contains the following two sample data information:

[0123] Customer A: age 30, income 5000, credit rating A, debt 1000, number of loan approval inquiries in the last 3 months 3, current credit card limit used 0.3, positive sample object (label = 1);

[0124] Customer B: age 45, income 7000, credit rating D, debt 2000, number of loan approval inquiries in the last 3 months 10, current credit card limit usage 0.95, negative sample object (label = 0);

[0125] In the preprocessing stage, all variables such as age, income, credit rating, etc. are converted into numerical data and integrated into a complete data table;

[0126] Step 2: Variable screening

[0127] Calculate the IV value for each variable in the data. For example, the order of the calculated IV values ​​from large to small is "credit rating", "number of loan approval inquiries in the last 3 months", "current credit card limit", "age", "income", and "debt". Through IV value screening, remove variables with values ​​below 0.5, and only keep those three variables (target attribute variables) that have a high degree of differentiation of customer credit, such as "credit rating", "number of loan approval inquiries in the last 3 months", and "current credit card limit".

[0128] Step 3: Binning

[0129] By maximizing the risk score gap between the positive sample objects and the negative sample objects in each bin, the variables of "credit rating", "number of loan approval inquiries in the last three months", and "current credit card limit" are binned.

[0130] Step 4: Calculate lift value

[0131] Calculate the lift value of each bin in each variable, as shown in Table 1.

[0132] Step 5: Threshold Optimization

[0133] The interval with lift value greater than 2 is selected as the final optimal threshold. For example, the threshold of "credit rating" is D, the threshold of "current credit card limit" is 0.85, and the threshold of "number of loan approval inquiries in the last 3 months" is 6. The thresholds of these bins are set as the key thresholds for risk assessment.

[0134] Corresponding to the risk assessment method provided in the above-mentioned embodiment, the present disclosure also provides a risk assessment device. Since the risk assessment device provided in the embodiment of the present disclosure corresponds to the risk assessment method provided in the above-mentioned embodiment, the implementation method of the risk assessment method is also applicable to the risk assessment device provided in the embodiment of the present disclosure, and will not be described in detail in the embodiment of the present disclosure.

[0135] Figure 5 It is a schematic diagram of the structure of a risk assessment device shown in the fifth embodiment of the present disclosure.

[0136] like Figure 5 As shown, the risk assessment device 500 includes: a first acquisition module 510 , a second acquisition module 520 and an assessment module 530 .

[0137] Among them, the first acquisition module 510 is used to obtain the risk control thresholds of multiple target attribute variables associated with the target business of the financial institution; wherein the risk control thresholds are determined based on the risk control labels of multiple sample objects and the attribute values ​​of the target attribute variables of the multiple sample objects, and the risk control labels are used to indicate the risk level of the sample objects in executing the target business; the second acquisition module 520 is used to obtain the first basic attribute data and the first risk control attribute data of the target object to execute the target business; wherein the first risk control attribute data includes the external risk control data of the target object outside the financial institution and the business risk control data within the financial institution; the evaluation module 530 is used to perform risk evaluation on the target object based on the risk control thresholds of multiple target attribute variables, the first basic attribute data and the first risk control attribute data to obtain the risk level of the target object in executing the target business.

[0138] As a possible implementation method of the embodiment of the present disclosure, the first acquisition module 510 is used to obtain the second basic attribute data, second risk control attribute data and risk control labels of multiple sample objects that execute the target business; wherein the second risk control attribute data includes the external risk control data of the corresponding sample objects outside the financial institution and the business risk control data within the financial institution; based on the risk control labels, the information values ​​of multiple candidate attribute variables are calculated; wherein the multiple candidate attribute variables are located in the second basic attribute data and the second risk control attribute data of the multiple sample objects; based on the information values ​​of the multiple candidate attribute variables, the risk control threshold of the target attribute variable among the multiple candidate attribute variables is determined.

[0139] As a possible implementation method of the embodiment of the present disclosure, the first acquisition module 510 is used to obtain multiple candidate bins of attribute values ​​of any candidate attribute variable for any candidate attribute variable; determine a first number of positive sample objects and a second number of negative sample objects in any candidate bin according to the risk control label of each sample object in any candidate bin; calculate the information sub-value of any candidate bin according to the first number and the second number; determine the information value of any candidate attribute variable according to the information sub-value of each candidate bin.

[0140] As a possible implementation method of the embodiment of the present disclosure, the first acquisition module 510 is used to screen multiple candidate attribute variables according to the information values ​​of the multiple candidate attribute variables to obtain at least one target attribute variable; for any target attribute variable, the attribute value of any target attribute variable is binned to obtain multiple target bins; according to the third number of positive sample objects and the fourth number of negative sample objects in each target bin, the lift of the negative sample objects to which the attribute values ​​in each target bin belong is calculated; wherein the positive sample object is a sample object whose risk control label is a positive label, and the negative sample object is a sample object whose risk control label is a negative label, the positive label is used to indicate that the risk level is lower than the set threshold, and the negative label is used to indicate that the risk level is not lower than the set threshold; based on the lift of each target bin, the risk control threshold of any target attribute variable is determined.

[0141] As a possible implementation method of the embodiment of the present disclosure, the first acquisition module 510 is used to sort multiple candidate attribute variables according to their information values ​​to obtain a candidate sequence; delete the candidate attribute variables with information values ​​lower than a set information threshold from the candidate sequence to obtain a target sequence; and use the candidate attribute variables in the target sequence as target attribute variables.

[0142] As a possible implementation method of an embodiment of the present disclosure, the first acquisition module 510 is used to determine, for any target bin, a first sum of a fourth number of negative sample objects and a third number of positive sample objects in any target bin; determine a first ratio based on the fourth number and the first sum; determine a second ratio between a fifth number of negative sample objects in multiple sample objects and the total number of multiple sample objects; and determine the lift of any target bin based on the first ratio and the second ratio.

[0143] As a possible implementation method of an embodiment of the present disclosure, multiple constraints are obtained; wherein the constraints are determined based on at least one of the mean of the first risk score of the positive sample objects, the mean of the second risk score of the negative sample objects, and the bin weights in any target attribute variable; under the constraints of the multiple constraints, an objective function for maximizing the risk score gap between the positive sample objects and the negative sample objects in each bin is solved to obtain multiple target bins.

[0144] As a possible implementation method of the embodiment of the present disclosure, the evaluation module 530 is used to obtain the attribute value of any target attribute variable from the first basic attribute data and the first risk control attribute data for any target attribute variable; in response to the attribute value of any target attribute variable being greater than or equal to the risk control threshold of any target attribute variable, determine that the risk level of the target object executing the target business is greater than the set risk level.

[0145] As a possible implementation method of the embodiment of the present disclosure, the evaluation module 530 is used to determine that the risk level of the target object executing the target business is not greater than the set risk level in response to the attribute values ​​of multiple target attribute variables in the first basic attribute data and the first risk control attribute data being less than the risk control thresholds of the multiple target attribute variables.

[0146] The risk assessment device of the embodiment of the present disclosure obtains risk control thresholds of multiple target attribute variables associated with the target business of the financial institution; wherein the risk control threshold is determined based on the risk control tags of multiple sample objects and the attribute values ​​of the target attribute variables of the multiple sample objects, and the risk control tags are used to indicate the risk level of the sample objects in executing the target business; obtains first basic attribute data and first risk control attribute data of the target object to execute the target business; wherein the first risk control attribute data includes external risk control data of the target object outside the financial institution and business risk control data within the financial institution; based on the risk control thresholds of the multiple target attribute variables, the first basic attribute data and the first risk control attribute data, the target object is risk assessed to obtain the risk level of the target object in executing the target business, thereby determining the risk control threshold based on the risk control tags of the sample objects and the attribute values ​​of the target attribute variables, realizing dynamic adjustment of the risk control threshold according to actual data, and improving the accuracy of the risk control threshold, and by combining the basic attribute data, external risk control data and internal business risk control data of the target object for risk assessment, it is possible to comprehensively assess the risk status of the target object from multiple angles, avoid the deviation that may be caused by a single data source, and thus improve the accuracy of risk assessment.

[0147] In an exemplary embodiment, an electronic device is also provided.

[0148] Among them, electronic equipment includes:

[0149] processor;

[0150] a memory for storing processor-executable instructions;

[0151] The processor is configured to execute instructions to implement the risk assessment method proposed in any of the aforementioned embodiments.

[0152] As an example, Figure 6 is a schematic diagram of the structure of an electronic device 600 shown in an exemplary embodiment of the present disclosure, such as Figure 6 As shown, the electronic device 600 may further include:

[0153] The memory 610 and the processor 620, a bus 630 connecting different components (including the memory 610 and the processor 620), the memory 610 stores a computer program, and when the processor 620 executes the program, the risk assessment method described in the embodiment of the present disclosure is implemented.

[0154] Bus 630 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. By way of example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0155] The electronic device 600 typically includes a variety of electronic device readable media. These media can be any available media that can be accessed by the electronic device 600, including volatile and non-volatile media, removable and non-removable media.

[0156] The memory 610 may also include computer system readable media in the form of volatile memory, such as random access memory (RAM) 640 and / or cache memory 650. The server 600 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 660 may be used to read and write non-removable, non-volatile magnetic media ( Figure 6 not shown, usually called a "hard drive"). Although Figure 6 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 630 via one or more data medium interfaces. The memory 610 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present disclosure.

[0157] A program / utility 680 having a set (at least one) of program modules 670 may be stored, for example, in the memory 610, such program modules 670 including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 670 generally perform the functions and / or methods of the embodiments described in the present disclosure.

[0158] The electronic device 600 may also communicate with one or more external devices 690 (e.g., keyboard, pointing device, display 691, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 600, and / or communicate with any device that enables the electronic device 600 to communicate with one or more other computing devices (e.g., network card, modem, etc.). Such communication may be performed via an input / output (I / O) interface 692. Furthermore, the electronic device 600 may also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN) and / or public network, such as the Internet) via a network adapter 693. As shown, the network adapter 693 communicates with other modules of the electronic device 600 via a bus 630. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0159] The processor 620 executes various functional applications and data processing by running the programs stored in the memory 610 .

[0160] It should be noted that the implementation process and technical principles of the electronic device of this embodiment refer to the aforementioned explanation of the risk assessment method of the embodiment of the present disclosure, and will not be repeated here.

[0161] In an exemplary embodiment, a computer-readable storage medium including instructions is also provided, such as a memory including instructions, and the above instructions can be executed by a processor of an electronic device to complete the risk assessment method proposed in any of the above embodiments. Optionally, the computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0162] In an exemplary embodiment, a computer program product is also provided, including a computer program / instruction, wherein the computer program / instruction implements the risk assessment method proposed in any of the above embodiments when executed by a processor.

[0163] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0164] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A risk assessment method, characterized in that: include: Obtaining risk control thresholds of multiple target attribute variables associated with a target business of a financial institution; wherein the risk control thresholds are determined based on risk control tags of multiple sample objects and attribute values ​​of the target attribute variables of the multiple sample objects, and the risk control tags are used to indicate the risk level of the sample objects in executing the target business; Acquire first basic attribute data and first risk control attribute data of a target object for executing a target business; wherein the first risk control attribute data includes external risk control data of the target object outside the financial institution and business risk control data within the financial institution; Based on the risk control thresholds of the multiple target attribute variables, the first basic attribute data and the first risk control attribute data, a risk assessment is performed on the target object to obtain a risk level of the target object in executing the target business.

2. The method according to claim 1, characterized in that The step of obtaining risk control thresholds of multiple target attribute variables associated with the target business of the financial institution includes: Acquire second basic attribute data, second risk control attribute data, and risk control labels of multiple sample objects that execute the target business; wherein the second risk control attribute data includes external risk control data of the corresponding sample objects outside the financial institution and business risk control data within the financial institution; Based on the risk control label, calculating information values ​​of a plurality of candidate attribute variables; wherein the plurality of candidate attribute variables are located in the second basic attribute data and the second risk control attribute data of the plurality of sample objects; Based on the information values ​​of the multiple candidate attribute variables, a risk control threshold of a target attribute variable among the multiple candidate attribute variables is determined.

3. The method according to claim 2, characterized in that The calculating information values ​​of a plurality of candidate attribute variables based on the risk control label includes: For any candidate attribute variable, obtain multiple candidate bins of the attribute value of any candidate attribute variable; Determine, according to the risk control label of each sample object in any candidate sub-bin, a first number of positive sample objects and a second number of negative sample objects in any candidate sub-bin; Calculate the information sub-value of any candidate bin according to the first quantity and the second quantity; The information value of any candidate attribute variable is determined according to the information sub-value of each candidate bin.

4. The method according to claim 2, characterized in that: The step of determining the risk control threshold of a target attribute variable among the plurality of candidate attribute variables based on the information values ​​of the plurality of candidate attribute variables includes: Screening the plurality of candidate attribute variables according to information values ​​of the plurality of candidate attribute variables to obtain at least one target attribute variable; For any target attribute variable, the attribute value of any target attribute variable is binned to obtain a plurality of target bins; According to the third number of positive sample objects and the fourth number of negative sample objects in each of the target bins, the lift of the negative sample objects to which the attribute values ​​in each of the target bins belong is calculated; wherein the positive sample objects are sample objects whose risk control labels are positive labels, and the negative sample objects are sample objects whose risk control labels are negative labels, the positive label is used to indicate that the risk level is lower than a set threshold, and the negative label is used to indicate that the risk level is not lower than the set threshold; Based on the lift of each of the target bins, a risk control threshold of any target attribute variable is determined.

5. The method according to claim 4, characterized in that The screening of the plurality of candidate attribute variables according to the information values ​​of the plurality of candidate attribute variables to obtain at least one target attribute variable includes: Sort the multiple candidate attribute variables according to the information values ​​of the multiple candidate attribute variables to obtain a candidate sequence; From the candidate sequence, delete the candidate attribute variables whose information values ​​are lower than a set information threshold to obtain a target sequence; The candidate attribute variables in the target sequence are used as the target attribute variables.

6. The method according to claim 4, characterized in that The calculating, according to the third number of positive sample objects and the fourth number of negative sample objects in each of the target bins, the lift of the negative sample objects to which the attribute values ​​in each of the target bins belong, comprises: For any target bin, determine a first sum of a fourth number of negative sample objects and a third number of positive sample objects in the any target bin; determining a first ratio according to the fourth quantity and the first sum; determining a second ratio between a fifth number of negative sample objects in the plurality of sample objects and a total number of the plurality of sample objects; The lift of any target bin is determined according to the first ratio and the second ratio.

7. The method according to claim 4, characterized in that For any target attribute variable, the attribute value of any target attribute variable is binned to obtain multiple target bins, including: Acquire multiple constraints; wherein the constraints are determined based on at least one of the mean of the first risk score of the positive sample object, the mean of the second risk score of the negative sample object, and the bin weight in any target attribute variable; Under the constraints of the multiple constraints, an objective function for maximizing the risk score gap between the positive sample objects and the negative sample objects in each bin is solved to obtain multiple target bins.

8. The method according to claim 1, characterized in that The risk control threshold based on the multiple target attribute variables, the first basic attribute data and the first risk control attribute data, performing risk assessment on the target object to obtain a risk level of the target object executing the target business, includes: For any target attribute variable, obtaining an attribute value of the any target attribute variable from the first basic attribute data and the first risk control attribute data; In response to the attribute value of any target attribute variable being greater than or equal to the risk control threshold of any target attribute variable, it is determined that the risk level of the target object executing the target business is greater than a set risk level.

9. The method according to claim 8, characterized in that The method further comprises: In response to the fact that the attribute values ​​of the multiple target attribute variables in the first basic attribute data and the first risk control attribute data are all less than the risk control thresholds of the multiple target attribute variables, it is determined that the risk level of the target object executing the target business is not greater than the set risk level.

10. A risk assessment device, characterized in that: include: A first acquisition module is used to acquire risk control thresholds of multiple target attribute variables associated with a target business of a financial institution; wherein the risk control thresholds are determined based on risk control tags of multiple sample objects and attribute values ​​of the target attribute variables of the multiple sample objects, and the risk control tags are used to indicate the risk level of the sample objects in executing the target business; A second acquisition module is used to acquire first basic attribute data and first risk control attribute data of a target object of a target business to be executed; wherein the first risk control attribute data includes external risk control data of the target object outside the financial institution and business risk control data within the financial institution; An evaluation module is used to perform a risk evaluation on the target object based on the risk control thresholds of the multiple target attribute variables, the first basic attribute data and the first risk control attribute data to obtain a risk level of the target object in executing the target business.

Images