CN114331671A - Loan risk monitoring method and device, server and storage medium - Google Patents
Loan risk monitoring method and device, server and storage medium Download PDFInfo
- Publication number
- CN114331671A CN114331671A CN202111582409.5A CN202111582409A CN114331671A CN 114331671 A CN114331671 A CN 114331671A CN 202111582409 A CN202111582409 A CN 202111582409A CN 114331671 A CN114331671 A CN 114331671A
- Authority
- CN
- China
- Prior art keywords
- risk
- data
- credit
- reputation
- score
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention provides a loan risk monitoring method, a loan risk monitoring device, a server and a storage medium, wherein the method comprises the steps of obtaining the operation data, the reputation data and the credit data of an object to be monitored; respectively inputting an operation condition evaluation model, a reputation evaluation model and a credit evaluation model, and outputting an operation risk score, a reputation risk score and a credit risk score; acquiring a preset known credit model, and calculating the risk score of an object to be monitored; determining a model weight coefficient, and calculating the final risk total score of the object to be monitored; and comparing the operation risk score, the reputation risk score, the credit risk score, the risk score and the final risk total score with respective corresponding set risk thresholds, and performing early warning according to a comparison result. According to the scheme, business data is taken as a main body, data sources such as reputation data and credit data are integrated, data latitude is enriched, a brand-new post-credit monitoring strategy is set, early warning decision is more flexible, characteristics of customer groups are more conformed, and therefore the risk control effect is improved.
Description
Technical Field
The invention relates to the technical field of bank wind control, in particular to a loan risk monitoring method, a loan risk monitoring device, a loan risk monitoring server and a loan risk monitoring storage medium.
Background
At present, with the increase of the supporting force of the country to the small and micro enterprises and the implementation of direct connection of the bank tax, the rise of the finance of the small and micro enterprises is driven, and a series of financial products and service platforms facing the small and micro enterprises emerge in recent two years. The small and micro financial products have strong impact on the traditional offline examination and approval mode around the online examination and approval mode of the establishment of the direct connection mode of the bank tax. In addition, as the small and micro enterprise user group belongs to a lower-layer client group, the production and operation fluctuation is large, the risk resistance is poor compared with that of a traditional B-end business client group, and as the small and micro enterprise user has the problems of non-standard tax payment behaviors and the like, the traditional post-loan monitoring mode is poor in performance on the client group due to the lack of the support of financial statement related data and cash flow tables concerned by traditional B-end financial products, and the traditional post-loan monitoring strategy has the following problems for pure credit loan products: the traditional B-end service has single observation data in the post-credit monitoring aspect and mainly observes judicial relevant information and credit investigation information; the traditional B-end service mostly adopts a one-time mode in the aspect of post-credit monitoring, and cannot be subjected to industry subdivision.
Disclosure of Invention
The invention provides a loan risk monitoring method, a loan risk monitoring device, a loan risk monitoring server and a loan risk monitoring storage medium, which mainly solve the technical problems that: the traditional B-end service observation data is single, real data support is lacked, and the post-loan monitoring algorithm is performed by one-time cutting, so that the risk control effect is poor.
In order to solve the above technical problems, the present invention provides a loan risk monitoring method, including:
acquiring operation data, reputation data and credit data of an object to be monitored;
inputting the operation data into an operation condition evaluation model, and outputting an operation risk value; inputting the reputation data into a reputation evaluation model and outputting a reputation risk score; inputting the credit data into a credit evaluation model and outputting a credit risk score; the operation condition evaluation model, the reputation evaluation model and the credit evaluation model are obtained by machine learning and construction by using a logistic regression algorithm;
acquiring a preset known credit model, and calculating the risk score of the object to be monitored;
determining model weight coefficients of the operation condition evaluation model, the reputation evaluation model, the credit evaluation model and the preset known credit model, and calculating a final risk total score of the object to be monitored;
and comparing the operation risk score, the reputation risk score, the credit risk score, the risk score and the final risk total score with respective corresponding set risk thresholds, and performing early warning according to comparison results.
