CN113807943A - Multi-factor valuation method, system, medium and equipment for bad assets - Google Patents
Multi-factor valuation method, system, medium and equipment for bad assets Download PDFInfo
- Publication number
- CN113807943A CN113807943A CN202110951468.9A CN202110951468A CN113807943A CN 113807943 A CN113807943 A CN 113807943A CN 202110951468 A CN202110951468 A CN 202110951468A CN 113807943 A CN113807943 A CN 113807943A
- Authority
- CN
- China
- Prior art keywords
- repayment
- asset
- assets
- bad
- data
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention provides a multi-factor valuation method, a multi-factor valuation system, a multi-factor valuation medium and a multi-factor valuation device for bad assets. According to the invention, the valuation of the asset fund return is carried out by utilizing multiple influence factors related to the bad assets, so that the valuation accuracy for individual single loan is greatly improved, and the matching requirement of a diversified collection urging scheme can be met.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a multi-factor valuation method, a system, a medium and equipment for bad assets.
Background
The poor assets refer to assets which cannot be dealt with by a borrower according to time and volume or credit enterprises in the field of financial credit, such as mortgage houses and the like. In order to reduce the economic loss caused by the poor assets, credit enterprises utilize a big data method to estimate the asset value of the poor asset data so as to accurately customize the collection promotion scheme.
At present, no mature solution is available for personal bad asset valuation, the overall package can be roughly valued only by means of recovery experiences in relevant fields such as public bad asset recovery, overdue credit card recovery and the like, the influence of factors such as repayment willingness, repayment capacity and overdue single-loan duration of a single-loan repayment person on the final single-loan repayment rate cannot be fully considered, and the personalized collection urging scheme is difficult to customize subsequently to maximize the debt repayment rate.
Disclosure of Invention
In view of the above problems, the invention provides a multi-factor valuation method, a system, a medium and equipment for bad assets, which perform multi-factor evaluation on repayment willingness, repayment capacity and discount rate of borrowers, and make a selectable fraud filtering strategy to finally obtain a comprehensive repayment index and a pre-estimated recovery amount of a single loan.
According to a first aspect of the present invention, there is provided a multi-factor valuation method of bad assets, comprising:
acquiring asset characteristic data of the bad assets to be processed;
performing data preprocessing on the asset characteristic data to obtain processed characteristic data;
performing feature prediction processing on the feature data by using the trained feature prediction model to obtain a repayment willingness evaluation value, a repayment capacity evaluation value and an asset reduction rate aiming at the unhealthy assets;
and carrying out refund calculation on the bad assets according to the repayment wish valuation, the repayment capacity valuation and the asset discount rate to obtain the refund valuation of the bad assets.
Optionally, the calculating the refund of the undesirable asset according to the repayment willingness valuation, the repayment ability valuation and the asset discount rate to obtain the refund valuation of the undesirable asset includes:
calculating a comprehensive repayment index of the bad assets according to the repayment wish valuation, the repayment capacity valuation and the asset discount rate, wherein the calculation formula is as follows:
G(x)＝H(W(x),V(x))*(1-P(x))
wherein x is a numerical characteristic of the user; g (x) is the bad asset comprehensive repayment index for user x; h (W (x), V (x)) is the ideal refund rate of the bad assets of the user x; w (x) valuation of repayment willingness for user x for the undesirable asset; v (x) estimates for user x's repayment capacity for the undesirable asset; p (x) is the rate of reduction of bad assets for user x;
calculating the refund valuation of the bad assets according to the comprehensive repayment index and the amount to be paid corresponding to the bad assets, wherein the calculation formula is as follows:
R(x)＝G(x)*B(x)
wherein R (x) is an estimate of refund for user x's bad assets; b (x) a refund amount corresponding to the bad asset of user x.
Optionally, before performing the feature prediction processing on the feature data according to the trained feature prediction model, the method further includes:
constructing a characteristic prediction model; the characteristic prediction model comprises a repayment willingness prediction submodel, a repayment capacity prediction submodel and an asset reduction rate prediction submodel;
acquiring original characteristic data as model training sample data; the original characteristic data is asset characteristic data of original bad assets which are collected in advance and subjected to data preprocessing;
and performing model training on the repayment willingness prediction submodel and the repayment ability prediction submodel according to the model training sample data to obtain a trained feature prediction model.
Optionally, the asset characteristic data comprises at least one of basic data, asset data, financial attribute data, and borrowing data of the user;
the data preprocessing at least comprises one of numerical processing, characteristic normalization processing, abnormal value processing and characteristic screening processing.
Optionally, the constructing a feature prediction model includes:
constructing the repayment willingness prediction submodel and the repayment capacity prediction submodel based on a deep learning algorithm;
the repayment intention forecasting submodel is used for forecasting a repayment intention valuation aiming at the bad asset according to the characteristic data, and the repayment ability forecasting submodel is used for forecasting a repayment ability valuation aiming at the bad asset according to the characteristic data;
and the number of the first and second groups,
constructing the asset reduction rate prediction sub-model based on asset reduction rules;
wherein the asset reduction rate predictor model is used for predicting the asset reduction rate of the bad assets; the asset reduction rule is a preset rule according to the overdue time of repayment of the bad asset, and the shorter the overdue time of repayment of the bad asset is, the smaller the asset reduction rate is; the longer the overdue duration of the repayment of the bad asset is, the larger the reduction rate of the asset is.
