CN110634060A - User credit risk assessment method, system, device and storage medium - Google Patents
User credit risk assessment method, system, device and storage medium Download PDFInfo
- Publication number
- CN110634060A CN110634060A CN201810643043.XA CN201810643043A CN110634060A CN 110634060 A CN110634060 A CN 110634060A CN 201810643043 A CN201810643043 A CN 201810643043A CN 110634060 A CN110634060 A CN 110634060A
- Authority
- CN
- China
- Prior art keywords
- time characteristic
- characteristic information
- user
- implicit
- evaluation
- 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
- 238000000034 method Methods 0.000 title claims abstract description 66
- 238000012502 risk assessment Methods 0.000 title claims abstract description 31
- 238000007418 data mining Methods 0.000 claims abstract description 74
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 67
- 238000004364 calculation method Methods 0.000 claims abstract description 32
- 238000012911 target assessment Methods 0.000 claims abstract description 4
- 238000011156 evaluation Methods 0.000 claims description 151
- 238000012545 processing Methods 0.000 claims description 44
- 230000008859 change Effects 0.000 claims description 21
- 238000013527 convolutional neural network Methods 0.000 claims description 17
- 230000008569 process Effects 0.000 claims description 16
- 230000015654 memory Effects 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 10
- 230000007787 long-term memory Effects 0.000 claims description 10
- 238000011176 pooling Methods 0.000 claims description 10
- 230000006403 short-term memory Effects 0.000 claims description 10
- 238000012216 screening Methods 0.000 claims description 8
- 230000002123 temporal effect Effects 0.000 claims description 5
- 230000009286 beneficial effect Effects 0.000 abstract description 4
- 238000013528 artificial neural network Methods 0.000 description 19
- 239000011159 matrix material Substances 0.000 description 8
- 238000006243 chemical reaction Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 238000013507 mapping Methods 0.000 description 6
- 238000007781 pre-processing Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 4
- 238000005065 mining Methods 0.000 description 4
- 238000013058 risk prediction model Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 3
- 238000003066 decision tree Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 230000000306 recurrent effect Effects 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012805 post-processing Methods 0.000 description 2
- 238000011946 reduction process Methods 0.000 description 2
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2216/00—Indexing scheme relating to additional aspects of information retrieval not explicitly covered by G06F16/00 and subgroups
- G06F2216/03—Data mining
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)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a user credit risk assessment method, based on time characteristic information of a user to be tested, firstly, a data mining algorithm is utilized to obtain target hidden characteristics hidden behind the time characteristic information, then, target assessment weights are obtained according to a corresponding table recorded with corresponding relations between the hidden characteristics and the assessment weights, and finally, the product of the hidden characteristics and the assessment weights is used as credit risk assessment data of the user to be tested according to a weighting calculation method. The application also discloses a system and a device for evaluating the credit risk of the user and a computer readable storage medium, which have the beneficial effects.
Description
Technical Field
The present application relates to the field of risk assessment technologies, and in particular, to a method, a system, an apparatus, and a computer-readable storage medium for assessing a user credit risk.
Background
In the field of financial lending, financial institutions often use various techniques and means to assess the credit risk of a user by collecting various data of the user to assess the user's credit rating and deciding whether to ultimately credit the user and the amount of the loan based on the assessed user credit rating.
With the advancement of technology, the work of such user credit risk assessment is gradually shifted from traditional manual operation to automatic or semi-automatic by machine. In recent years, even a classification prediction model is developed through a machine learning technology, so that automatic evaluation of credit risks of users is realized.
With the continuous popularization and popularity of machine learning technology, a user credit risk assessment model is continuously evolved from an initial scoring card model to a decision tree, then to a random forest and finally to a combined decision tree, and the model is essentially a tree-structured classification algorithm although the scale of the model is larger and the complexity is higher. The prediction model is established essentially based on a condition judgment strategy, a condition judgment window needs to be established after manual analysis of characteristic information, and meanwhile, a limit condition based on quantifiable data needs to be met during condition judgment, so that analysis of user characteristic data stays in a shallow layer, and complex user characteristics which may affect the credit risk of a user along with time change cannot be observed. Therefore, the condition judgment and prediction model with the tree structure cannot adapt to the situations of larger and larger data scale, more types of user characteristic information and various expression forms, and cannot accurately evaluate the credit risk of the user under the situations.
Therefore, it is an urgent need for those skilled in the art to solve the above technical problems, how to overcome various technical defects of the existing user credit risk assessment mechanism, and provide a credit risk assessment mechanism of a credit risk prediction model that is established without manually setting a condition determination window and in combination with the influence that various types of user feature information may cause on final credit risk assessment over time.
Disclosure of Invention
The method comprises the steps of firstly obtaining target hidden characteristics hidden behind time characteristic information by using a data mining algorithm on the basis of the time characteristic information of a user to be detected, then obtaining target evaluation weights according to a corresponding table recorded with corresponding relations between the hidden characteristics and the evaluation weights, and finally taking the product of the hidden characteristics and the target evaluation weights as credit risk evaluation data of the user to be detected according to a weighting calculation method.
Another object of the present application is to provide a system, an apparatus and a computer-readable storage medium for evaluating a user credit risk.
In order to achieve the above object, the present application provides a method for evaluating a credit risk of a user, including:
acquiring time characteristic information of a user to be detected;
processing the time characteristic information by using a data mining algorithm to obtain a target hidden characteristic;
inquiring a preset corresponding table to obtain a target evaluation weight corresponding to the target implicit characteristic; wherein, the corresponding relation between the implicit characteristic and the evaluation weight is recorded in the corresponding table;
and calculating the target hidden features and the target evaluation weight according to a weighted calculation method to obtain credit risk evaluation data of the user to be tested.
Optionally, the generating process of the correspondence table includes:
acquiring historical time characteristic information; the historical time characteristic information is time characteristic information of a stock user;
processing the historical time characteristic information by using the data mining algorithm to obtain an implicit characteristic set;
respectively calculating the evaluation weight of each implicit feature in the implicit feature set by utilizing a full-connection layer to obtain an evaluation weight set;
and establishing the corresponding table according to the implicit characteristic set and the evaluation weight set.