Optionally, the inputting the business data into the business situation assessment model includes: determining the industry characteristic of the object to be monitored, calling an operation condition evaluation model corresponding to the industry characteristic of the object to be monitored so as to input the operation data into the operation condition evaluation model; and aiming at different industry characteristic samples, constructing corresponding operation condition evaluation models in advance based on the industry characteristic samples.
Optionally, the industry characteristics include: industry category and enterprise size.
Optionally, the industry categories include production, trade, retail, and service; the enterprise size is divided according to revenue.
Optionally, the operation data includes tax data and industry fluctuation data, where the tax data includes tax payment change condition, income fluctuation condition, and stockholder change condition;
the reputation data comprises judicial data, public opinion data, bidding and financing investment data associated with the enterprise and/or enterprise high management;
the credit data includes third party credit data and credit investigation data.
Optionally, the industry fluctuation data includes a business prosperity index of the enterprise to be monitored, an industry rank of the enterprise to be monitored in the near-three-month revenue scale, an annular-ratio industry rank of the enterprise to be monitored in the near-three-month revenue scale, and a same-ratio industry rank of the enterprise in the near-three-month revenue scale.
Optionally, the performing early warning according to the comparison result includes:
and when the business risk score is smaller than a set risk threshold corresponding to the business risk score, or the reputation risk score is smaller than a set risk threshold corresponding to the business risk score, the credit risk score is smaller than a set risk threshold corresponding to the business risk score, the risk score is smaller than a set risk threshold corresponding to the business risk score, or the final risk total score is smaller than a set risk threshold corresponding to the final risk total score, early warning is carried out.
The invention also provides a loan risk monitoring device, comprising:
the first acquisition module is used for acquiring the operation data, the reputation data and the credit data of the object to be monitored;
the data processing module is used for inputting the operation data into the operation condition evaluation model and outputting an operation risk value; inputting the reputation data into a reputation evaluation model and outputting a reputation risk score; inputting the credit data into a credit evaluation model and outputting a credit risk score; the operation condition evaluation model, the reputation evaluation model and the credit evaluation model are obtained by machine learning and construction by using a logistic regression algorithm;
the second acquisition module is used for acquiring a preset known credit model and calculating the risk score of the object to be monitored;
the monitoring and early warning module is used for determining model weight coefficients of the operation condition evaluation model, the reputation evaluation model, the credit evaluation model and the preset known credit model and calculating the final risk total score of the object to be monitored; and comparing the operation risk score, the reputation risk score, the credit risk score, the risk score and the final risk total score with respective corresponding set risk thresholds, and performing early warning according to comparison results.
The invention also provides a server, which comprises a processor, a memory and a communication bus;
the communication bus is used for realizing connection communication between the processor and the memory;
the processor is configured to execute one or more programs stored in the memory to implement the steps of the loan risk monitoring method as described in any one of the above.
The present invention also provides a storage medium storing one or more programs executable by one or more processors to perform the steps of the loan risk monitoring method as described in any one of the above.
The invention has the beneficial effects that:
according to the loan risk monitoring method, the loan risk monitoring device, the loan risk monitoring server and the storage medium, the method comprises the steps of obtaining operation data, reputation data and credit data of an object to be monitored; respectively inputting an operation condition evaluation model, a reputation evaluation model and a credit evaluation model, and outputting an operation risk score, a reputation risk score and a credit risk score; the business situation assessment model, the reputation assessment model and the credit assessment model are obtained by constructing a logistic regression algorithm through machine learning; acquiring a preset known credit model, and calculating the risk score of an object to be monitored; determining model weight coefficients of an operation condition evaluation model, a reputation evaluation model, a credit evaluation model and a preset known credit model, and calculating a final risk total score of an object to be monitored; and comparing the operation risk score, the reputation risk score, the credit risk score, the risk score and the final risk total score with respective corresponding set risk thresholds, and performing early warning according to a comparison result. According to the scheme, business data is taken as a main body, data sources such as reputation data and credit data are integrated, data latitude is enriched, a brand-new post-credit monitoring strategy is set, early warning decision is more flexible, characteristics of customer groups are more conformed, and therefore the risk control effect is improved.