Optionally, the performing model training on the repayment willingness prediction submodel and the repayment ability prediction submodel according to the model training sample data includes:
matching corresponding model attribute information to the model training sample data according to the user age and/or the amount to be returned in the model training sample data based on a preset feature level division rule;
adding corresponding repayment willingness valuation and repayment capacity valuation to the model training sample data according to the model attribute information;
performing model training on the repayment intention forecasting submodel by using the model training sample data and the corresponding repayment intention estimation value; and
and performing model training on the repayment ability prediction submodel by using the model training sample data and the corresponding repayment ability estimation value.
Optionally, the method further comprises:
and carrying out fraud analysis on the asset characteristic data of the bad assets to be processed, and selecting fraud assets in the bad assets, wherein the fraud assets specifically comprise:
and judging whether the user behavior exists at least one of the following behaviors according to the asset characteristic data: borrowing and lending multi-identity cards, excessive debt, multi-head borrowing, serious overdue, new borrowing and old borrowing, and lost connection;
and if so, the bad assets are fraudulent assets, and the return money evaluation of the fraudulent assets is ignored.
According to a second aspect of the present invention, there is provided a multi-factor valuation system for undesirable assets, comprising:
the data acquisition module is used for acquiring asset characteristic data of the to-be-processed bad assets;
the data processing module is used for carrying out data preprocessing on the asset characteristic data to obtain processed characteristic data;
the model prediction module is used for performing feature prediction processing on the feature data by using the trained feature prediction model to obtain repayment willingness valuation, repayment capacity valuation and asset discount rate aiming at the undesirable assets;
and the repayment calculation module is used for carrying out repayment calculation on the bad assets according to the repayment wish valuation, the repayment capacity valuation and the asset discount rate to obtain the repayment valuation of the bad assets.
According to a third aspect of the present invention, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, carries out the steps of a method for multi-factor valuation of poor assets according to any one of the first aspect of the present invention.
According to a fourth aspect of the present invention, there is provided a computer apparatus comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor when executing the computer program implementing the steps of the method for multi-factor estimation of undesirable assets according to any one of the first aspect of the present invention.
According to the scheme, by obtaining asset characteristic data of the to-be-processed bad asset, data preprocessing is carried out on the asset characteristic data to obtain processed characteristic data, a trained characteristic prediction model is used for carrying out characteristic prediction processing on the characteristic data to obtain repayment willingness valuation, repayment capacity valuation and asset reduction rate aiming at the bad asset, and the refund calculation is carried out on the bad asset according to the repayment willingness valuation, the repayment capacity valuation and the asset reduction rate to obtain the refund valuation of the bad asset. And the valuation is carried out on the asset refund by utilizing the multiple influence factors related to the bad assets, so that the valuation accuracy of individual single loan is improved, and the matching requirement of a diversified collection urging scheme is met.
The foregoing description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the above and other objects, features, and advantages of the present invention more clearly understandable.
The above and other objects, advantages and features of the present invention will become more apparent to those skilled in the art from the following detailed description of specific embodiments thereof, taken in conjunction with the accompanying drawings.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
FIG. 1 is a flow diagram illustrating a multi-factor valuation method for undesirable assets according to an embodiment of the present invention;
FIG. 2 is a schematic diagram illustrating a multi-factor valuation system for undesirable assets according to an embodiment of the present invention;
FIG. 3 illustrates a schematic structural diagram of a multi-factor valuation system for undesirable assets according to an embodiment of the present invention;
fig. 4 shows a physical structure diagram of a computer device according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
The embodiment of the invention provides a multi-factor estimation method for bad assets, as shown in fig. 1, the method at least comprises the following steps S101 to S104:
and step S101, acquiring asset characteristic data of the bad assets to be processed.
Asset characteristic data in embodiments of the present invention are items of information data that characterize an individual user as having been generated in a single credit transaction and that are related to repayment information. Before asset characteristic data of the bad assets to be processed is obtained, the bad asset information to be processed needs to be obtained first, the bad asset information can comprise basic information of a borrower and detailed information of a single borrowing, wherein the basic information of the borrower can comprise a name of the borrower, a mobile phone number, an identity card number and the like, and the detailed information of the single borrowing can comprise information of money to be returned, interest to be returned, a date to be successfully borrowed, a total amount to be returned and the like.
Further, asset characterization data associated with the undesirable asset information to be processed according to the above is extended. The asset characteristic data may include, among other things, the user's underlying data, asset data, financial attribute data, borrowing data, and the like. The basic data of the user is data of the age, sex, marital status, academic history, work and the like of the user, the asset data is whether a house exists, whether a car exists, the number of cars in the house and the like, the financial attribute data is attribute content used for representing financial transactions of the user, and comprises credit situations such as public deposit, wages, financing, bank card collection, credit cards and network loan and repayment situations, and the borrowing data is borrowing amount, borrowing interest, returned amount, unretained amount, overdue time and the like.
Since the asset characteristic data cannot be directly obtained, in the embodiment of the present invention, the current execution end performs agreement cooperation with a plurality of enterprise platforms with credit functions to accurately obtain the asset characteristic data of the bad assets of the user.
And S102, performing data preprocessing on the asset characteristic data to obtain processed characteristic data.
In order to further improve the processing efficiency of the asset characteristic data, the asset characteristic data is subjected to data preprocessing, which may include a digitization processing, a characteristic normalization processing, an abnormal value processing, a characteristic screening processing, and the like.
The digitization processing is to convert data such as texts into numerical data so as to be used as input of model training for training operation. The feature normalization processing means that after the digitization processing, in order to eliminate the dimensional influence between the asset feature data, the data is normalized so that the asset feature data are in the same order of magnitude, and the feature normalization processing may be performed by a linear scale conversion method, a 0-mean normalization method, or the like. In the process, in order to avoid the influence of abnormal data on the model training efficiency, abnormal value processing can be performed, namely, the numerical values in the abnormal range are screened out, and data with stable numerical value distribution are reserved. In addition, in order to improve subsequent data processing efficiency, feature screening can be performed, some asset feature data with more null values are deleted, and more effective feature data are screened out according to feature importance, feature correlation and the like. Wherein, the characteristic data with stronger characteristic importance or element can be screened out through models such as random forest, xgboost and the like; the Pearson correlation coefficient and the Spearman correlation coefficient may also be used to screen feature data with strong correlation, and the like, and the embodiment of the present invention is not particularly limited.
And preprocessing the asset characteristic data to obtain characteristic data, wherein the characteristic data is used for subsequently inputting a characteristic prediction model to perform characteristic prediction processing, and calculating the repayment willingness valuation, the repayment capacity valuation and the asset discount rate of the bad asset.
And step S103, performing characteristic prediction processing on the characteristic data by using the trained characteristic prediction model to obtain repayment willingness estimation, repayment capacity estimation and asset discount rate aiming at the unhealthy assets.
In the embodiment of the present invention, before performing the feature prediction processing on the feature data, a feature prediction model needs to be constructed and trained, which may specifically include steps S103-1 to S103-3:
step S103-1: and constructing a characteristic prediction model.
The characteristic prediction model comprises a repayment willingness prediction submodel, a repayment capacity prediction submodel and an asset reduction rate prediction submodel.
Specifically, a repayment willingness prediction submodel and a repayment capacity prediction submodel can be constructed based on a deep learning algorithm; the repayment intention prediction submodel is used for predicting repayment intention valuation aiming at the bad assets according to the characteristic data, and the repayment ability prediction submodel is used for predicting repayment ability valuation aiming at the bad assets according to the characteristic data.
That is, a repayment willingness prediction submodel and a repayment ability prediction submodel are constructed through a selected deep learning algorithm, including but not limited to xgboost, random forest, linear regression, and the like. Preferably, in the embodiment of the present invention, xgboost may be selected to construct a repayment willingness prediction sub-model and a repayment ability prediction sub-model, xgboost is an integrated model, a basic model of the model is a cart tree structure, and a residual value of a previous tree is fitted by gradually generating a new cart tree to form a final model.
In the embodiment of the invention, the repayment intention of the user for the undesirable asset is a subjective factor influencing the recoverable range of the undesirable asset, and the higher the estimated value of the repayment intention of the undesirable asset is, the higher the repayment intention of the user for the undesirable asset is, the higher the estimated value of the repayment intention of the user for the undesirable asset is; the repayment capacity of the user for the bad assets is an objective factor influencing the recoverable range of the bad assets, and the higher the estimated value of the repayment capacity of the bad assets is predicted by the repayment capacity prediction submodel, the higher the repayment capacity of the user for the bad assets is.
Further, constructing an asset reduction rate prediction sub-model based on asset reduction rules; the asset reduction rate prediction sub-model is used for predicting the asset reduction rate of the bad assets; the asset reduction rule is a preset rule according to the overdue time of the repayment of the bad asset, and the shorter the overdue time of the repayment of the bad asset is, the smaller the asset reduction rate is; the longer the overdue duration of the repayment of the bad asset, the greater the asset discount rate.
The asset reduction rate is used for measuring the reduction condition of the undesirable assets, the value is 0 or decimal between 0 and 1, the higher the asset reduction rate is, the larger the asset reduction degree is, and the asset reduction rule can be as follows: when the overdue time length of repayment of the bad asset is [0, 30 ], the asset discount rate is 0; when the overdue time of repayment of the bad assets is [30, 90 ], the asset discount rate is 0.1; when the overdue time of repayment of the bad assets is [90, 180 ], the asset discount rate is 0.2; when the overdue time length of repayment of the bad asset is [180, 300 ], the asset reduction rate is 0.4.
The asset reduction rule provided in the embodiment of the invention is obtained according to expert experience, and in practical application, the asset reduction rule can be set according to actual requirements, which is not limited by the invention.
Step S103-2: acquiring original characteristic data as model training sample data; the original characteristic data is the asset characteristic data of the original bad assets which are collected in advance and subjected to data preprocessing.
In the embodiment of the present invention, before training the feature prediction model, training sample data for the model needs to be acquired first. The property characteristic data of the original bad assets can be collected in advance, and the original bad assets are not the above-mentioned bad assets to be processed but the related bad assets which are collected for individual single loan. The original characteristic data is the asset characteristic data of original bad assets, and comprises user basic data, asset data, financial attribute data, borrowing data and the like which are subjected to data preprocessing. The data preprocessing process is the same as the data preprocessing method mentioned above, and may include a digitization processing, a feature normalization processing, an abnormal value processing, a feature screening processing, and the like.
The original characteristic data after data preprocessing can be used as model training sample data to train the characteristic prediction model.
Step S103-3: and performing model training on the repayment willingness prediction submodel and the repayment capacity prediction submodel according to model training sample data to obtain a trained feature prediction model.
Specifically, the corresponding model attribute information can be matched with the model training sample data according to the user age and the amount to be paid in the model training sample data based on a preset feature level division rule; adding corresponding repayment willingness expert experience values and repayment ability expert experience values to the model training sample data according to the model attribute information; performing model training on the repayment intention prediction submodel by using the model training sample data and the corresponding repayment intention expert experience value; and performing model training on the repayment ability prediction submodel by using the model training sample data and the corresponding repayment ability expert experience value.
The model training sample data is matched with corresponding model attribute information based on a preset feature level division rule according to the age of a user and the amount to be paid in the model training sample data, and it can be understood that the model attribute information is obtained by analyzing the age information of the user and the amount to be paid, the emphasis feature information is used for extracting the emphasis feature data in the model training sample data so as to evaluate the repayment willingness expert experience value and the repayment ability expert experience value of the user, and the obtained expert experience value is the model target value used in model training.
The characteristic hierarchy dividing rule provided by the embodiment of the invention is obtained according to expert experience. For example, the user ages are classified into 5 ages, wherein the 5 ages are 18 years or more and less than 22 years old (academic calendar and work are unstable), 22 years or more and less than 27 years old (academic calendar and work properties are basically stable), 27 years or more and 50 years old (mostly enter marriage and child education stages), 50 years or more and less than 65 years old (influence of child education factors is reduced), 65 years old or more (residual loans need to be returned, the loans are beyond the loan application age, and the loans cannot be continuously borrowed); and carrying out hierarchical division on the amount to be refunded to obtain the hierarchical division of 3 amounts to be refunded of a low-amount layer, a medium-amount layer and a high-amount layer. And corresponding emphasis characteristic information is set for the age level and the amount to be paid level of a specific user.
Analyzing the emphasis characteristic information aiming at the repayment willingness, wherein when the age level of the user is more than or equal to 18 years and less than 22 years, and the to-be-refunded amount level is in a low-amount level, the emphasis characteristic information is the active repayment condition of the user on each platform in the near future; when the user is in the same age level and the to-be-paid amount level is in a medium or high-amount level, the emphasis on the characteristic information is to emphasize the active payment condition and the overdue payment condition of the user on each platform in the near future. For another example, when the age hierarchy of the user is greater than or equal to 22 years and less than 27 years, the work condition and the academic history condition can be added to the emphasis feature information needing to be emphasized; when the user age level is greater than or equal to 27 years and less than 50 years, the work condition, the marital condition and the child condition can be added into the emphasis feature information needing to be emphasized; when the user age hierarchy is greater than 65 years old, the retirement status may be added to the emphasis feature information that needs to be focused on, and so on.
After the emphasis characteristic information of the payment willingness of each level of user is obtained, emphasis characteristic data are extracted from model training sample data according to the emphasis characteristic information, and the emphasis characteristic data can comprise the active payment condition, the overdue payment condition, the working condition, the academic condition, the marital condition, the child condition, the retired fund condition and the like of the user on each platform. And evaluating the repayment intention of the user according to the emphasis feature data to obtain a repayment intention expert experience value, wherein the obtained repayment intention expert experience value is used as a model target value during the repayment intention model training. The higher the repayment wish of the user is, the higher the experience value of a repayment wish expert is, and the value range is 0-1.
Further, aiming at analysis of the emphasis characteristic information of the repayment willingness, when the age level of the user is more than or equal to 18 years and less than 22 years, and the amount to be refunded is in a low-sum level, the emphasis characteristic information is the payment and payment condition of the paying user; when the users are in the same age level and the to-be-paid amount level is in the middle-amount level, the emphasis on the characteristic information is to pay attention to the collection and payment condition of the users and the payment condition of each platform; when the users are in the same age level and the to-be-paid amount level is in a high-value level, the emphasis characteristic information is the user collection and payment condition, the payment condition of each platform and the existing asset condition; when the age level of the user is more than or equal to 22 years and less than 27 years, the working condition and the academic condition can be added into the emphasis characteristic information needing to be emphasized; when the user age hierarchy is greater than or equal to 27 years and less than 50 years, the work situation, the marital status, the child status, and the like may be added to the emphasis feature information that needs to be emphasized.
After the emphasis characteristic information of the repayment capacity of each level of users is obtained, emphasis characteristic data are extracted from model training sample data according to the emphasis characteristic information, and the emphasis characteristic data can comprise the payment and payment conditions, the translational repayment conditions, the working conditions, the academic conditions, the marital conditions, the child conditions, the existing asset conditions and the like of the users. And evaluating the repayment capacity of the user according to the emphasis feature data to obtain a repayment capacity expert experience value, wherein the obtained repayment capacity expert experience value is used as a model target value during the repayment capacity model training. The stronger the repayment ability of the user is, the higher the experience value of the repayment ability expert is, and the value range is 0-1.
The evaluation of the feature level division rule, the repayment intention expert experience value and the repayment ability expert experience value provided by the embodiment of the invention is completed through expert experience setting, and in practical application, the setting of the feature level division rule, the repayment intention expert experience value and the repayment ability expert experience value can be set according to practical situations, which is not limited by the invention.
After a repayment intention expert experience value and a repayment ability expert experience value corresponding to model training sample data are obtained, taking the model training sample data as a model input value, taking the corresponding repayment intention expert experience value or the corresponding repayment ability expert experience value as a model target value, dividing a training set, a testing set and a verification set for the model input value and the model target value, carrying out model training, testing and verification on a repayment intention prediction submodel and a repayment ability prediction submodel until the repayment intention prediction submodel and the repayment ability prediction submodel reaching a preset accuracy rate are obtained, and finishing training. The trained repayment intention forecasting submodel is used for estimating a repayment intention valuation of the user according to the input characteristic data; and the trained repayment ability prediction submodel is used for estimating the repayment ability estimation value of the user according to the input characteristic data.
And step S104, carrying out refund calculation on the bad assets according to the repayment intention estimation value, the repayment capacity estimation value and the asset reduction rate estimated by the model to obtain the refund estimation value of the bad assets.
Calculating the comprehensive repayment index of the bad assets according to the repayment wish valuation, the repayment capacity valuation and the asset discount rate, wherein the calculation formula is as follows:
G(x)＝H(W(x),V(x))*(1-P(x))
wherein x is the numerical characteristic of the user and represents the user x; g (x) is the bad asset comprehensive repayment index for user x; h (W (x), V (x)) is the ideal refund rate of the bad assets of the user x; p (x) is the rate of reduction of bad assets for user x;
the calculation formula of the ideal refund rate of the bad assets is as follows:
wherein W (x) estimates the repayment willingness of the user x for the undesirable asset; v (x) estimates for user x's repayment capacity for the undesirable asset;
calculating the refund valuation of the bad assets according to the comprehensive repayment index and the amount to be refunded corresponding to the bad assets, wherein the calculation formula is as follows:
R(x)＝G(x)*B(x)
wherein R (x) is an estimate of refund for user x's bad assets; b (x) a refund amount corresponding to the bad asset of user x.
Optionally, the multi-factor valuation method for the bad assets provided in the embodiment of the present invention may further perform fraud analysis on the asset feature data of the bad assets to be processed, and select a fraud asset in the bad assets, specifically including: and judging whether the user behavior exists at least one of the following behaviors according to the asset characteristic data: borrowing and lending multi-identity cards, excessive debt, multi-head borrowing, serious overdue, new borrowing and old borrowing, and lost connection; and if so, the bad asset is a fraud asset, and the return money evaluation of the fraud asset is ignored.
That is, for an overall asset pack of different types of people, a fraud filtering policy may be used to make a conservative estimate of the asset recovery of the overall asset pack, i.e., the recovery involving fraudulent assets is ignored in calculating the overall asset pack estimate. Fraudulent activities may include multiple identity card loans, excess liability, long-term loans, severe overdue, new loans and old loans, unleashed, etc.
The multi-identity card lending refers to the lending behavior of binding different identity cards with the same mobile phone number; excessive liability means that the user's recent income cannot support his debt; multi-headed loan refers to the act of a single borrower submitting a loan demand to two or more financial institutions; the serious overdue means that the overdue time of repayment exceeds a certain number of days; the loan repayment refers to the behavior that the loan cannot be timely withdrawn in the form of money after the loan is due, and the loan is re-issued for returning part or all of the original loan; the lost connection means that the borrowing mobile phone number is currently a null number.
The embodiment of the invention provides a multi-factor valuation method of bad assets, which comprises the steps of obtaining asset characteristic data of the bad assets to be processed, carrying out data preprocessing on the asset characteristic data to obtain processed characteristic data, carrying out characteristic prediction processing on the characteristic data by using a trained characteristic prediction model to obtain repayment intention valuation, repayment ability valuation and asset discount rate aiming at the bad assets, and carrying out refund calculation on the bad assets according to the repayment intention valuation, the repayment ability valuation and the asset discount rate to obtain the refund valuation of the bad assets. Modeling repayment willingness and repayment capacity of the repayment person by using characteristics such as sample data repayment person loan information, social attributes and economic conditions, obtaining an asset discount rate model by depending on expert experience, obtaining indexes such as the repayment willingness, the repayment capacity and the discount rate of the repayment person in each loan in a newly-entered financial package according to the model, obtaining a comprehensive debt repayment index of a debtor, and finally realizing the prediction of the predicted payable amount of a single debt right and the prediction of the whole package recovery amount of a bad financial package by combining with an optional fraud filtering strategy. The valuation accuracy of the individual single borrowing and whole packet recovery amount is improved, and the matching requirement of diversified collection urging schemes is met.
Further, as a specific implementation of fig. 1, an embodiment of the present invention provides a system for multi-factor valuation of bad assets, as shown in fig. 2, the apparatus may include: a data acquisition module 210, a data processing module 220, a model prediction module 230, and a refund calculation module 240.
The data acquisition module 210 may be configured to acquire asset characterization data of the undesirable asset to be processed.
The data processing module 220 may be configured to perform data preprocessing on the asset feature data to obtain processed feature data.
The model prediction module 230 may be configured to perform feature prediction processing on the feature data by using the trained feature prediction model, so as to obtain a repayment willingness estimation, a repayment capability estimation, and an asset discount rate for the undesirable asset.
And the refund calculating module 240 can be used for carrying out refund calculation on the bad assets according to the repayment wish valuation, the repayment capacity valuation and the asset discount rate to obtain the refund valuation of the bad assets.
Optionally, the repayment calculation module 240 may be further configured to calculate a comprehensive repayment index of the undesirable asset according to the repayment wish valuation, the repayment ability valuation and the asset discount rate, where the calculation formula is:
G(x)＝H(W(x),V(x))*(1-P(x))
wherein x is a numerical characteristic of the user; g (x) is the bad asset comprehensive repayment index for user x; h (W (x), V (x)) is the ideal refund rate of the bad assets of the user x; w (x) valuation of repayment willingness for user x for the undesirable asset; v (x) estimates for user x's repayment capacity for the undesirable asset; p (x) is the rate of reduction of bad assets for user x;
calculating the refund valuation of the bad assets according to the comprehensive repayment index and the amount to be refunded corresponding to the bad assets, wherein the calculation formula is as follows:
R(x)＝G(x)*B(x)
wherein R (x) is an estimate of refund for user x's bad assets; b (x) a refund amount corresponding to the bad asset of user x.
Optionally, as shown in fig. 3, the system for multi-factor estimation of undesirable assets according to an embodiment of the present invention may further include: a model building module 250 and a fraud analysis module 260.
A model construction module 250, which may be used to construct a feature prediction model; the characteristic prediction model comprises a repayment willingness prediction submodel, a repayment capacity prediction submodel and an asset reduction rate prediction submodel;
acquiring original characteristic data as model training sample data; the original characteristic data is the asset characteristic data of original bad assets which are collected in advance and subjected to data preprocessing;
and performing model training on the repayment willingness prediction submodel and the repayment capacity prediction submodel according to model training sample data to obtain a trained feature prediction model.
Optionally, the model building module 250 may also be configured to build a repayment willingness prediction submodel and a repayment ability prediction submodel based on a deep learning algorithm;
the repayment intention prediction submodel is used for predicting repayment intention valuation aiming at the bad assets according to the characteristic data, and the repayment ability prediction submodel is used for predicting repayment ability valuation aiming at the bad assets according to the characteristic data;
and the number of the first and second groups,
constructing an asset reduction rate prediction sub-model based on asset reduction rules;
the asset reduction rate prediction sub-model is used for predicting the asset reduction rate of the bad assets; the asset reduction rule is a preset rule according to the overdue time of the repayment of the bad asset, and the shorter the overdue time of the repayment of the bad asset is, the smaller the asset reduction rate is; the longer the overdue duration of the repayment of the bad asset is, the greater the reduction rate of the asset is;
matching corresponding model attribute information to the model training sample data according to the user age and/or the amount to be paid in the model training sample data based on a preset feature level division rule;
adding corresponding repayment willingness valuation and repayment capacity valuation to model training sample data according to the model attribute information;
carrying out model training on the repayment willingness prediction submodel by utilizing the model training sample data and the corresponding repayment willingness estimation value; and
and performing model training on the repayment ability prediction submodel by using the model training sample data and the corresponding repayment ability estimation value.
Optionally, the fraud analysis module 260 may be configured to perform fraud analysis on the asset characteristic data of the to-be-processed bad assets, and select a fraud asset from the bad assets, where the fraud analysis module specifically includes:
and judging whether the user behavior exists at least one of the following behaviors according to the asset characteristic data: borrowing and lending multi-identity cards, excessive debt, multi-head borrowing, serious overdue, new borrowing and old borrowing, and lost connection;
and if so, the bad asset is a fraud asset, and the return money evaluation of the fraud asset is ignored.
It should be noted that other corresponding descriptions of the functional modules involved in the multi-factor valuation system for poor assets provided by the embodiment of the present invention may refer to the corresponding descriptions of the method shown in fig. 1, and are not described herein again.
Based on the method shown in fig. 1, correspondingly, the embodiment of the invention further provides a computer-readable storage medium, on which a computer program is stored, and the computer program, when executed by a processor, implements the steps of the method for multi-factor valuation of poor assets according to any of the embodiments.
Based on the above embodiments of the method shown in fig. 1 and the apparatus shown in fig. 3, an embodiment of the present invention further provides an entity structure diagram of a computer device, as shown in fig. 4, the computer device may include a communication bus, a processor, a memory, and a communication interface, and may further include an input/output interface and a display device, where the functional units may complete communication with each other through the bus. The memory stores computer programs, and the processor is used for executing the programs stored in the memory and executing the steps of the multi-factor estimation method of the bad assets described in the embodiment.
It is clear to those skilled in the art that the specific working processes of the above-described systems, devices, modules and units may refer to the corresponding processes in the foregoing method embodiments, and for the sake of brevity, further description is omitted here.
In addition, the functional units in the embodiments of the present invention may be physically independent of each other, two or more functional units may be integrated together, or all the functional units may be integrated in one processing unit. The integrated functional units may be implemented in the form of hardware, or in the form of software or firmware.
Those of ordinary skill in the art will understand that: the integrated functional units, if implemented in software and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computing device (e.g., a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention when the instructions are executed. And the aforementioned storage medium includes: u disk, removable hard disk, Read Only Memory (ROM), Random Access Memory (RAM), magnetic or optical disk, and other various media capable of storing program code.
Alternatively, all or part of the steps of implementing the foregoing method embodiments may be implemented by hardware (such as a computing device, e.g., a personal computer, a server, or a network device) associated with program instructions, which may be stored in a computer-readable storage medium, and when the program instructions are executed by a processor of the computing device, the computing device executes all or part of the steps of the method according to the embodiments of the present invention.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments can be modified or some or all of the technical features can be equivalently replaced within the spirit and principle of the present invention; such modifications or substitutions do not depart from the scope of the present invention.
Claims (10)
1. A method for multi-factor valuation of undesirable assets, comprising:
acquiring asset characteristic data of the bad assets to be processed;
performing data preprocessing on the asset characteristic data to obtain processed characteristic data;
performing feature prediction processing on the feature data by using the trained feature prediction model to obtain a repayment willingness evaluation value, a repayment capacity evaluation value and an asset reduction rate aiming at the unhealthy assets;
and carrying out refund calculation on the bad assets according to the repayment wish valuation, the repayment capacity valuation and the asset discount rate to obtain the refund valuation of the bad assets.
2. The method of claim 1, wherein calculating a refund for the undesirable asset based on the estimate of the willingness to repay, the estimate of the ability to repay, and the reduction rate of the asset to obtain an estimate of the refund for the undesirable asset comprises:
calculating a comprehensive repayment index of the bad assets according to the repayment wish valuation, the repayment capacity valuation and the asset discount rate, wherein the calculation formula is as follows:
G(x)＝H(W(x),V(x))*(1-P(x))
wherein x is a numerical characteristic of the user; g (x) is the bad asset comprehensive repayment index for user x; h (W (x), V (x)) is the ideal refund rate of the bad assets of the user x; w (x) valuation of repayment willingness for user x for the undesirable asset; v (x) estimates for user x's repayment capacity for the undesirable asset; p (x) is the rate of reduction of bad assets for user x;
calculating the refund valuation of the bad assets according to the comprehensive repayment index and the amount to be paid corresponding to the bad assets, wherein the calculation formula is as follows:
R(x)＝G(x)*B(x)
wherein R (x) is an estimate of refund for user x's bad assets; b (x) a refund amount corresponding to the bad asset of user x.
3. The method of claim 1, wherein before performing the feature prediction processing on the feature data according to the trained feature prediction model, the method further comprises:
constructing a characteristic prediction model; the characteristic prediction model comprises a repayment willingness prediction submodel, a repayment capacity prediction submodel and an asset reduction rate prediction submodel;
acquiring original characteristic data as model training sample data; the original characteristic data is asset characteristic data of original bad assets which are collected in advance and subjected to data preprocessing;
and performing model training on the repayment willingness prediction submodel and the repayment ability prediction submodel according to the model training sample data to obtain a trained feature prediction model.
4. The method according to claim 1 or 3,
the asset characteristic data comprises at least one of basic data, asset data, financial attribute data and borrowing data of a user;
the data preprocessing at least comprises one of numerical processing, characteristic normalization processing, abnormal value processing and characteristic screening processing.
5. The method of claim 3, wherein constructing the feature prediction model comprises:
constructing the repayment willingness prediction submodel and the repayment capacity prediction submodel based on a deep learning algorithm;
the repayment intention forecasting submodel is used for forecasting a repayment intention valuation aiming at the bad asset according to the characteristic data, and the repayment ability forecasting submodel is used for forecasting a repayment ability valuation aiming at the bad asset according to the characteristic data;
and the number of the first and second groups,
constructing the asset reduction rate prediction sub-model based on asset reduction rules;
wherein the asset reduction rate predictor model is used for predicting the asset reduction rate of the bad assets; the asset reduction rule is a preset rule according to the overdue time of repayment of the bad asset, and the shorter the overdue time of repayment of the bad asset is, the smaller the asset reduction rate is; the longer the overdue duration of the repayment of the bad asset is, the larger the reduction rate of the asset is.
6. The method of claim 4, wherein the model training of the willingness to repayment prediction submodel and the repayment ability prediction submodel according to the model training sample data comprises:
matching corresponding model attribute information to the model training sample data according to the user age and/or the amount to be returned in the model training sample data based on a preset feature level division rule;
adding corresponding repayment willingness valuation and repayment capacity valuation to the model training sample data according to the model attribute information;
performing model training on the repayment intention forecasting submodel by using the model training sample data and the corresponding repayment intention estimation value; and
and performing model training on the repayment ability prediction submodel by using the model training sample data and the corresponding repayment ability estimation value.
7. The method according to any one of claims 1-6, comprising:
and carrying out fraud analysis on the asset characteristic data of the bad assets to be processed, and selecting fraud assets in the bad assets, wherein the fraud assets specifically comprise:
and judging whether the user behavior exists at least one of the following behaviors according to the asset characteristic data: borrowing and lending multi-identity cards, excessive debt, multi-head borrowing, serious overdue, new borrowing and old borrowing, and lost connection;
and if so, the bad assets are fraudulent assets, and the return money evaluation of the fraudulent assets is ignored.
8. A multi-factor valuation system for undesirable assets comprising:
the data acquisition module is used for acquiring asset characteristic data of the to-be-processed bad assets;
the data processing module is used for carrying out data preprocessing on the asset characteristic data to obtain processed characteristic data;
the model prediction module is used for performing feature prediction processing on the feature data by using the trained feature prediction model to obtain repayment willingness valuation, repayment capacity valuation and asset discount rate aiming at the undesirable assets;
and the repayment calculation module is used for carrying out repayment calculation on the bad assets according to the repayment wish valuation, the repayment capacity valuation and the asset discount rate to obtain the repayment valuation of the bad assets.
9. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method for multi-factor valuation of poor assets of any one of claims 1 to 7.
10. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor when executing the computer program implements the steps of the method for multi-factor valuation of bad assets of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110951468.9A CN113807943A (en) | 2021-08-18 | 2021-08-18 | Multi-factor valuation method, system, medium and equipment for bad assets |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110951468.9A CN113807943A (en) | 2021-08-18 | 2021-08-18 | Multi-factor valuation method, system, medium and equipment for bad assets |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113807943A true CN113807943A (en) | 2021-12-17 |
Family
ID=78941511
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110951468.9A Pending CN113807943A (en) | 2021-08-18 | 2021-08-18 | Multi-factor valuation method, system, medium and equipment for bad assets |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113807943A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115660811A (en) * | 2022-11-07 | 2023-01-31 | 杭州度言软件有限公司 | Asset management method for improving recovery rate of consumption financial overdue assets |
-
2021
- 2021-08-18 CN CN202110951468.9A patent/CN113807943A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115660811A (en) * | 2022-11-07 | 2023-01-31 | 杭州度言软件有限公司 | Asset management method for improving recovery rate of consumption financial overdue assets |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20100287093A1 (en) | System and Method for Collections on Delinquent Financial Accounts | |
| CN110458693A (en) | A kind of automatic measures and procedures for the examination and approval of business loan, device, storage medium and electronic equipment | |
| CN108665159A (en) | A kind of methods of risk assessment, device, terminal device and storage medium | |
| CN110503564B (en) | Security case processing method, system, equipment and storage medium based on big data | |
| CN110070430A (en) | Assess method and device, the storage medium, electronic equipment of refund risk | |
| CN110796539A (en) | Credit investigation evaluation method and device | |
| CN111708883A (en) | Credit credit limit determination method and device based on machine learning and equipment fingerprint | |
| CN113034263A (en) | Credit approval method, credit approval device, credit approval server and credit approval medium | |
| CN110363642A (en) | Loan data processing method, device, readable storage medium storing program for executing and program product | |
| CN112116245A (en) | Credit risk assessment method, credit risk assessment device, computer equipment and storage medium | |
| CN113034046A (en) | Data risk metering method and device, electronic equipment and storage medium | |
| CN111798304A (en) | Risk loan determination method and device, electronic equipment and storage medium | |
| CN111369345A (en) | Big data credit user credit assessment method and system of cloud computing block chain | |
| CN113807943A (en) | Multi-factor valuation method, system, medium and equipment for bad assets | |
| KR101927893B1 (en) | Method and system for the fundrevolving through the management of the credit rating. | |
| JP2003114977A (en) | Method and system for calculating customer's lifelong value | |
| CN116596659A (en) | Enterprise intelligent credit approval method, system and medium based on big data wind control | |
| CN114240605A (en) | Loan calculation method and device, computer equipment and storage medium | |
| CN114240599A (en) | Loan calculation method and device, computer equipment and storage medium | |
| JP2012238073A (en) | Credit purchase assessment support system and credit purchase assessment support method | |
| CN113269629A (en) | Credit limit determining method, electronic equipment and related product | |
| Li et al. | Semiparametric estimation of default probability: Evidence from the Prosper online credit market | |
| CN112116356B (en) | Asset characteristic information processing method and device | |
| KR102133668B1 (en) | Lending Meditation Platform System and Credit Estimating Apparatus | |
| KR102566466B1 (en) | Alternative Credit Rating System for Evaluating Personal Credit |
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 |