Optionally, before the computing the evaluation weight of each implicit feature in the set of implicit features by using the fully-connected layer, the method further includes:
screening the implicit features contained in the implicit feature set by using a long-term and short-term memory network to obtain an optimal credit feature set;
respectively calculating the evaluation weight of each implicit feature in the implicit feature set by using a full-connection layer to obtain an evaluation weight set, wherein the evaluation weight set specifically comprises the following steps: respectively calculating the evaluation weight of each piece of preferred feature information in the preferred credit feature set by using the full connection layer to obtain a preferred evaluation weight set;
the establishing of the corresponding table according to the implicit characteristic set and the evaluation weight set specifically comprises the following steps: and establishing the corresponding table by using the preferred credit feature set and the preferred evaluation weight set.
Optionally, before processing the historical time characteristic information by using the data mining algorithm, the method further includes:
converting the historical time characteristic information into a historical time characteristic map according to a preset format;
the processing of the historical time characteristic information by using the data mining algorithm specifically comprises the following steps: and processing the historical time characteristic map by using the data mining algorithm.
Optionally, converting the historical time feature information into a historical time feature map according to a preset format, including:
merging the historical time characteristic information according to different time characteristic types to obtain a characteristic information set with the same number as the time characteristic types;
arranging the characteristic elements in each characteristic information set on a time axis according to the sequence of the generation time to obtain a parameter change table of the time characteristic of the corresponding type along with the time change;
and arranging the parameter change tables according to a preset arrangement mode to obtain the historical time characteristic map.
Optionally, before processing the time characteristic information by using a data mining algorithm, the method further includes:
converting the time characteristic information into a time characteristic map according to the preset format;
the processing of the time characteristic information by using a data mining algorithm specifically comprises the following steps: and processing the time characteristic map by using the data mining algorithm.
Optionally, the evaluation method further includes:
and adjusting the implicit feature set, the evaluation weight set and the corresponding table by using credit risk actual data of the inventory user.
Optionally, processing the time characteristic information by using a data mining algorithm includes:
and processing the time characteristic information by sequentially utilizing a convolution layer and a pooling layer in the convolutional neural network.
In order to achieve the above object, the present application further provides a system for evaluating a credit risk of a user, including:
the time characteristic information acquisition unit is used for acquiring the time characteristic information of the user to be detected;
the first data mining unit is used for processing the time characteristic information by using a data mining algorithm to obtain a target hidden characteristic;
the target evaluation weight query unit is used for obtaining a target evaluation weight corresponding to the target hidden feature by querying from a preset corresponding table; wherein, the corresponding relation between the implicit characteristics and the evaluation weight is recorded in the corresponding table;
and the credit risk assessment data calculation unit is used for calculating the target hidden features and the target assessment weights according to a weighting calculation method to obtain the credit risk assessment data of the user to be tested.
Optionally, the evaluation system further comprises:
a historical time characteristic information obtaining unit for obtaining historical time characteristic information; the historical time characteristic information is time characteristic information of a stock user;
the second data mining unit is used for processing the historical time characteristic information by using the data mining algorithm to obtain an implicit characteristic set;
the evaluation weight calculation unit is used for calculating the evaluation weight of each implicit feature in the implicit feature set by utilizing a full connection layer to obtain an evaluation weight set;
and the corresponding table establishing unit is used for establishing the corresponding table according to the implicit characteristic set and the evaluation weight set.
Optionally, the evaluation system further comprises:
the preferred credit feature screening unit is used for screening the hidden features contained in the hidden feature set by using a long-term and short-term memory network to obtain a preferred credit feature set;
the evaluation weight calculation unit specifically includes: respectively calculating the evaluation weight of each piece of preferred feature information in the preferred credit feature set by using the full connection layer to obtain a preferred evaluation weight set;
the correspondence table establishing unit specifically comprises: and establishing the corresponding table by using the preferred credit feature set and the preferred evaluation weight set.
Optionally, the evaluation system further comprises:
the first format conversion unit is used for converting the historical time characteristic information into a historical time characteristic map according to a preset format;
the second data mining unit specifically comprises: and processing the historical time characteristic map by using the data mining algorithm.
Optionally, the first format conversion unit includes:
the merging processing subunit is used for merging the historical time characteristic information according to different time characteristic types to obtain a characteristic information set with the same number as the time characteristic types;
the arrangement subunit is used for arranging the characteristic elements in each characteristic information set on a time axis according to the sequence of the generation time to obtain a parameter change table of the time characteristics of the corresponding types along with the time change;
and the historical time characteristic map generating subunit is used for arranging the parameter change tables according to a preset arrangement mode to obtain the historical time characteristic map.
Optionally, the evaluation system further comprises:
the second format conversion unit is used for converting the time characteristic information into a time characteristic map according to the preset format;
the first data mining unit specifically comprises: and processing the time characteristic map by using the data mining algorithm.
Optionally, the evaluation system further comprises:
and the parameter adjusting unit is used for adjusting the implicit feature set, the evaluation weight set and the corresponding table by using credit risk actual data of the inventory user.
Optionally, the first data mining unit includes:
and the convolutional neural network processing subunit is used for processing the time characteristic information by sequentially utilizing a convolutional layer and a pooling layer in the convolutional neural network.
In order to achieve the above object, the present application further provides an apparatus for evaluating a credit risk of a user, including:
a memory for storing a computer program;
a processor for implementing the steps of the method for assessing a user's credit risk as described above when executing said computer program.
To achieve the above object, the present application also provides a computer-readable storage medium having stored thereon a computer program, which when executed by a processor, implements the steps of the method for assessing a user's credit risk as described above.
Obviously, the evaluation method for the user credit risk provided by the application is based on the time characteristic information of the user to be tested, firstly, the data mining algorithm is utilized to obtain the target hidden characteristics hidden behind the time characteristic information, then, the target evaluation weight is obtained according to the corresponding table recorded with the corresponding relation between each hidden characteristic and the evaluation weight, and finally, the product of the hidden characteristics and the evaluation weight is used as the credit risk evaluation data of the user to be tested according to the weighting calculation method. The application also provides a system and a device for evaluating the credit risk of the user and a computer readable storage medium, which have the beneficial effects and are not repeated herein.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a flowchart of a method for evaluating a user credit risk according to an embodiment of the present application;
fig. 2 is a flowchart of a method for generating a corresponding table in the method for evaluating a user credit risk according to the embodiment of the present application;
fig. 3 is a flowchart of another method for generating a corresponding table in the method for evaluating a credit risk of a user according to the embodiment of the present application;
FIG. 4 is a flowchart of another method for generating a mapping table in the method for evaluating a credit risk of a user according to the embodiment of the present application
FIG. 5 is a flowchart of another method for assessing a user's credit risk according to an embodiment of the present disclosure;
fig. 6 is a logic diagram illustrating a time feature map building process and a dimension reduction process using a convolutional neural network in the method for evaluating a user credit risk according to the embodiment of the present application;
fig. 7 is a logic diagram of a one-dimensional time feature matrix sequentially processed by a long-term and short-term memory network and a full connection layer in the method for evaluating a user credit risk according to the embodiment of the present application;
fig. 8 is a block diagram of a structure of a system for evaluating a user credit risk according to an embodiment of the present application.