Drawings
FIG. 1 is a schematic flow chart illustrating a loan risk monitoring method according to a first embodiment of the invention;
FIG. 2 is a schematic structural diagram of a loan risk monitoring apparatus according to a second embodiment of the invention;
fig. 3 is a schematic structural diagram of a server according to a third embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following detailed description and accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The first embodiment is as follows:
at present, in the traditional post-credit monitoring, business data and credit investigation data of a main reference user are used for formulating a strategy, a unified threshold is adopted for monitoring and early warning of customers in a customer group, observation data are single, a post-credit monitoring algorithm is simple, and the problem of poor risk control effect exists. Particularly, for small micro-enterprises, due to the change of client groups, the users of the small micro-enterprises do not have the same complete financial and tax report data as the client groups in the traditional B-end service, or the financial and tax report is self-reported to cause data distortion; in addition, the small and micro enterprise user group has poor risk resistance and high operation fluctuation, and the traditional post-credit monitoring strategy has high alarm triggering rate, so that the early warning efficiency is low. Meanwhile, since most of small and micro financial products are credit credits, and the distribution difference of the industry is large, the post-credit strategy of the unified standard has no universality.
In view of the above problems, the present embodiment provides a loan risk monitoring method, which takes enterprise business data as a main body, integrates data sources such as reputation data and credit data, enriches data latitude, and establishes a brand-new post-loan monitoring policy, so that an early warning decision is more flexible, and a problem of one-time decision of the same standard is avoided, thereby better fitting customer group characteristics, and improving a risk control effect.
Referring to fig. 1, the loan risk monitoring method includes the following steps:
s101, obtaining management data, business reputation data and credit data of an object to be monitored.
The business data includes tax data and industry fluctuation data.
The tax data comprises tax payment change condition, income fluctuation condition and stockholder change condition. By utilizing the tax related data, the recent operation behavior of the enterprise, including the short-term and medium-term tax payment change, income fluctuation condition, stockholder change and other information of the enterprise, can be restored as much as possible, thereby reflecting the recent operation stability of the enterprise.
The tax change data includes, but is not limited to, whether the tax is owed (e.g., set to 0, and set to 1, if yes), the current number of consecutive zero-declaration-period, the number of zero-declaration-period in the last twelve months of the enterprise, and the number of late-fund-existence periods in the last six months of the enterprise.
The income fluctuation conditions include, but are not limited to, the revenue scale of the enterprise in approximately three months, the revenue scale of the enterprise in approximately six months, the revenue share ratio of the enterprise in approximately three business and the revenue ring ratio of the enterprise in approximately three business.
The shareholder change data includes, but is not limited to, whether shareholder reductions have recently occurred to the enterprise (e.g., set to 0, if not, set to 1).
And the industry fluctuation data laterally reflects the future income trend of the enterprise according to the overall change trend of the future income of the industry by means of research and analysis of the market, judgment of industry scene degree and the like.
In this embodiment, the industry fluctuation data includes the industry prosperity index of the enterprise to be monitored, the industry rank of the revenue scale of the enterprise to be monitored in approximately three months, the ring-ratio industry rank of the revenue scale of the enterprise to be monitored in approximately three months, and the industry rank of the revenue scale of the enterprise in approximately three months.
The industry prosperity index is an index which processes and summarizes qualitative indexes in enterprise prosperity survey through a quantitative method and comprehensively reflects the state or development trend of a specific survey group or a social phenomenon. The prosperity index is between 0 and 200, the higher the prosperity index value is, the economic condition tends to rise or improve, and the lower the prosperity index value is, the economic condition tends to decline or worsen, and the gas is in a bad state. The industry prosperity index is an index for measuring the prosperity degree of the industry and is a reflection of the future development trend of the industry. For financial institutions, a certain enterprise is selected for investment, and the prospect index of the industry of the enterprise has important reference significance.
The industry prosperity index can be calculated as follows: one is to calculate the business prospect index by means of weighted averaging for a set of questions. For example, the individual index of each problem is calculated through the indexes of ordering, production, sales, profit, labor consumption, investment and the like closely related to the production and operation activities of enterprises, corresponding weight is given to each problem, and then the comprehensive prospect index is calculated through weighted average. And secondly, only aiming at the answer calculation of one comprehensive question, namely calculating the business prosperity index according to the judgment and the expectation of the comprehensive business condition of the enterprise. Since the national statistics institute formally carries out institutional enterprise business investigation in 1999, the second method has been adopted. The industry prosperity index can be calculated by any conventional method, and the embodiment is not limited to this.