Detailed Description
The core of the application is to provide an evaluation method of user credit risk, based on time characteristic information of a user to be tested, firstly, a data mining algorithm is utilized to obtain a target hidden characteristic hidden behind the time characteristic information, then, a target evaluation weight is obtained according to a corresponding table recorded with corresponding relations between the hidden characteristics and the evaluation weights, and finally, the product of the hidden characteristic and the target evaluation weight is used as credit risk evaluation data of the user to be tested according to a weighting calculation method. The application also provides a system and a device for evaluating the credit risk of the user and a computer readable storage medium, and the system and the device have the beneficial effects.
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
With reference to fig. 1, fig. 1 is a flowchart of a method for evaluating a user credit risk according to an embodiment of the present application, which specifically includes the following steps:
s101: acquiring time characteristic information of a user to be detected;
the step aims to obtain the time characteristic information of the user to be tested, and because the step possibly comprises not only the time-related characteristic information but also other non-time-related characteristic information within the range of the type of the characteristic information of the user allowed to be collected, the purpose of the step is to evaluate the credit risk of the user and judge whether to approve the loan service application of the user and the safe loan amount upper limit according to the evaluation result, the related condition is the fund condition and the credit condition of the user, and in combination with long-term real cases, the characteristic information influencing the final credit risk evaluation result is often related to the time characteristic information of the user.
By taking the situation of fund flow change (expenditure and entry) of the user along with time as an example, the consumption habit and the repayment capacity of the corresponding user can be approximately seen, so that the application starts from the time characteristic information of the user to find out the characteristic information capable of influencing the credit risk assessment result of the user from the time characteristic information.
It should be noted that the user to be tested in this step refers to all users who need to perform credit risk assessment on the user, including users who register an application but do not apply for or are applying for a loan service (there is no payment for the previous loan service), and also includes users who have previously applied for a loan service (there is payment for the previous loan service), and since these users may have to apply for a loan service again, it is necessary to prepare for applying for a loan service next time by combining the previous time characteristic information.
S102: processing the time characteristic information by using a data mining algorithm to obtain a target hidden characteristic;
on the basis of S101, this step aims to process the acquired time characteristic information by using a data mining algorithm to obtain a target implicit characteristic hidden in the time characteristic information.
The time characteristic information obtained from S101 usually includes different types of time characteristics, for example, time characteristic information that may affect the credit risk assessment result of the user when viewed from the surface, such as a fund flow change situation, a credit investigation change situation, and an opening number of a certain shopping application of the user, and also includes time characteristic information that may affect the assessment result when viewed from the surface, such as a date change situation, a countdown of a memorial day, and the like, but the time characteristic information that may not affect the credit risk assessment result specifically when viewed from the surface may have a myriad of connections with the assessment result, which is also a disadvantage of manually setting up a classification window in the prior art.
Therefore, it is necessary to mine implicit features hidden behind the time feature information to minimize missing of feature types and specific feature information to which the feature types affect the evaluation result, because if the evaluation result is inaccurate, the repayment capability of a user may be evaluated incorrectly, and the probability of bad accounts and dead accounts is increased. The step of mining uses a data mining algorithm to mine the target hidden characteristics behind the time characteristic information so as to evaluate the credit risk of the tested user based on the target hidden characteristics.
The data mining algorithm comprises a plurality of specific algorithms, and as the similarity meter algorithm and decision tree method based on comparison have been proved to have defects by the prior art, the data mining algorithm described in the application is specifically an algorithm based on Machine learning, and can be roughly divided into supervised and unsupervised algorithms, wherein the supervised algorithm is commonly used in supervision and comprises regression analysis (SVM, support vector Machine) and statistical classification (KNN, k-nearest neighbor, neighborhood analysis algorithm), and the unsupervised algorithm mainly comprises various clustering algorithms.
The neural network can be further subdivided, and comprises a plurality of layers of neural networks, convolutional neural networks and recurrent neural networks which are added into convolutional layers and pooling layers, and the like, different types of neural networks are more suitable for being applied to some special application scenes according to different characteristics of the neural networks, the neural networks are not particularly limited, the most suitable data mining algorithm can be flexibly selected according to actual conditions, hidden target hidden characteristics can be preferably mined from the time characteristic information as much as possible, and the omission of characteristic information which can affect the credit risk assessment result is reduced as much as possible.
S103: inquiring a preset corresponding table to obtain a target evaluation weight corresponding to the target implicit characteristic; wherein, the corresponding relation between the implicit characteristic and the evaluation weight is recorded in the corresponding table;
on the basis of S102, the step aims to obtain the target evaluation weight corresponding to the mined target implicit characteristics by combining with the query of a preset corresponding table. According to actual conditions, the types of the target hidden features mined from the time feature information of the tested user may exist in multiple numbers, but the influence degrees of different types of target hidden features on the final credit risk assessment result obtained through calculation may be different, and the assessment weight is an abstract parameter for measuring different influence degrees, so that different influences of various target hidden features on the credit risk assessment result are simulated as truly as possible.
It should be noted that, the correspondence table records the correspondence between each hidden feature and each evaluation weight, so that after the target hidden feature is mined from the time feature information of the user to be tested, the target evaluation weight corresponding to the target hidden feature can be found according to the correspondence table. It should be further noted that, the correspondence table can record the correspondence between each implicit feature and each evaluation weight, and is obtained by performing the same data mining operation on historical time features obtained by summarizing training samples with a longer time period, more inventory users and a larger data magnitude, and because the training samples have a larger magnitude and cover the most comprehensive time feature types, the types of the implicit features obtained based on the correspondence are more, and the calculation of the corresponding evaluation weights is obtained based on the calculation of the full connection layer of the neural network, and the special framework based on the full connection layer may automatically score each input implicit feature so as to use the scoring result as the evaluation weight of the corresponding implicit feature.
Further, in the process of obtaining the correspondence table based on the historical time feature information, some implicit features which appear less frequently in a longer time period and have longer intervals but have larger influence on the final credit risk assessment data are often judged incorrectly, taking a common neural network as an example, because the existence of recursion factors on a time line is not considered in the design, incorrect judgment is made on the features, and even the important features are ignored, so that the final credit risk assessment result is not accurate, therefore, in order to solve the problem, in the process of establishing the correspondence table, the implicit features can be screened by using the recursion neural network to judge the important features correctly.
One way includes, but is not limited to, using the more classical long-short term memory network in the recurrent neural network to filter implicit features.
Furthermore, if the disordered time characteristic information is directly used as input data to be processed by a data mining algorithm, the difficulty of data mining may be increased due to the disorder, so that the workload of data mining and the difficulty of data mining are reduced, the time characteristic information can be converted in advance according to a preset format, the conversion aims at preprocessing the time characteristic information, the time characteristic information which is seemingly one type of time characteristic information can be merged to classify the time characteristic information, and the difficulty of data mining is reduced on the basis of preprocessing. It should be noted that, in the process of generating the mapping table based on the historical time characteristic information, the same preprocessing method needs to be executed before the time characteristic information of the user to be tested is subjected to data mining, so as to ensure consistency of the pre-processing and the post-processing.
After the corresponding table is preliminarily established, credit risk evaluation data can be adjusted according to credit risk real data in the inventory users, namely, the data in the corresponding table is more accurate by adjusting parameters of a data mining algorithm and parameters of full-connection layer calculation evaluation weights.
S104: and calculating the target hidden features and the target evaluation weight according to a weighting calculation method to obtain credit risk evaluation data of the user to be tested.
On the basis of S103, the step aims to calculate the target hidden features and the target evaluation weights according to a weighting calculation method to obtain credit risk evaluation data of the user to be tested. For example, when the target implicit features exist in A, B, C, D types and E five types, and the weights of the target implicit features in the correspondence table are 0.1, 0.5, 0.2, and 0.1 respectively after the calculation of the full connection layer, the calculation process according to the weighting calculation method is as follows: a × 0.1+ B × 0.1+ C × 0.5+ D × 0.2+ E × 0.1 is credit risk assessment data, and it can also be seen that C, a time characteristic, has a large influence on the final assessment result.
It should be noted that the sum of the weights of the 5 target implicit features in the above example is exactly 1, and there are also cases where the sum of the weights obtained after the lookup according to the correspondence table is not 1, for example, when there are only A, B, D target implicit features, i.e. the sum weight is only 0.4, the original evaluation weights can be expanded proportionally, i.e. after keeping a: b: and under the condition that the weight ratio of the D is not changed, multiplying by 0.25 to respectively obtain new evaluation weights of 0.25, 0.25 and 0.5, and adding the new evaluation weights to 1 to ensure that the credit risk evaluation data under different conditions are transversely compared with the same evaluation standard.
Furthermore, on the basis of using a weighting calculation method, a correction coefficient and an auxiliary adjustment parameter can be added by combining all possible special conditions under the actual condition, so that the finally obtained credit risk assessment data is more accurate.
Based on the technical scheme, the method for evaluating the credit risk of the user provided by the embodiment of the application is based on the time characteristic information of the user to be tested, firstly, the data mining algorithm is utilized to obtain the target hidden characteristics hidden behind the time characteristic information, then, the target evaluation weight is obtained according to the corresponding table recorded with the corresponding relation between each hidden characteristic and the evaluation weight, and finally, the product of the hidden characteristics and the evaluation weight is used as the credit risk evaluation data of the user to be tested according to the weighting calculation method.
Example two
With reference to fig. 2, fig. 2 is a flowchart of a method for generating a mapping table in an evaluation method for a user credit risk provided in the embodiment of the present application, where the embodiment provides a specific method for generating a mapping table on the basis of the first embodiment, and the specific method specifically includes the following steps:
s201: acquiring historical time characteristic information; the historical time characteristic information is the time characteristic information of the stock user;
s202: processing the historical time characteristic information by using a data mining algorithm to obtain an implicit characteristic set;
in order to ensure consistency in the pre-and post-processing processes, the data mining algorithm used in this step is consistent with the algorithm used in the data mining of the time characteristic information of the user to be tested in the embodiment. Taking the example of using the neural network algorithm to perform data mining on the historical time characteristic information and finally generating a corresponding table, when performing data mining on the time characteristic information of the user to be tested, the same neural network algorithm is also required to be used to obtain the target implicit characteristic.
S203: respectively calculating the evaluation weight of each hidden feature in the hidden feature set by utilizing the full-connection layer to obtain an evaluation weight set;
on the basis of S202, in this step, the implicit feature set is used as input data of the fully-connected layer, so as to obtain an evaluation weight for each implicit feature in the implicit feature set by using a special structure calculation of the fully-connected layer, thereby obtaining an evaluation weight set.
S204: and establishing a corresponding table according to the implicit characteristic set and the evaluation weight set.
On the basis of S202 and S203, this step aims to establish a correspondence table according to the implicit feature set and the evaluation weight set, that is, the correspondence table is established according to the correspondence between each implicit feature and the corresponding evaluation weight, specifically, the correspondence in the correspondence table should be one-to-one, that is, one implicit feature corresponds to a unique evaluation weight, and of course, the numerical values of the evaluation weights respectively corresponding to different implicit features may also be the same.
EXAMPLE III
With reference to fig. 3, fig. 3 is a flowchart of another method for generating a corresponding table in the method for evaluating a user credit risk provided in the embodiment of the present application, where the embodiment adds a scheme for screening implicit features using a recursive network on the basis of the second embodiment, and the method specifically includes the following steps:
s301: acquiring historical time characteristic information; the historical time characteristic information is the time characteristic information of the stock user;
s302: processing the historical time characteristic information by using a data mining algorithm to obtain an implicit characteristic set;
s303: screening the hidden features contained in the hidden feature set by using a long-term and short-term memory network to obtain an optimal credit feature set;
in the embodiment, the long-term and short-term memory network is adopted to screen the implicit features contained in the implicit feature set to obtain the preferred credit feature set.
S304: respectively calculating the evaluation weight of each preferred feature information in the preferred credit feature set by using the full connection layer to obtain a preferred evaluation weight set;
s305: and establishing a corresponding table according to the preferred credit feature set and the preferred evaluation weight set.
S304, on the basis of S303, only the evaluation weight of the preferred credit feature in the preferred credit feature set is calculated by utilizing the full connection layer, so that a preferred evaluation weight set is obtained; s305 only needs to establish a preferred mapping table by using the preferred credit feature set and the preferred evaluation weight set based on S303 and S304. The implementation can make the final result more accurate because a recurrent neural network of the long-short term memory network is added to screen the implicit characteristics.
Example four
With reference to fig. 4, fig. 4 is a flowchart of another method for generating a corresponding table in the method for evaluating a user credit risk provided in the embodiment of the present application, where the embodiment adds a scheme for preprocessing a historical time feature on the basis of the second embodiment, and specifically includes the following steps
S401: acquiring historical time characteristic information; the historical time characteristic information is the time characteristic information of the stock user;
s402: converting the historical time characteristic information into a historical time characteristic map according to a preset format;
in the embodiment, the historical time characteristic information is converted into the historical time characteristic map according to a preset format, the original disordered historical time characteristic information is sorted according to different time characteristic types by the converted historical time characteristic map, and the change condition of various types of time characteristics on a time sequence is formed in the form of the map, so that the difficulty of a subsequent data mining algorithm in data mining is reduced.
One method, including but not limited to, converting historical temporal feature information into a historical temporal feature map is:
merging the historical time characteristic information according to different time characteristic types to obtain a characteristic information set with the same number as the time characteristic types; arranging the characteristic elements in each characteristic information set on a time axis according to the sequence of the generation time to obtain a parameter change table of time characteristic of the corresponding type along with time change; and arranging the parameter change tables according to a preset arrangement mode to obtain a historical time characteristic map.
S403: processing the historical time feature map by using a data mining algorithm to obtain an implicit feature set;
s404: respectively calculating the evaluation weight of each hidden feature in the hidden feature set by utilizing the full-connection layer to obtain an evaluation weight set;
s405: and establishing a corresponding table according to the implicit characteristic set and the evaluation weight set.
The third embodiment and the fourth embodiment are respectively added with different schemes on the basis of the second embodiment to improve some small defects existing in the process of forming the corresponding table so as to increase the accuracy of data parameters in the finally formed corresponding table, so that the credit risk assessment data of the tested user is more accurate and closer to the real credit condition of the user. And the third embodiment and the fourth embodiment also respectively provide a complete scheme for generating the corresponding table, so that the third embodiment and the fourth embodiment can respectively form different preferred embodiments independently from the first embodiment, and a new embodiment can be formed by combining the improved schemes provided from two aspects of the third embodiment and the fourth embodiment on the principle of obtaining a more accurate credit risk assessment result.
EXAMPLE five
With reference to fig. 5, fig. 5 is a flowchart of another method for evaluating a user credit risk according to an embodiment of the present application, where a scheme provided by the present application corresponds to a scheme provided by the fourth embodiment for obtaining a historical time feature map by converting historical time feature information according to a preset format, that is, on the basis of forming a mapping table scheme by using the scheme of the fourth embodiment, credit risk evaluation data of a user to be tested is obtained by using the method of the present embodiment, so as to ensure accuracy of the evaluation data on the basis of ensuring consistency between before and after the user is evaluated:
s501: acquiring time characteristic information of a user to be detected;
s502: converting the time characteristic information into a time characteristic map according to a preset format;
according to the implementation, on the basis that the time characteristic information of the detected user is obtained in S501, the time characteristic information is converted into the time characteristic map according to the preset format adopted when the historical time characteristic information is converted into the historical time characteristic map, so that the time characteristic map is processed by using a data mining algorithm in S503 to reduce the difficulty of data mining.
S503: processing the time characteristic map by using a data mining algorithm to obtain target hidden characteristics;
s504: inquiring a preset corresponding table to obtain a target evaluation weight corresponding to the target implicit characteristic; wherein, the corresponding relation between the implicit characteristic and the evaluation weight is recorded in the corresponding table;
s505: and calculating the target hidden features and the target evaluation weight according to a weighting calculation method to obtain credit risk evaluation data of the user to be tested.
For the parts that are not explained in the above embodiments, the related descriptions appear in other embodiments, and the other embodiments do not adjust the parts, so the same descriptions can be referred to, and the details are not repeated herein.
Based on any one of the first to fifth embodiments, the present application further provides a step of performing data mining when a preferred data mining algorithm is adopted, that is, performing data mining by using a convolutional neural network. The Network depends on the complexity of the system, and achieves the purpose of processing information by adjusting the interconnection relationship among a large number of nodes inside, thereby adding a Convolutional layer and a pooling layer.
Because the convolutional neural network comprises a convolutional layer and a pooling layer, the convolutional layer is processed before and then processed by the pooling layer, and the main purposes of the two layers of processing are to reduce parameters (dimension reduction) and calculation amount while keeping main characteristics, prevent overfitting and improve generalization capability, because original time characteristic information may be complex information and has higher dimension, namely the information is established on the basis of a plurality of other information, the two layers of processing are to reduce the dimension of the information so as to find more implicit characteristics from low-dimensional characteristics.
In order to further enable the implicit time characteristics obtained by data mining to be more accurate, on the basis that one characteristic mining layer is formed by one convolution layer and one pooling layer, multiple characteristic mining can be carried out by arranging multiple characteristic mining layers, and the finally obtained implicit time characteristics are more accurate.
Because the situation is complicated and cannot be illustrated by a list, a person skilled in the art can realize that many examples exist according to the basic method principle provided by the application and the practical situation, and the protection scope of the application should be protected without enough inventive work.
EXAMPLE six
Referring to fig. 6 and 7, fig. 6 is a logic diagram illustrating a time feature map building logic to a dimension reduction process using a convolutional neural network in the method for assessing user credit risk according to the embodiment of the present application; fig. 7 is a logic diagram of a one-dimensional time feature matrix processed by a long-term and short-term memory network and a full connection layer in sequence in the method for evaluating a user credit risk provided in the embodiment of the present application.
The scheme of the embodiment uses a convolutional neural network and a long-short term memory network which are generally applied in the deep learning field for reference, and a credit risk prediction model based on the neural network is combined, wherein the main responsibility of the convolutional neural network is to learn (mine and mashup) the features which influence the credit risk assessment of the user, and the main responsibility of the long-short term memory network is to screen based on the results after feature learning and combined with a time sequence.