The earning scale is divided into three aspects of the amount to be earned, the annual percentage of the previous month and the year-on-month growth situation ranking, and the operation situation of the enterprise to be monitored is comprehensively considered. The ranking condition is based on revenue and earning data of all enterprises of the system, and ranking can be performed by different industries.
Reputation data includes judicial data, public opinion data, bidding and financing investment data associated with a business and/or business high administration. The judicial data and the public opinion data reflect the reputation risk of the user, the judgment documents related to the judicial reflect the possibility of fund change of the user laterally, and the bidding data and the financing investment information reflect the trend of the fund change of the user in a short period laterally.
The judicial data includes, but is not limited to, whether the enterprise is an executed enterprise (e.g., set to 0, if not, set to 1), whether the enterprise is a lost credit executed enterprise (e.g., set to 0, if not, set to 1), the current number of outstanding cases of the enterprise, whether the enterprise highly manages the current number of outstanding cases, whether the enterprise highly manages the person to be executed (e.g., set to 0, if not, set to 1), and whether the enterprise highly manages the person to be executed (e.g., set to 0, if not, set to 1).
The public opinion data is divided into positive public opinion data and negative public opinion data, including but not limited to the number of positive public opinion information, the popularity of positive public opinion, the number of negative public opinion information, and the popularity of negative public opinion. Positive public sentiment is positive number, and negative public sentiment is negative number. It should be understood that the public sentiment quantity can be obtained based on some existing known platform statistics, and the public sentiment popularity can also be calculated based on the existing manner, which is not limited by the embodiment. Such as a hundred-degree hot search.
Bidding includes, but is not limited to, a bid amount; the financing investment includes, but is not limited to, a financing amount and an investment amount.
The credit data includes third party credit data and credit investigation data. The credit data of the third party is used for measuring the risk of multi-head borrowing and lending of the user, the credit investigation data can reflect the recent fund demand of the user and the recent fund looseness of the user, and the overdue probability of the repayment of the user in the next account can be accurately measured by combining the income of the user and the expected refund amount of the next account. The third party credit data includes but is not limited to credit data given by payment treasures, WeChat, banks, and other financial institutions, and the credit data gives credit data given by evaluation to people's banks.
The credit data comprises the total amount of the latest one-month credit repayment of the enterprise, the total amount of the latest three-month credit repayment of the enterprise, the total amount of the latest six-month credit repayment of the enterprise, the number of the latest half-year performance scenes of the enterprise, the number of the latest one-month actively inquired financial institutions of the enterprise, the number of the latest three-month actively inquired financial institutions of the enterprise, the number of the latest six-month actively inquired financial institutions of the enterprise, and the income-debt ratio of the latest six months of the enterprise, the total amount of credit type repayment of the enterprise grand manager in the last month, the total amount of credit type repayment of the enterprise grand manager in the last three months, the total amount of credit type repayment of the enterprise grand manager in the last six months, the number of the business senior manager in the last half year of performance scene, the number of financial institutions actively inquired by the enterprise grand manager in the last one month, the number of the business senior manager in the last three months, the number of the business senior manager in the last six months, and the income-debt ratio of the enterprise grand manager in the last six months.
See table 1 below:
TABLE 1
And (4) operating risk: and according to the information of short-term and medium-term tax payment change, income fluctuation condition, stockholder change and the like of the enterprise, reflecting the recent operation stability of the enterprise.
Reputation risk: the impact force brought to the reputation of the enterprise is evaluated by judging newly added case-involved information in the enterprise and the enterprise high-management near term.
Industry fluctuation: by means of research and analysis on the market, judgment on the business scene degree and the like, the future income trend of an enterprise is reflected laterally according to the overall change trend of the future income of the industry.
And (4) opening funds: the recent force of the enterprise user on the capital requirement and the asset liability ratio are reflected through the analysis of credit information of the enterprise and the enterprise owner, and the probability of the multi-head risk of the enterprise owner is laterally evaluated through the observation and analysis of credit data of a third party, so that the overdue risk of the user is reflected. In view of the characteristic that small and micro enterprises are not classified into public and private, when the credit investigation related indexes are considered, the scheme not only considers the credit investigation change condition of an application main body, but also judges the personal credit investigation change of an applicant (generally a legal person and an enterprise high-level administration).
The method has rich index dimensionality, fully considers the possible business fluctuation condition of the enterprise in a short period, and further can fully and accurately measure the risk after the loan. Comprehensively considering the aspects of operational risk, reputation risk, industry fluctuation, fund exposure and the like, the method is more suitable for the characteristics of the customer group, thereby improving the risk control effect.
And S102, inputting the operation data into the operation condition evaluation model and outputting the operation risk value.
Inputting the business data into the business situation assessment model comprises: determining the industry characteristic of an object to be monitored, calling an operation condition evaluation model corresponding to the industry characteristic of the object to be monitored so as to input operation data into the operation condition evaluation model; and aiming at different industry characteristic samples, constructing corresponding operation condition evaluation models in advance based on the industry characteristic samples.
The industry characteristics include: industry category and enterprise size. Industry categories include, but are not limited to, production, trade, retail, and service; the enterprise size is divided according to the business income. Enterprise sizes include, but are not limited to, micro-scale enterprises, small-scale enterprises, medium-scale enterprises, large-scale enterprises, super-large-scale enterprises, and the like.
Specifically, the samples are classified according to the industry characteristics, and different industry categories or different enterprise scales can be classified into one category, such as production type micro enterprises, production type small enterprises, production type medium enterprises and the like. And obtaining an operation condition evaluation model corresponding to the industry characteristics by adopting a logistic regression algorithm based on the sample of each industry characteristic to obtain each index weight, so that the corresponding operation risk score can be output based on the input operation data. According to the invention, through scene segmentation and combination of user operation data, different volatility of different industries is fully considered, so that a segmented user operation condition evaluation model is constructed for different industries, and the problem of user early warning failure caused by the uniformity standard of a traditional strategy is solved.
In this embodiment, the industry characteristic division standard can be referred to as the following table 2:
TABLE 2
S103, inputting the reputation data into a reputation evaluation model and outputting a reputation risk score.
In the embodiment, a potential reputation evaluation model of an enterprise is constructed by using relevant data of judicial, bidding and investment fusion; and (3) evaluating the risk coefficient of the user by adopting a minimum risk decision algorithm:
decision alphai: classifying a sample x to be identified into wiClass (c);
loss of lambdaij: to belong to the truth of wiClass samples x to wiLosses due to class;
conditional Risk R (. alpha.)i| x): post-decision alpha is taken for sample xiTotal risk of the latter
Classification decision rules:
and S104, inputting the credit data into the credit evaluation model and outputting a credit risk score.
In the embodiment, the credit evaluation model adopts a logistic regression model to construct the credit evaluation model according to the existing sample labels, and the weight of each index is obtained;
the core principle of the logistic regression algorithm is as follows:
logistic function:
LR definition:
P(Y＝1x)＝δ(x)，P(Y＝0|x)＝1-δ(x)
loss function:
substituting into the Logistic function yields:
the gradient function is:
the gradient is updated as:
it should be understood that the execution order of steps S102 to S104 is not limited, and may be processed in series or in parallel.
And S105, acquiring a preset known credit model, and calculating the risk score of the object to be monitored.
It should be noted that, a known credit model is preset as a currently existing mature credit model, and the risk score of the object to be monitored is output based on a plurality of existing credit models by acquiring the plurality of existing credit models in the present embodiment. For example, the risk scores output by the credit models are subjected to mean processing to serve as the risk scores of the objects to be monitored.
And S106, determining model weight coefficients of the business situation evaluation model, the reputation evaluation model, the credit evaluation model and the preset known credit model.
Optionally, an expert experience method is used to respectively assign weight coefficients to the business situation assessment model, the reputation assessment model, the credit assessment model and the preset known credit model.
And S107, calculating the final risk total score of the object to be monitored.