Firstly, historical characteristic information of inventory users is required to be used for training to construct a prediction model:
1. all user characteristics related to time are extracted from the historical characteristic information, the number of the time characteristic types of the users is assumed to be H, and the time characteristics are stored as a time sequence. Assuming that the length of the time series is W observation periods (default is day), so that the characteristic data of each user can form a W x H time characteristic map;
2. then, the formed time characteristic map is used as input data of a convolutional neural network, the convolutional neural network automatically extracts the implicit characteristics of the user on the time sequence from the time characteristic map by utilizing a convolutional layer and a pooling layer according to the input W x H time characteristic matrix, and an N x 1 output matrix containing the implicit time characteristics is obtained;
3. and converting the N x 1 output matrix into a list (T x W matrix, and the product of T and W is N) which is used as input data of the long-term and short-term memory network, after the long-term and short-term memory network comprehensively compares T implicit time characteristics, automatically scoring the processed data through N full-connection layers (fc) according to characteristic weights, and finally returning a prediction result. And repeatedly comparing the returned prediction result with the actual performance of the inventory user, continuously adjusting the parameters of the convolutional neural network and the long-term and short-term memory network, and finally obtaining a prediction model with strong generalization capability, namely extracting a corresponding table containing implicit characteristics and evaluation weights from the finally formed prediction model.
After the prediction model is built, when a new user applies for a service, the time characteristics of the new user can be generated into an H-W time characteristic matrix in the same mode, the H-W time characteristic matrix is input into the trained prediction model (namely, the target hidden characteristics are obtained through the same data mining algorithm, the corresponding table is used for inquiring to obtain the evaluation weight corresponding to the target hidden characteristics), the credit risk evaluation data of the new user can be calculated by using a weighting calculation method based on the target hidden characteristics and the corresponding evaluation weight, the credit risk grade of the new user is determined according to the credit risk evaluation data, and whether overdue risk and a specific credit amount upper limit exist is judged according to the credit risk evaluation data.
Generally, collected user historical characteristic data are behavior habit information of the user, the relationship between the behavior habit information and financial loan business is very weak, and the generalization capability of a traditional tree structure classification prediction model is poor in user credit risk assessment. Because the neural network is greatly successful in image recognition, the idea of using the neural network to mine hidden characteristics hidden behind surface user behavior habit information is generated, in order to ensure the effect of the neural network and meet the requirement of actual situations, a combined mode of a convolutional neural network and a long-short term memory network is used for constructing a final credit risk prediction model, evaluation indexes are errors between prediction results and actual conditions, and the smaller the error is, the better the credit risk prediction model is.
Compared with the prior art, the scheme provided by the embodiment has the following beneficial effects:
1. compared with a traditional tree structure classification prediction model, the neural network can sense nonlinear feature data in data and can find more complex user features;
2. the input data comprises time dimension, so that the problem that the tree-structure classification prediction model is difficult to process time sequence can be effectively solved;
3. the convolution kernel of the convolution neural network can be continuously adjusted, and the optimal time window can be captured effectively with the increase of the network scale, so that the efficiency is higher and more accurate than the efficiency of artificially setting the time window;
4. the long-short term memory network is more suitable for processing and predicting important events with very long intervals and delays in the time sequence, so that the characteristics of low occurrence frequency, long time intervals and high importance degree are taken into consideration, and the accuracy of the final result is higher.
It should be noted that, in the steps of describing the execution of each algorithm mentioned in the present application, how to perform specific processing inside the algorithm (for example, how to process data by convolutional layer) is not explained in detail, because the specific use of the algorithm as a well-known algorithm has been disclosed, the present application focuses on how to use the algorithms originally used in other fields and solving other problems to solve the problem of how to accurately evaluate the credit risk of the user, and the application field and the use idea of the algorithms are greatly expanded.
Referring to fig. 8, fig. 8 is a block diagram illustrating a structure of a system for evaluating a user credit risk according to an embodiment of the present application, where the system for evaluating a user credit risk includes:
a time characteristic information obtaining unit 100, configured to obtain time characteristic information of a user to be detected;
the first data mining unit 200 is configured to process the time feature information by using a data mining algorithm to obtain a target hidden feature;
a target evaluation weight query unit 300, configured to query a preset correspondence table to obtain a target evaluation weight corresponding to the target implicit feature; wherein, the corresponding relation between the implicit characteristics and the evaluation weight is recorded in the corresponding table;
and the credit risk assessment data calculation unit 400 is configured to perform calculation between the target hidden feature and the target assessment weight according to a weighting calculation method to obtain credit risk assessment data of the user to be tested.
Further, the evaluation system may further include:
a historical time characteristic information obtaining unit for obtaining historical time characteristic information; the historical time characteristic information is the time characteristic information of the stock user;
the second data mining unit is used for processing the historical time characteristic information by using a data mining algorithm to obtain an implicit characteristic set;
the evaluation weight calculation unit is used for calculating the evaluation weight of each implicit feature in the implicit feature set by utilizing the full-connection layer to obtain an evaluation weight set;
and the corresponding table establishing unit is used for establishing a corresponding table according to the implicit characteristic set and the evaluation weight set.
Further, the evaluation system may further include:
the preferred credit feature screening unit is used for screening the hidden features contained in the hidden feature set by using the long-short term memory network to obtain a preferred credit feature set;
the evaluation weight calculation unit specifically includes: respectively calculating the evaluation weight of each preferred feature information in the preferred credit feature set by using the full connection layer to obtain a preferred evaluation weight set;
the corresponding table establishing unit is specifically as follows: and establishing a corresponding table by utilizing the preferred credit feature set and the preferred evaluation weight set.
Further, the evaluation system may further include:
the first format conversion unit is used for converting the historical time characteristic information into a historical time characteristic map according to a preset format;
the second data mining unit specifically comprises: and processing the historical time characteristic map by using a data mining algorithm.
Wherein, the first format conversion unit may include:
the merging processing subunit is used for merging the historical time characteristic information according to different time characteristic types to obtain a characteristic information set with the same number as the time characteristic types;
the arrangement subunit is used for arranging the characteristic elements in each characteristic information set on a time axis according to the sequence of the generation time to obtain a parameter change table of the time characteristics of the corresponding types along with the time change;
and the historical time characteristic map generating subunit is used for arranging the parameter change tables according to a preset arrangement mode to obtain the historical time characteristic map.
Further, the evaluation system may further include:
the second format conversion unit is used for converting the time characteristic information into a time characteristic map according to a preset format;
the first data mining unit specifically comprises: and processing the time characteristic map by using a data mining algorithm.
Further, the evaluation system may further include:
and the parameter adjusting unit is used for adjusting the implicit characteristic set, the evaluation weight set and the corresponding table by using the credit risk actual data of the inventory user.
Wherein the first data mining unit may include:
and the convolutional neural network processing subunit is used for processing the time characteristic information by sequentially utilizing a convolutional layer and a pooling layer in the convolutional neural network.
Based on the foregoing embodiments, the present application further provides an apparatus for evaluating a user credit risk, where the apparatus may include a memory and a processor, where the memory stores a computer program, and the processor may implement the steps provided in the foregoing embodiments when calling the computer program in the memory. Of course, the evaluation device may also include various necessary network interfaces, power supplies, other components, and the like.
The present application also provides a computer-readable storage medium, on which a computer program is stored, which, when executed by an execution terminal or processor, can implement the steps provided by the above-mentioned embodiments. The storage medium may include: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The embodiments are described in a progressive manner in the specification, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. The device disclosed by the embodiment corresponds to the method disclosed by the embodiment, so that the description is simple, and the relevant points can be referred to the method part for description.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
The principles and embodiments of the present application are explained herein using specific examples, which are provided only to help understand the method and the core idea of the present application. It will be apparent to those skilled in the art that various changes and modifications can be made in the present invention without departing from the principles of the invention, and these changes and modifications also fall within the scope of the claims of the present application.
It is further noted that, in the present specification, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
Claims (11)
1. A method for assessing a credit risk of a user, comprising:
acquiring time characteristic information of a user to be detected;
processing the time characteristic information by using a data mining algorithm to obtain a target hidden characteristic;
inquiring a preset corresponding table to obtain a target evaluation weight corresponding to the target implicit characteristic; wherein, the corresponding relation between the implicit characteristic and the evaluation weight is recorded in the corresponding table;
and calculating the target hidden features and the target evaluation weight according to a weighted calculation method to obtain credit risk evaluation data of the user to be tested.
2. The evaluation method according to claim 1, wherein the generation process of the correspondence table includes:
acquiring historical time characteristic information; the historical time characteristic information is time characteristic information of a stock user;
processing the historical time characteristic information by using the data mining algorithm to obtain an implicit characteristic set;
respectively calculating the evaluation weight of each implicit feature in the implicit feature set by utilizing a full-connection layer to obtain an evaluation weight set;
and establishing the corresponding table according to the implicit characteristic set and the evaluation weight set.
3. The method of claim 2, further comprising, before calculating the evaluation weight of each implicit feature in the set of implicit features separately using a fully-connected layer:
screening the implicit features contained in the implicit feature set by using a long-term and short-term memory network to obtain an optimal credit feature set;
respectively calculating the evaluation weight of each implicit feature in the implicit feature set by using a full-connection layer to obtain an evaluation weight set, wherein the evaluation weight set specifically comprises the following steps: respectively calculating the evaluation weight of each piece of preferred feature information in the preferred credit feature set by using the full connection layer to obtain a preferred evaluation weight set;
the establishing of the corresponding table according to the implicit characteristic set and the evaluation weight set specifically comprises the following steps: and establishing the corresponding table by using the preferred credit feature set and the preferred evaluation weight set.
4. The evaluation method according to claim 2 or 3, prior to processing the historical temporal feature information using the data mining algorithm, further comprising:
converting the historical time characteristic information into a historical time characteristic map according to a preset format;
the processing of the historical time characteristic information by using the data mining algorithm specifically comprises the following steps:
and processing the historical time characteristic map by using the data mining algorithm.
5. The evaluation method according to claim 4, wherein converting the historical time characteristic information into a historical time characteristic map according to a preset format comprises:
merging the historical time characteristic information according to different time characteristic types to obtain a characteristic information set with the same number as the time characteristic types;
arranging the characteristic elements in each characteristic information set on a time axis according to the sequence of the generation time to obtain a parameter change table of the time characteristic of the corresponding type along with the time change;
and arranging the parameter change tables according to a preset arrangement mode to obtain the historical time characteristic map.
6. The evaluation method of claim 5, prior to processing the temporal feature information using a data mining algorithm, further comprising:
converting the time characteristic information into a time characteristic map according to the preset format;
the processing of the time characteristic information by using a data mining algorithm specifically comprises the following steps: and processing the time characteristic map by using the data mining algorithm.
7. The evaluation method according to claim 2, further comprising:
and adjusting the implicit feature set, the evaluation weight set and the corresponding table by using credit risk actual data of the inventory user.
8. The evaluation method of claim 1, wherein processing the temporal feature information using a data mining algorithm comprises:
and processing the time characteristic information by utilizing a convolutional layer and a pooling layer in the convolutional neural network.
9. A system for assessing a user's credit risk, comprising:
the time characteristic information acquisition unit is used for acquiring the time characteristic information of the user to be detected;
the first data mining unit is used for processing the time characteristic information by using a data mining algorithm to obtain a target hidden characteristic;
the target evaluation weight query unit is used for obtaining a target evaluation weight corresponding to the target hidden feature by querying from a preset corresponding table; wherein, the corresponding relation between the implicit characteristics and the evaluation weight is recorded in the corresponding table;
and the credit risk assessment data calculation unit is used for calculating the target hidden features and the target assessment weights according to a weighting calculation method to obtain the credit risk assessment data of the user to be tested.
10. An apparatus for assessing a credit risk of a user, comprising:
a memory for storing a computer program;
a processor for implementing the steps of the method for assessing a user's credit risk according to any one of claims 1 to 8 when executing said computer program.
11. A computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, which computer program, when being executed by a processor, carries out the steps of the method for assessing a user's credit risk according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810643043.XA CN110634060A (en) | 2018-06-21 | 2018-06-21 | User credit risk assessment method, system, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810643043.XA CN110634060A (en) | 2018-06-21 | 2018-06-21 | User credit risk assessment method, system, device and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110634060A true CN110634060A (en) | 2019-12-31 |
Family
ID=68967636
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810643043.XA Pending CN110634060A (en) | 2018-06-21 | 2018-06-21 | User credit risk assessment method, system, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110634060A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111369139A (en) * | 2020-03-03 | 2020-07-03 | 电子科技大学 | Individual credit risk assessment method, system, terminal and storage medium |
| CN111507461A (en) * | 2020-04-13 | 2020-08-07 | 支付宝(杭州)信息技术有限公司 | Interpretability information determining method and device |
| CN111861701A (en) * | 2020-07-09 | 2020-10-30 | 深圳市富之富信息技术有限公司 | Wind control model optimization method and device, computer equipment and storage medium |
| WO2022062193A1 (en) * | 2020-09-28 | 2022-03-31 | 南京博雅区块链研究院有限公司 | Individual credit assessment and explanation method and apparatus based on time sequence attribution analysis, and device and storage medium |
| CN115713987A (en) * | 2022-11-17 | 2023-02-24 | 广州瑞博新材料技术研究有限公司 | Polycaprolactone test data analysis method and system |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140143134A1 (en) * | 2010-01-13 | 2014-05-22 | Corelogic Solutions, Llc | System and method of detecting and assessing multiple types of risks related to mortgage lending |
| CN106127363A (en) * | 2016-06-12 | 2016-11-16 | 腾讯科技（深圳）有限公司 | A kind of user credit appraisal procedure and device |
| CN106202491A (en) * | 2016-07-19 | 2016-12-07 | 中国银联股份有限公司 | A kind of based on seasonal effect in time series data digging method and system |
| CN106408184A (en) * | 2016-09-12 | 2017-02-15 | 中山大学 | User credit evaluation model based on multi-source heterogeneous data |
| CN107301577A (en) * | 2016-04-15 | 2017-10-27 | 阿里巴巴集团控股有限公司 | Training method, credit estimation method and the device of credit evaluation model |
| CN107316198A (en) * | 2016-04-26 | 2017-11-03 | 阿里巴巴集团控股有限公司 | Account risk identification method and device |
| CN107392217A (en) * | 2016-05-17 | 2017-11-24 | 上海点融信息科技有限责任公司 | Computer implemented information processing method and device |
| CN107563645A (en) * | 2017-09-04 | 2018-01-09 | 杭州云算信达数据技术有限公司 | A kind of Financial Risk Analysis method based on big data |
-
2018
- 2018-06-21 CN CN201810643043.XA patent/CN110634060A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140143134A1 (en) * | 2010-01-13 | 2014-05-22 | Corelogic Solutions, Llc | System and method of detecting and assessing multiple types of risks related to mortgage lending |
| CN107301577A (en) * | 2016-04-15 | 2017-10-27 | 阿里巴巴集团控股有限公司 | Training method, credit estimation method and the device of credit evaluation model |
| CN107316198A (en) * | 2016-04-26 | 2017-11-03 | 阿里巴巴集团控股有限公司 | Account risk identification method and device |
| CN107392217A (en) * | 2016-05-17 | 2017-11-24 | 上海点融信息科技有限责任公司 | Computer implemented information processing method and device |
| CN106127363A (en) * | 2016-06-12 | 2016-11-16 | 腾讯科技（深圳）有限公司 | A kind of user credit appraisal procedure and device |
| CN106202491A (en) * | 2016-07-19 | 2016-12-07 | 中国银联股份有限公司 | A kind of based on seasonal effect in time series data digging method and system |
| CN106408184A (en) * | 2016-09-12 | 2017-02-15 | 中山大学 | User credit evaluation model based on multi-source heterogeneous data |
| CN107563645A (en) * | 2017-09-04 | 2018-01-09 | 杭州云算信达数据技术有限公司 | A kind of Financial Risk Analysis method based on big data |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111369139A (en) * | 2020-03-03 | 2020-07-03 | 电子科技大学 | Individual credit risk assessment method, system, terminal and storage medium |
| CN111507461A (en) * | 2020-04-13 | 2020-08-07 | 支付宝(杭州)信息技术有限公司 | Interpretability information determining method and device |
| CN111507461B (en) * | 2020-04-13 | 2022-05-17 | 支付宝(杭州)信息技术有限公司 | Interpretability information determining method and device |
| CN111861701A (en) * | 2020-07-09 | 2020-10-30 | 深圳市富之富信息技术有限公司 | Wind control model optimization method and device, computer equipment and storage medium |
| WO2022062193A1 (en) * | 2020-09-28 | 2022-03-31 | 南京博雅区块链研究院有限公司 | Individual credit assessment and explanation method and apparatus based on time sequence attribution analysis, and device and storage medium |
| CN115713987A (en) * | 2022-11-17 | 2023-02-24 | 广州瑞博新材料技术研究有限公司 | Polycaprolactone test data analysis method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106127363B (en) | User credit assessment method and device | |
| CN110634060A (en) | User credit risk assessment method, system, device and storage medium | |
| US20180260891A1 (en) | Systems and methods for generating and using optimized ensemble models | |
| CN108898479B (en) | Credit evaluation model construction method and device | |
| CN110909984B (en) | Business data processing model training method, business data processing method and device | |
| CN110738564A (en) | Post-loan risk assessment method and device and storage medium | |
| CN109635010B (en) | User characteristic and characteristic factor extraction and query method and system | |
| CN114048436A (en) | Construction method and construction device for forecasting enterprise financial data model | |
| CN112700324A (en) | User loan default prediction method based on combination of Catboost and restricted Boltzmann machine | |
| CN110930038A (en) | Loan demand identification method, loan demand identification device, loan demand identification terminal and loan demand identification storage medium | |
| CN111242131B (en) | Method, storage medium and device for identifying images in intelligent paper reading | |
| CN114663002A (en) | Method and equipment for automatically matching performance assessment indexes | |
| CN115271442A (en) | Modeling method and system for evaluating enterprise growth based on natural language | |
| CN111652430A (en) | Internet financial platform default rate prediction method and system | |
| CN110738565A (en) | Real estate finance artificial intelligence composite wind control model based on data set | |
| CN113344692A (en) | Method for establishing network loan credit risk assessment model with multi-information-source fusion | |
| CN111815458A (en) | Dynamic investment portfolio configuration method based on fine-grained quantitative marking and integration method | |
| CN116188174A (en) | Insurance fraud detection method and system based on modularity and mutual information | |
| CN110472680B (en) | Object classification method, device and computer-readable storage medium | |
| CN112288571A (en) | Personal credit risk assessment method based on rapid construction of neighborhood coverage | |
| CN118196567B (en) | Data evaluation method, device, equipment and storage medium based on large language model | |
| CN118070775B (en) | Performance evaluation method and device of abstract generation model and computer equipment | |
| CN117291708A (en) | Enterprise credit assessment method, system, equipment and medium | |
| CN115409607A (en) | Method and device for determining credit granting data and electronic equipment | |
| CN116976187A (en) | Modeling variable determining method, abnormal data prediction model construction method and device |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191231 |
|
| RJ01 | Rejection of invention patent application after publication |