Wherein, each model is divided into 0-100 points, and the final risk total point of the object to be monitored is calculated according to the following formula:
score＝∑wi*xi
for example, the business situation assessment model, the reputation assessment model, the credit assessment model and the preset known credit model are respectively assigned with weight coefficients of 0.4, 0.3, 0.2 and 0.1 by using an expert experience method, then:
score 0.4 business risk Score +0.3 reputation risk Score +0.2 credit risk Score +0.1 existing model risk Score.
And S108, comparing the operation risk score, the reputation risk score, the credit risk score, the risk score and the final risk total score with respective corresponding set risk thresholds.
In this embodiment, the risk thresholds set for different industry characteristics are different, please refer to table 3 below:
TABLE 3
That is, for a production type small business, if the final risk total score is below 55, or the business risk score is below 50, or the reputation risk score is below 35, or the credit risk score is below 35, or the existing model score is below 40, then a pre-warning is triggered.
And S109, early warning is carried out according to the comparison result.
And when the operation risk score is smaller than the corresponding set risk threshold, or the reputation risk score is smaller than the corresponding set risk threshold, the credit risk score is smaller than the corresponding set risk threshold, the risk score is smaller than the corresponding set risk threshold, or the final risk total score is smaller than the corresponding set risk threshold, early warning is carried out.
The loan risk monitoring method provided by the invention takes tax data as a main body, integrates data sources such as public data, credit investigation data, expert experience and the like, and establishes a brand-new post-loan monitoring strategy. The user can be monitored in near real time or at regular time by a pure online mode. The problem of user early warning failure caused by the unification standard of the traditional strategy is solved through operations such as scene detail, the small micro-enterprise operation behavior is fully restored through analyzing tax payment behavior data of the small micro-enterprise, and the problem of failure caused by financial statement data loss of the traditional B-end early warning strategy is solved. And a plurality of data sources are introduced, the data latitude is enriched, and the early warning decision is more flexible and more suitable for the characteristics of the customer group.
Example two:
in this embodiment, on the basis of the first embodiment, a loan risk monitoring apparatus is provided to implement at least part of the steps of the loan risk monitoring method in the first embodiment, please refer to fig. 2, the loan risk monitoring apparatus mainly includes:
the first acquisition module 21 is configured to acquire business data, reputation data and credit data of an object to be monitored;
the data processing module 22 is used for inputting the operation data into the operation condition evaluation model and outputting the operation risk value; inputting the reputation data into a reputation evaluation model, and outputting a reputation risk score; inputting the credit data into a credit evaluation model and outputting a credit risk score; the business situation assessment model, the reputation assessment model and the credit assessment model are obtained by constructing a logistic regression algorithm through machine learning;
the second obtaining module 23 is used for obtaining a preset known credit model and calculating the risk score of the object to be monitored;
the monitoring and early warning module 24 is used for determining a model weight coefficient of the operation condition evaluation model, the reputation evaluation model, the credit evaluation model and a preset known credit model, and calculating the final risk total score of the object to be monitored; and comparing the operation risk score, the reputation risk score, the credit risk score, the risk score and the final risk total score with respective corresponding set risk thresholds, and performing early warning according to a comparison result.
The loan risk monitoring apparatus in this embodiment may be implemented on a server, and each functional module may implement the above functions through a hardware module such as a controller and a processor of the server, in combination with a specific software algorithm. For details, please refer to the description in the first embodiment, which is not repeated herein.
Example three:
in this embodiment, on the basis of the first embodiment, a server is provided for implementing at least part of the steps of the loan risk monitoring method in the first embodiment, please refer to fig. 3, the server mainly includes a processor 31, a memory 32 and a communication bus 33;
wherein, the communication bus 33 is used for realizing the connection communication between the processor 31 and the memory 32;
the processor 31 is configured to execute one or more programs stored in the memory 32 to implement the steps of the loan risk monitoring method as described in the first embodiment. For details, please refer to the description in the first embodiment, which is not repeated herein.
To achieve the above object, the present embodiment also provides a computer-readable storage medium storing one or more programs, which are executable by one or more processors to implement the steps of the loan risk monitoring method as described in the first embodiment. For details, please refer to the description in the first embodiment, which is not repeated herein.
It will be apparent to those skilled in the art that the modules or steps of the invention described above may be implemented in a general purpose computing device, they may be centralized on a single computing device or distributed across a network of computing devices, and optionally they may be implemented in program code executable by a computing device, such that they may be stored on a computer storage medium (ROM/RAM, magnetic disks, optical disks) and executed by a computing device, and in some cases, the steps shown or described may be performed in an order different than that described herein, or they may be separately fabricated into individual integrated circuit modules, or multiple ones of them may be fabricated into a single integrated circuit module. Thus, the present invention is not limited to any specific combination of hardware and software.
The foregoing is a more detailed description of the present invention that is presented in conjunction with specific embodiments, and the practice of the invention is not to be considered limited to those descriptions. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.
Claims (10)
1. A method for loan risk monitoring, comprising:
acquiring operation data, reputation data and credit data of an object to be monitored;
inputting the operation data into an operation condition evaluation model, and outputting an operation risk value; inputting the reputation data into a reputation evaluation model and outputting a reputation risk score; inputting the credit data into a credit evaluation model and outputting a credit risk score; the operation condition evaluation model, the reputation evaluation model and the credit evaluation model are obtained by machine learning and construction by using a logistic regression algorithm;
acquiring a preset known credit model, and calculating the risk score of the object to be monitored;
determining model weight coefficients of the operation condition evaluation model, the reputation evaluation model, the credit evaluation model and the preset known credit model, and calculating a final risk total score of the object to be monitored;
and comparing the operation risk score, the reputation risk score, the credit risk score, the risk score and the final risk total score with respective corresponding set risk thresholds, and performing early warning according to comparison results.
2. The loan risk monitoring method of claim 1, wherein the entering the business data into a business situation assessment model comprises: determining the industry characteristic of the object to be monitored, calling an operation condition evaluation model corresponding to the industry characteristic of the object to be monitored so as to input the operation data into the operation condition evaluation model; and aiming at different industry characteristic samples, constructing corresponding operation condition evaluation models in advance based on the industry characteristic samples.
3. The loan risk monitoring method of claim 2, wherein the industry characteristic comprises: industry category and enterprise size.
4. A loan risk monitoring method as claimed in claim 3, wherein the industry categories include production, trade, retail and service; the enterprise size is divided according to revenue.
5. The loan risk monitoring method of any one of claims 1 to 4, wherein the business data includes tax data and industry fluctuation data, wherein the tax data includes tax payment variation, income fluctuation, stockholder variation;
the reputation data comprises judicial data, public opinion data, bidding and financing investment data associated with the enterprise and/or enterprise high management;
the credit data includes third party credit data and credit investigation data.
6. The loan risk monitoring method according to claim 5, wherein the industry fluctuation data includes an industry prosperity index of the enterprise to be monitored, an industry rank of the revenue scale of the enterprise to be monitored in approximately three months, a circle-to-circle industry rank of the revenue scale of the enterprise to be monitored in approximately three months, and a unity-to-scale industry rank of the revenue scale of the enterprise in approximately three months.
7. The loan risk monitoring method of claim 5, wherein the pre-warning based on the comparison comprises:
and when the business risk score is smaller than a set risk threshold corresponding to the business risk score, or the reputation risk score is smaller than a set risk threshold corresponding to the business risk score, the credit risk score is smaller than a set risk threshold corresponding to the business risk score, the risk score is smaller than a set risk threshold corresponding to the business risk score, or the final risk total score is smaller than a set risk threshold corresponding to the final risk total score, early warning is carried out.
8. A loan risk monitoring apparatus, comprising:
the first acquisition module is used for acquiring the operation data, the reputation data and the credit data of the object to be monitored;
the data processing module is used for inputting the operation data into the operation condition evaluation model and outputting an operation risk value; inputting the reputation data into a reputation evaluation model and outputting a reputation risk score; inputting the credit data into a credit evaluation model and outputting a credit risk score; the operation condition evaluation model, the reputation evaluation model and the credit evaluation model are obtained by machine learning and construction by using a logistic regression algorithm;
the second acquisition module is used for acquiring a preset known credit model and calculating the risk score of the object to be monitored;
the monitoring and early warning module is used for determining model weight coefficients of the operation condition evaluation model, the reputation evaluation model, the credit evaluation model and the preset known credit model and calculating the final risk total score of the object to be monitored; and comparing the operation risk score, the reputation risk score, the credit risk score, the risk score and the final risk total score with respective corresponding set risk thresholds, and performing early warning according to comparison results.
9. A server, comprising a processor, a memory, and a communication bus;
the communication bus is used for realizing connection communication between the processor and the memory;
the processor is configured to execute one or more programs stored in the memory to implement the steps of the loan risk monitoring method according to any of claims 1 to 7.
10. A storage medium storing one or more programs, the one or more programs being executable by one or more processors to perform the steps of the loan risk monitoring method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111582409.5A CN114331671A (en) | 2021-12-22 | 2021-12-22 | Loan risk monitoring method and device, server and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111582409.5A CN114331671A (en) | 2021-12-22 | 2021-12-22 | Loan risk monitoring method and device, server and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114331671A true CN114331671A (en) | 2022-04-12 |
Family
ID=81055451
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111582409.5A Pending CN114331671A (en) | 2021-12-22 | 2021-12-22 | Loan risk monitoring method and device, server and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114331671A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115456753A (en) * | 2022-09-07 | 2022-12-09 | 安徽省优质采科技发展有限责任公司 | Enterprise credit information analysis method and system for bidding platform |
| CN115578007A (en) * | 2022-10-12 | 2023-01-06 | 南京数聚科技有限公司 | Method and system for integrating calculation of points and task in tax industry |
-
2021
- 2021-12-22 CN CN202111582409.5A patent/CN114331671A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115456753A (en) * | 2022-09-07 | 2022-12-09 | 安徽省优质采科技发展有限责任公司 | Enterprise credit information analysis method and system for bidding platform |
| CN115578007A (en) * | 2022-10-12 | 2023-01-06 | 南京数聚科技有限公司 | Method and system for integrating calculation of points and task in tax industry |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11868941B2 (en) | Task-level answer confidence estimation for worker assessment | |
| Herrero-Lopez | Social interactions in P2P lending | |
| US7720732B2 (en) | System and method for generating transaction based recommendations | |
| CN111861698B (en) | Pre-loan approval early warning method and system based on loan multi-head data | |
| CN114331671A (en) | Loan risk monitoring method and device, server and storage medium | |
| US20220004923A1 (en) | Systems and methods for model explanation | |
| CN115545886A (en) | Overdue risk identification method, overdue risk identification device, overdue risk identification equipment and storage medium | |
| Barbon et al. | NFT Bubbles | |
| Li et al. | Credit Risk management of P2P network Lending | |
| Xu et al. | Research on credit risk assessment of listed companies in technology sector based on support vector machine integration | |
| Han | Semi-supervised learning classification based on generalized additive logistic regression for corporate credit anomaly detection | |
| CN115034685A (en) | Customer value evaluation method, customer value evaluation device and computer-readable storage medium | |
| CN115619270A (en) | Data asset management method, device, computer equipment and storage medium | |
| Kasaian et al. | Understanding the difference between opportunistic and needy customers’ borrowing behavior: a new approach to segment current credit card customers | |
| CN117252677A (en) | Credit line determination method and device, electronic equipment and storage medium | |
| CN114626940A (en) | Data analysis method and device and electronic equipment | |
| CN113506173A (en) | Credit risk assessment method and related equipment thereof | |
| Karminsky et al. | Comparison of bank financial stability factors in CIS countries | |
| Lu et al. | Predicting P2P lenders' decisions: the prospect theory approach | |
| Andreevna | The Investment Analysis of IT Companies: A Case Study of Yandex | |
| Ghasemkhani et al. | Actions Speak Louder Than Words: Imputing Users' Reputation from Transaction History | |
| TWI776370B (en) | Investment risk scoring method and system for fund commodities | |
| Brezigar Masten et al. | Predicting bankruptcy with semi-parametric single-index model | |
| CN117114812A (en) | Financial product recommendation method and device for enterprises | |
| Tang | The Influence of Managerial Ability on Corporate Financialization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |