CN111062444A - Credit risk prediction method, system, terminal and storage medium - Google Patents
Credit risk prediction method, system, terminal and storage medium Download PDFInfo
- Publication number
- CN111062444A CN111062444A CN201911331410.3A CN201911331410A CN111062444A CN 111062444 A CN111062444 A CN 111062444A CN 201911331410 A CN201911331410 A CN 201911331410A CN 111062444 A CN111062444 A CN 111062444A
- Authority
- CN
- China
- Prior art keywords
- credit risk
- sample
- preset
- algorithm
- trained
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000013058 risk prediction model Methods 0.000 claims abstract description 86
- 238000012549 training Methods 0.000 claims abstract description 59
- 238000012550 audit Methods 0.000 claims abstract description 24
- 238000004458 analytical method Methods 0.000 claims abstract description 12
- 238000004422 calculation algorithm Methods 0.000 claims description 124
- 238000007635 classification algorithm Methods 0.000 claims description 71
- 239000011159 matrix material Substances 0.000 claims description 70
- 230000004927 fusion Effects 0.000 claims description 39
- 230000006399 behavior Effects 0.000 claims description 21
- 238000009826 distribution Methods 0.000 claims description 16
- 238000007781 pre-processing Methods 0.000 claims description 14
- 238000001914 filtration Methods 0.000 claims description 10
- 238000013507 mapping Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 8
- 238000002790 cross-validation Methods 0.000 claims description 4
- 230000009471 action Effects 0.000 abstract description 4
- 230000008569 process Effects 0.000 description 10
- 238000004891 communication Methods 0.000 description 5
- 238000010606 normalization Methods 0.000 description 4
- 230000009466 transformation Effects 0.000 description 4
- 239000013598 vector Substances 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Finance (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Accounting & Taxation (AREA)
- Evolutionary Biology (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a credit risk prediction method, a system, a terminal and a storage medium, wherein the method comprises the following steps: training a credit risk prediction model to be trained by using the information of the user who has paid the action, and obtaining a preliminarily trained credit risk prediction model; predicting the credit risk grade of the user who fails in loan audit by adopting a preliminarily trained credit risk prediction model to obtain the credit risk prediction grade of the user who fails in loan audit; and training the preliminarily trained credit risk prediction model to obtain a final credit analysis prediction model according to the user information of the user who has made the payment behavior, the corresponding actual credit risk level, the user information of the user who fails in loan approval and the corresponding credit risk prediction level. The method and the device solve the problem that the accuracy rate of predicting the credit risk of the user in the existing credit scoring model is low.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a credit risk prediction method, a credit risk prediction system, a credit risk prediction terminal and a computer readable storage medium.
Background
In recent years, with the continuous development of internet finance, the online P2P loan market has gradually merged into human daily life. The online P2P loan market provides a convenient service that allows direct loan transactions between users. But this convenience also presents a significant potential risk to users, particularly investors. Therefore, how to predict the credit risk of the borrower becomes a problem to be solved urgently in the online P2P loan market.
The credit scoring model is presented to alleviate the problem to some extent, but the traditional credit scoring model is constructed based on the information of the users allowed to loan, and lacks the information of other users refused to loan, so that the model still has bias on the credit risk prediction of the users, and the accuracy of the risk prediction is low
Disclosure of Invention
The invention mainly aims to provide a credit risk prediction method, a credit risk prediction system, a credit risk prediction terminal and a computer readable storage medium, and aims to solve the technical problem that the accuracy rate of predicting the credit risk of a user in the existing credit scoring model is low.
In order to achieve the above object, the present invention provides a credit risk prediction method, comprising the steps of:
collecting information of a user who has paid back behavior as a first sample, and marking the actual credit risk level of the first sample according to a preset mapping relation between the payment behavior of the user and the credit risk level;
collecting user information which fails in loan audit as a second sample;
training a credit risk prediction model to be trained according to the plurality of first samples and the corresponding actual credit risk grades to obtain a preliminarily trained credit risk prediction model;
predicting the credit risk grade of the second sample by adopting a preliminarily trained credit risk prediction model to obtain the credit risk prediction grade of the second sample;
and training the preliminarily trained credit risk prediction model by using the plurality of first samples and the corresponding actual credit risk grades and the plurality of second samples and the corresponding credit risk prediction grades to obtain a final credit analysis prediction model.
Optionally, the step of collecting information of the user who has made the refund act as the first sample includes:
collecting information of a user who has paid a refund;
sensitive information filtering and/or preprocessing is carried out on the user information of which the payment returning behavior occurs;
and taking the filtered and/or preprocessed user information of the refund behavior as a first sample.
Optionally, the step of collecting the user information that the loan audit fails as the second sample includes:
collecting user information that the loan audit fails;
sensitive information filtering and/or preprocessing is carried out on the user information which fails in loan audit;
the filtered and/or preprocessed loan approval failing user information is used as a second sample.
Optionally, if the to-be-trained credit risk prediction model includes at least one preset different classification algorithm, at least one preset different clustering algorithm, and a to-be-trained fusion algorithm, the step of training the to-be-trained credit risk prediction model according to the plurality of first samples and the corresponding actual credit risk classes to obtain the preliminarily-trained credit risk prediction model includes:
simultaneously and respectively inputting a plurality of first samples into each preset classification algorithm and each preset clustering algorithm to obtain a first classification result output by each preset classification algorithm and a first clustering result output by each preset clustering algorithm, wherein the first classification result output by each preset classification algorithm comprises the probability that each first sample belongs to each credit risk grade and the credit risk prediction grade of each first sample, and the first clustering result output by each preset clustering algorithm comprises clusters with the same number as the number of the preset credit risk grades, the probability that each cluster belongs to each credit risk grade and the cluster to which each first sample belongs;
and training the fusion algorithm to be trained according to the preset cluster actual probability matrix, the actual credit risk grade of each first sample, the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm to obtain the preliminarily trained fusion algorithm.
Optionally, the training the fusion algorithm to be trained according to the preset cluster actual probability matrix, the actual credit risk level of each first sample, the first classification result output by each preset classification algorithm, and the first clustering result output by each preset clustering algorithm, and the step of obtaining the preliminarily trained fusion algorithm includes:
constructing a sample actual probability matrix according to the actual credit risk level of each first sample;
constructing a sample prediction average probability matrix, a cluster prediction average probability matrix, a distribution matrix and a membership matrix according to the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm;
inputting a preset cluster actual probability matrix, a sample prediction average probability matrix, a cluster prediction average probability matrix, a distribution matrix and a degree of identity matrix into a fusion algorithm to be trained, and acquiring initial parameters in the fusion algorithm to be trained by adopting a block coordinate descent algorithm.
Optionally, the step of training the credit risk prediction model to be trained according to the plurality of first samples and the corresponding actual credit risk classes to obtain the preliminarily trained credit risk prediction model further includes:
and training the credit risk prediction model to be trained according to the K-ten fold cross validation method and the plurality of first samples and the corresponding actual credit risk grades to obtain the preliminarily trained credit risk prediction model.
Optionally, the step of predicting the credit risk level of the second sample by using the preliminarily trained credit risk prediction model includes:
inputting a plurality of second samples into each preset classification algorithm and each preset clustering algorithm to obtain second classification results respectively output by each preset classification algorithm and second clustering results respectively output by each preset clustering algorithm;
and inputting the second classification result output by each preset classification algorithm and the second clustering result output by each preset clustering algorithm into a preliminary training fusion algorithm, and outputting the credit risk prediction grade of each second sample.
In addition, to achieve the above object, the present invention also provides a credit risk prediction system, including:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring information of a user who has paid back, serving as a first sample, and marking the actual credit risk level of the first sample according to a preset mapping relation between the payment behavior of the user and the credit risk level;
the second acquisition module is used for acquiring the user information which fails in loan audit as a second sample;
the first training module is used for training the credit risk prediction model to be trained according to the plurality of first samples and the corresponding actual credit risk grades to obtain a preliminarily trained credit risk prediction model;
the prediction module is used for predicting the credit risk grade of the second sample by adopting the preliminarily trained credit risk prediction model to obtain the credit risk prediction grade of the second sample;
and the second training module is used for training the preliminarily trained credit risk prediction model by using the plurality of first samples and the corresponding actual credit risk grades and the plurality of second samples and the corresponding credit risk prediction grades to obtain a final credit analysis prediction model.
Furthermore, to achieve the above object, the present invention also provides a terminal comprising a memory, a processor and a computer program stored on the memory and operable on the processor, the computer program, when executed by the processor, implementing the steps of the credit risk prediction method as described above.
Furthermore, to achieve the above object, the present invention also provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the credit risk prediction method as described above.
According to the credit risk prediction method, the credit risk prediction system, the credit risk prediction terminal and the computer readable storage medium, information of a user who has paid back is collected to serve as a first sample, and the actual credit risk level of the first sample is marked according to the preset mapping relation between the payment back behavior of the user and the credit risk level; collecting user information which fails in loan audit as a second sample; training a credit risk prediction model to be trained according to the plurality of first samples and the corresponding actual credit risk grades to obtain a preliminarily trained credit risk prediction model; predicting the credit risk grade of the second sample by adopting a preliminarily trained credit risk prediction model to obtain the credit risk prediction grade of the second sample; and training the preliminarily trained credit risk prediction model by using the plurality of first samples and the corresponding actual credit risk grades and the plurality of second samples and the corresponding credit risk prediction grades to obtain a final credit analysis prediction model. When the credit analysis prediction model is constructed, the model is initially trained by using the user information of the allowed loan, and then the model is retrained by using the user information of the allowed loan and the user information of the refused loan together, so that the obtained model has high accuracy in risk prediction of potential users meeting loan conditions, and the accuracy in risk prediction of users not meeting the loan conditions is improved, thereby integrally improving the accuracy in credit risk prediction of the model on the users.
Drawings
FIG. 1 is a schematic diagram of a hardware operating environment according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating a credit risk prediction method according to a first embodiment of the present invention;
FIG. 3 is a flowchart illustrating a detailed process of step S30 in the first embodiment of the credit risk prediction method according to the present invention;
FIG. 4 is a detailed flowchart of step S40 in the first embodiment of the credit risk prediction method of the present invention
FIG. 5 is a functional block diagram of a credit risk prediction system according to the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, fig. 1 is a schematic diagram of a hardware structure of a terminal provided in various embodiments of the present invention. The terminal comprises a communication module 01, a memory 02, a processor 03 and the like. Those skilled in the art will appreciate that the terminal shown in fig. 1 may also include more or fewer components than shown, or combine certain components, or a different arrangement of components. The processor 03 is connected to the memory 02 and the communication module 01, respectively, and the memory 02 stores a computer program, which is executed by the processor 03 at the same time.
The communication module 01 may be connected to an external device through a network. The communication module 01 may receive data sent by an external device, and may also send data, instructions, and information to the external device, where the external device may be an electronic device such as a mobile phone, a tablet computer, a notebook computer, and a desktop computer.
The memory 02 may be used to store software programs and various data. The memory 02 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program (building a distribution matrix) required for at least one function, and the like; the storage data area may store data or information created according to the use of the terminal, or the like. Further, the memory 02 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device.
The processor 03, which is a control center of the terminal, connects various parts of the entire terminal using various interfaces and lines, and performs various functions of the terminal and processes data by operating or executing software programs and/or modules stored in the memory 02 and calling data stored in the memory 02, thereby integrally monitoring the terminal. Processor 03 may include one or more processing units; preferably, the processor 03 may integrate an application processor, which mainly handles operating systems, user interfaces, application programs, etc., and a modem processor, which mainly handles wireless communications. It will be appreciated that the modem processor described above may not be integrated into the processor 03.
Although not shown in fig. 1, the terminal may further include a circuit control module, where the circuit control module is used for being connected to a mains supply to implement power control and ensure normal operation of other components.
Those skilled in the art will appreciate that the terminal structure shown in fig. 1 is not intended to be limiting and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
Various embodiments of the method of the present invention are presented in terms of the above-described hardware architecture.
Referring to fig. 2, in a first embodiment of the credit risk prediction method of the present invention, the credit risk prediction method includes the steps of:
step S10, collecting information of a user who has paid the action as a first sample, and marking the actual credit risk level of the first sample according to the preset mapping relation between the payment action of the user and the credit risk level;
in the scheme, information of a user who has paid is collected as a first sample, wherein the paying behaviors of the user include punctual paying, short-delay paying, long-delay paying and non-paying. The terminal marks the actual credit risk level of the first sample according to a preset mapping relation between the repayment behavior of the user and the credit risk level, the credit risk level can be set to be high, medium and low, or to be 5 levels of 1-5, and the setting of the credit risk level is not limited in the example.
Specifically, the step of collecting information of the user who has made the payment in step S10 as a first sample includes:
step S11, collecting the information of the user who has returned the money;
step S12, sensitive information filtering and/or preprocessing is carried out on the user information of the refund behavior;
and step S13, taking the filtered and/or preprocessed user information of the refund behavior as a first sample.
Since the collected user information may have sensitive information, such as an identification number, a name, a family member, and the like, which reveals the privacy of the user, it is necessary to automatically identify the sensitive information by using a keyword identification method, and filter, i.e., delete, the sensitive information. In order to improve the training effect on the model, before the model training is performed by using the information, the user information may be preprocessed, and the preprocessing includes normalization, standardization and the like, for example, the class type data is subjected to Onehot transformation, and the numerical type data is normalized. It is understood that the data pre-processing method includes, but is not limited to, One hot transformation and normalization as used in this example. After sensitive information filtering and/or preprocessing is carried out on the user information of the behavior of the refund, the filtered and/or preprocessed user information of the behavior of the refund can be directly used as a first sample.
Step S20, collecting user information which fails in loan audit as a second sample;
the terminal collects the information of users who fail in loan audit as a second sample, and the users do not have repayment behavior because the users are refused because the user does not pass the audit in the loan application process, so the second sample has no actual risk level.
Specifically, the step S20 includes:
step S21, collecting user information that the loan audit fails;
step S22, sensitive information filtering and/or preprocessing is carried out on the user information which fails in loan audit;
in step S23, the filtered and/or preprocessed user information that fails the loan audit is used as a second sample.
Since the collected user information may have sensitive information, such as an identification number, a name, a family member, and the like, which reveals the privacy of the user, it is necessary to automatically identify the sensitive information by using a keyword identification method, and filter, i.e., delete, the sensitive information. In order to improve the training effect on the model, before the model training is performed by using the information, the user information may be preprocessed, and the preprocessing includes normalization, standardization and the like, for example, the class type data is subjected to Onehot transformation, and the numerical type data is normalized. It is understood that the data pre-processing method includes, but is not limited to, One hot transformation and normalization as used in this example. After sensitive information filtering and/or preprocessing is carried out on the user information of the refused loan, the filtered and/or preprocessed user information of the refused loan can be directly used as a second sample.
Step S30, training a credit risk prediction model to be trained according to the plurality of first samples and the corresponding actual credit risk grades to obtain a preliminarily trained credit risk prediction model;
and training a credit risk prediction model to be trained by adopting a plurality of first samples and the actual credit risk grade corresponding to each sample to obtain initial parameters of the credit prediction model to be trained, and taking the credit risk prediction model with the initial parameters as a preliminarily trained credit risk prediction model.
Specifically, referring to fig. 3, fig. 3 is a flowchart illustrating a process of training a to-be-trained credit risk prediction model according to a plurality of first samples and corresponding actual credit risk classes if the to-be-trained credit risk prediction model includes at least one preset different classification algorithm and at least one preset different clustering algorithm, so as to obtain a preliminarily-trained credit risk prediction model, according to the above embodiment, the step S30 includes:
step S31, a plurality of first samples are simultaneously and respectively input into each preset classification algorithm and each preset clustering algorithm, and first classification results output by each preset classification algorithm and first clustering results output by each preset clustering algorithm are obtained, wherein the first classification results output by each preset classification algorithm comprise the probability that each first sample belongs to each credit risk grade and the credit risk prediction grade of each first sample, and the first clustering results output by each preset clustering algorithm comprise clusters with the same number as the number of the preset credit risk grades, the probability that each cluster belongs to each credit risk grade and the cluster to which each first sample belongs;
when the constructed risk prediction model to be trained comprises at least one preset different classification algorithm, at least one preset different clustering algorithm and a fusion algorithm to be trained, namely the risk prediction model is a semi-supervised learning model-a semi-supervised learning model based on a combination of a classification algorithm and a clustering algorithm. The classification algorithm adopted by the risk prediction model can be one or more of an NBC (Native Bayesian Classifier, naive Bayesian classification) algorithm, a logistic regression algorithm, various decision tree algorithms, an SVM (Support Vector Machine) algorithm, a K nearest neighbor algorithm, a neural network algorithm and the like; the adopted clustering algorithm can be one or more of a K mean value clustering algorithm, a K-MEDOIDS clustering algorithm, a hierarchical clustering algorithm, a GMM Gaussian mixture model, a graph group detection algorithm, a density-based clustering algorithm and the like. The preset classification algorithms and clustering algorithms in the risk prediction model to be trained are trained, and the classification algorithms and clustering algorithms are trained and verified by adopting a plurality of first samples before the risk prediction model to be trained is constructed.
And simultaneously and respectively inputting the plurality of first samples into each preset classification algorithm and each clustering algorithm in the risk prediction model to be trained by the terminal to obtain a first classification result output by each preset classification algorithm and a first clustering result output by each preset clustering algorithm. The first classification result output by each preset classification algorithm comprises the probability that each first sample belongs to each credit risk grade and the credit risk prediction grade of each first sample, and the first clustering result output by each preset clustering algorithm comprises clusters with the same number as the preset credit risk grades, the probability that each cluster belongs to each credit risk grade and the cluster class to which each first sample belongs
Step S32, training the fusion algorithm to be trained according to the preset cluster actual probability matrix, the actual credit risk level of each first sample, the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm, and obtaining the preliminarily trained fusion algorithm.
And training the fusion algorithm to be trained by using a preset cluster actual probability matrix, the actual credit risk grade of each first sample, the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm to obtain the preliminarily trained fusion algorithm.
The preset cluster actual probability matrix is as follows:
Specifically, the step S32 includes:
step S321, constructing a sample actual probability matrix according to the actual credit risk level of each first sample;
and constructing a sample actual probability matrix according to the actual credit risk level of each first sample, wherein the sample actual probability matrix is as follows:
Step S322, constructing a sample prediction average probability matrix, a cluster prediction average probability matrix, a distribution matrix and a membership matrix according to the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm;
constructing a sample prediction average probability matrix according to the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm, wherein the sample prediction average probability matrix is as follows:
According to the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm, a cluster prediction average probability matrix is constructed, wherein the cluster prediction average probability matrix is as follows:
According to the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm, a cluster distribution matrix is constructed, wherein the distribution matrix is as follows:
Constructing a membership matrix according to the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm, wherein the membership matrix is as follows:
Step S323, inputting a preset cluster actual probability matrix, a sample prediction average probability matrix, a cluster prediction average probability matrix, a distribution matrix and a degree of similarity matrix into a fusion algorithm to be trained, and acquiring initial parameters in the fusion algorithm to be trained by adopting a block coordinate descent algorithm.
Inputting a preset cluster actual probability matrix, a sample prediction average probability matrix, a cluster prediction average probability matrix, a distribution matrix and a membership matrix into a fusion algorithm to be trained, solving the fusion algorithm to be trained by adopting a block coordinate descent algorithm to obtain a preliminary parameter, and taking the fusion algorithm comprising the preliminary parameter as the fusion algorithm for the preliminary training.
The fusion algorithm is as follows:
The process of training the fusion algorithm is to solve and optimize the parameters of the fusion algorithm.
Step S40, predicting the credit risk grade of the second sample by adopting the credit risk prediction model of the preliminary training to obtain the credit risk prediction grade of the second sample;
and after the preliminarily trained credit risk prediction model is obtained, preliminarily predicting the credit risk grades of the plurality of second samples by adopting the preliminarily trained credit risk prediction model, and taking the preliminarily predicted result as the credit risk prediction grade of each second sample.
Specifically, referring to fig. 4, fig. 4 is a flowchart illustrating a step of predicting the credit risk level of the second sample by using a preliminarily trained credit risk prediction model in another embodiment of the present invention to obtain the credit risk prediction level of the second sample, where, based on the above embodiment, the step S40 includes:
step S41, inputting a plurality of second samples into each preset classification algorithm and each preset clustering algorithm to obtain second classification results respectively output by each preset classification algorithm and second clustering results respectively output by each preset clustering algorithm;
and inputting the plurality of second samples into each preset classification algorithm and each preset clustering algorithm in the preliminarily trained credit risk prediction model to obtain second classification results respectively output by each preset classification algorithm and second clustering results respectively output by each preset clustering algorithm.
And step S42, inputting the second classification results respectively output by the preset classification algorithms and the second clustering results respectively output by the preset clustering algorithms into the preliminarily trained fusion algorithm, and outputting the credit risk prediction grade of each second sample.
And then inputting the second classification result and the second classification result into a preliminarily trained fusion algorithm in a preliminarily trained credit risk prediction model, calculating each probability value corresponding to different credit risk prediction levels of each second sample, and taking the credit risk prediction level corresponding to the maximum probability value in the probability values of the sample as the credit risk prediction level of the sample.
And step S50, training the preliminarily trained credit risk prediction model by using the plurality of first samples and the corresponding actual credit risk grades and the plurality of second samples and the corresponding credit risk prediction grades to obtain a final credit analysis prediction model.
Taking the obtained multiple first samples and corresponding actual credit risk grades and the multiple second samples and corresponding credit risk prediction grades as training samples of the preliminarily trained credit risk prediction model, training the preliminarily trained credit risk prediction model, wherein the specific process of the preliminarily trained credit risk prediction model is the same as that of the step S30, and is not repeated here, and finally obtaining a final credit analysis prediction model.
According to the embodiment, the information of the user who has paid the money is collected to serve as a first sample, and the actual credit risk level of the first sample is marked according to the preset mapping relation between the payment behavior of the user and the credit risk level; collecting user information which fails in loan audit as a second sample; training a credit risk prediction model to be trained according to the plurality of first samples and the corresponding actual credit risk grades to obtain a preliminarily trained credit risk prediction model; predicting the credit risk grade of the second sample by adopting a preliminarily trained credit risk prediction model to obtain the credit risk prediction grade of the second sample; and training the preliminarily trained credit risk prediction model by using the plurality of first samples and the corresponding actual credit risk grades and the plurality of second samples and the corresponding credit risk prediction grades to obtain a final credit analysis prediction model. When the credit analysis prediction model is constructed, the model is initially trained by using the user information of the allowed loan, and then the model is retrained by using the user information of the allowed loan and the user information of the refused loan together, so that the obtained model has high accuracy in risk prediction of potential users meeting loan conditions, and the accuracy in risk prediction of users not meeting the loan conditions is improved, thereby integrally improving the accuracy in credit risk prediction of the model on the users.
Further, a second embodiment of the credit risk prediction method according to the first embodiment of the present application is proposed, in this embodiment, the step S30 further includes:
and step S33, training the credit risk prediction model to be trained according to the K-ten fold cross validation method and the plurality of first samples and the corresponding actual credit risk grades to obtain the preliminarily trained credit risk prediction model.
In this embodiment, the plurality of first samples are divided into ten groups (generally, equally divided) of sample subsets, each group of sample subsets is respectively used as a verification sample set, and the remaining K-1 groups of sample subsets are used as training sample sets, so that 10 pairs of training sample sets-verification sample sets are obtained, then each training sample set is used to train the credit risk prediction model to be trained, the stability of the corresponding candidate credit risk prediction model is obtained to evaluate, the stability evaluation result is obtained, and then the credit risk prediction model with the best stability performance result is selected as the primarily trained credit risk prediction model according to the stability evaluation result corresponding to each candidate credit risk prediction model. By adopting the K-V cross verification method, the occurrence of under-fitting and over-fitting can be effectively avoided, and the finally obtained model is also comparatively convincing.
The invention also provides a credit risk prediction system, which comprises:
the first acquisition module 10 is used for acquiring information of a user who has paid back, taking the information as a first sample, and marking the actual credit risk level of the first sample according to a preset mapping relation between the payment behavior of the user and the credit risk level;
the second acquisition module 20 is used for acquiring the user information which fails in loan audit as a second sample;
the first training module 30 is configured to train the credit risk prediction model to be trained according to the plurality of first samples and the corresponding actual credit risk grades, so as to obtain a preliminarily trained credit risk prediction model;
the prediction module 40 is configured to predict the credit risk level of the second sample by using the preliminarily trained credit risk prediction model, and obtain the credit risk prediction level of the second sample;
and the second training module 50 is used for training the preliminarily trained credit risk prediction model by using the plurality of first samples and the corresponding actual credit risk grades and the plurality of second samples and the corresponding credit risk prediction grades to obtain a final credit analysis prediction model.
Optionally, the first acquisition module 10 further includes:
the first acquisition unit is used for acquiring information of a user who has paid a payment;
the first processing unit is used for filtering and/or preprocessing sensitive information in the user information of which the payment returning action occurs;
and the first sample generating unit is used for taking the filtered and/or preprocessed user information of the refund behavior as a first sample.
Optionally, the second acquisition module 20 further comprises:
the second acquisition unit is used for acquiring the user information of which the loan audit fails;
the second processing unit is used for filtering and/or preprocessing sensitive information in the user information which fails in loan audit;
and the second sample generating unit is used for taking the user information which is filtered and/or preprocessed and fails in loan audit as a second sample.
Optionally, if the credit risk prediction model to be trained includes at least one preset different classification algorithm, at least one preset different clustering algorithm, and a fusion algorithm to be trained, the first training module 30 includes:
the first input unit is used for simultaneously and respectively inputting a plurality of first samples into each preset classification algorithm and each preset clustering algorithm to obtain a first classification result output by each preset classification algorithm and a first clustering result output by each preset clustering algorithm, wherein the first classification result output by each preset classification algorithm comprises the probability that each first sample belongs to each credit risk grade and the credit risk prediction grade of each first sample, and the first clustering result output by each preset clustering algorithm comprises clusters with the same number as the number of the preset credit risk grades, the probability that each cluster belongs to each credit risk grade and the cluster to which each first sample belongs;
and the first training unit is used for training the fusion algorithm to be trained according to the preset cluster actual probability matrix, the actual credit risk grade of each first sample, the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm to obtain the preliminarily trained fusion algorithm.
Optionally, the first training unit comprises:
the first construction subunit is used for constructing a sample actual probability matrix according to the actual credit risk level of each first sample;
the second construction subunit is used for constructing a sample prediction average probability matrix, a cluster prediction average probability matrix, a distribution matrix and a membership matrix according to the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm;
and the obtaining subunit is used for inputting a preset cluster actual probability matrix, a sample prediction average probability matrix, a cluster prediction average probability matrix, a distribution matrix and a degree of identity matrix into the fusion algorithm to be trained, and obtaining initial parameters in the fusion algorithm to be trained by adopting a block coordinate descent algorithm.
Optionally, the first training module 30 further comprises:
and the second training unit is used for training the credit risk prediction model to be trained according to the K-ten fold cross validation method, the plurality of first samples and the corresponding actual credit risk grades to obtain the preliminarily trained credit risk prediction model.
Optionally, the prediction module 40 comprises:
the second input unit is used for inputting the plurality of second samples into each preset classification algorithm and each preset clustering algorithm to obtain second classification results respectively output by each preset classification algorithm and second clustering results respectively output by each preset clustering algorithm;
and the output unit is used for inputting the second classification results respectively output by the preset classification algorithms and the second clustering results respectively output by the preset clustering algorithms into the preliminarily trained fusion algorithm and outputting the credit risk prediction grade of each second sample.
The invention also proposes a computer-readable storage medium on which a computer program is stored. The computer readable storage medium may be the Memory 02 in the credit risk prediction terminal of fig. 1, and may also be at least one of a ROM (Read-only Memory)/RAM (Random Access Memory), a magnetic disk and an optical disk, and the computer readable storage medium includes several pieces of information for enabling the credit risk prediction terminal to perform the method according to the embodiments of the present invention.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system 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 system. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (10)
1. A method for predicting credit risk, the method comprising the steps of:
collecting information of a user who has paid back behavior as a first sample, and marking the actual credit risk level of the first sample according to a preset mapping relation between the payment behavior of the user and the credit risk level;
collecting user information which fails in loan audit as a second sample;
training a credit risk prediction model to be trained according to the plurality of first samples and the corresponding actual credit risk grades to obtain a preliminarily trained credit risk prediction model;
predicting the credit risk grade of the second sample by adopting a preliminarily trained credit risk prediction model to obtain the credit risk prediction grade of the second sample;
and training the preliminarily trained credit risk prediction model by using the plurality of first samples and the corresponding actual credit risk grades and the plurality of second samples and the corresponding credit risk prediction grades to obtain a final credit analysis prediction model.
2. The method of claim 1, wherein the step of collecting information of the user who has made a refund act as the first sample comprises:
collecting information of a user who has paid a refund;
sensitive information filtering and/or preprocessing is carried out on the user information of which the payment returning behavior occurs;
and taking the filtered and/or preprocessed user information of the refund behavior as a first sample.
3. The method of claim 1, wherein the step of collecting user information that the loan audit has failed as a second sample comprises:
collecting user information that the loan audit fails;
sensitive information filtering and/or preprocessing is carried out on the user information which fails in loan audit;
the filtered and/or preprocessed loan approval failing user information is used as a second sample.
4. The method according to any one of claims 1 to 3, wherein if the credit risk prediction model to be trained includes at least one of a predetermined different classification algorithm, at least one of a predetermined different clustering algorithm, and a fusion algorithm to be trained, the step of training the credit risk prediction model to be trained according to the plurality of first samples and the corresponding actual credit risk levels to obtain the preliminarily trained credit risk prediction model includes:
simultaneously and respectively inputting a plurality of first samples into each preset classification algorithm and each preset clustering algorithm to obtain a first classification result output by each preset classification algorithm and a first clustering result output by each preset clustering algorithm, wherein the first classification result output by each preset classification algorithm comprises the probability that each first sample belongs to each credit risk grade and the credit risk prediction grade of each first sample, and the first clustering result output by each preset clustering algorithm comprises clusters with the same number as the number of the preset credit risk grades, the probability that each cluster belongs to each credit risk grade and the cluster to which each first sample belongs;
and training the fusion algorithm to be trained according to the preset cluster actual probability matrix, the actual credit risk grade of each first sample, the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm to obtain the preliminarily trained fusion algorithm.
5. The method for predicting credit risk according to claim 4, wherein the step of training the fusion algorithm to be trained according to the preset cluster actual probability matrix, the actual credit risk level of each first sample, the first classification result output by each preset classification algorithm, and the first clustering result output by each preset clustering algorithm to obtain the preliminarily trained fusion algorithm comprises:
constructing a sample actual probability matrix according to the actual credit risk level of each first sample;
constructing a sample prediction average probability matrix, a cluster prediction average probability matrix, a distribution matrix and a membership matrix according to the first classification result output by each preset classification algorithm and the first clustering result output by each preset clustering algorithm;
inputting a preset cluster actual probability matrix, a sample prediction average probability matrix, a cluster prediction average probability matrix, a distribution matrix and a degree of identity matrix into a fusion algorithm to be trained, and acquiring initial parameters in the fusion algorithm to be trained by adopting a block coordinate descent algorithm.
6. The method according to claim 5, wherein the step of training the credit risk prediction model to be trained according to the plurality of first samples and the corresponding actual credit risk classes to obtain the preliminarily trained credit risk prediction model further comprises:
and training the credit risk prediction model to be trained according to the K-ten fold cross validation method and the plurality of first samples and the corresponding actual credit risk grades to obtain the preliminarily trained credit risk prediction model.
7. The method according to claim 5, wherein the step of predicting the credit risk level of the second sample using the preliminarily trained credit risk prediction model comprises:
inputting a plurality of second samples into each preset classification algorithm and each preset clustering algorithm to obtain second classification results respectively output by each preset classification algorithm and second clustering results respectively output by each preset clustering algorithm;
and inputting the second classification result output by each preset classification algorithm and the second clustering result output by each preset clustering algorithm into a preliminary training fusion algorithm, and outputting the credit risk prediction grade of each second sample.
8. A credit risk level prediction system, the system comprising:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring information of a user who has paid back, serving as a first sample, and marking the actual credit risk level of the first sample according to a preset mapping relation between the payment behavior of the user and the credit risk level;
the second acquisition module is used for acquiring the user information which fails in loan audit as a second sample;
the first training module is used for training the credit risk prediction model to be trained according to the plurality of first samples and the corresponding actual credit risk grades to obtain a preliminarily trained credit risk prediction model;
the prediction module is used for predicting the credit risk grade of the second sample by adopting the preliminarily trained credit risk prediction model to obtain the credit risk prediction grade of the second sample;
and the second training module is used for training the preliminarily trained credit risk prediction model by using the plurality of first samples and the corresponding actual credit risk grades and the plurality of second samples and the corresponding credit risk prediction grades to obtain a final credit analysis prediction model.
9. A terminal, characterized in that the terminal comprises a memory, a processor and a computer program stored on the memory and executable on the processor, which computer program, when executed by the processor, carries out the steps of the credit risk prediction method according to any one of claims 1 to 7.
10. A computer-readable storage medium, having stored thereon a computer program which, when being executed by a processor, carries out the steps of the credit risk prediction method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911331410.3A CN111062444B (en) | 2019-12-21 | 2019-12-21 | Credit risk prediction method, credit risk prediction system, credit risk prediction terminal and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911331410.3A CN111062444B (en) | 2019-12-21 | 2019-12-21 | Credit risk prediction method, credit risk prediction system, credit risk prediction terminal and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111062444A true CN111062444A (en) | 2020-04-24 |
| CN111062444B CN111062444B (en) | 2023-12-08 |
Family
ID=70301409
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911331410.3A Active CN111062444B (en) | 2019-12-21 | 2019-12-21 | Credit risk prediction method, credit risk prediction system, credit risk prediction terminal and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111062444B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111582565A (en) * | 2020-04-26 | 2020-08-25 | 支付宝(杭州)信息技术有限公司 | Data fusion method and device and electronic equipment |
| CN111695989A (en) * | 2020-06-18 | 2020-09-22 | 新分享科技服务(深圳)有限公司 | Modeling method and platform of wind-control credit model |
| CN111798047A (en) * | 2020-06-30 | 2020-10-20 | 平安普惠企业管理有限公司 | Wind control prediction method and device, electronic equipment and storage medium |
| CN113744049A (en) * | 2021-09-18 | 2021-12-03 | 中国银行股份有限公司 | Personal credit investigation score obtaining method, device, server and medium |
| CN116228139A (en) * | 2023-01-10 | 2023-06-06 | 南京田雨企业咨询管理有限公司 | Automobile financial risk auditing method and device, electronic equipment and storage medium |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130346284A1 (en) * | 2012-06-26 | 2013-12-26 | Lauren Nicole Cozart Stubbs | Novel systems and processes for enhanced microlending |
| CN106897918A (en) * | 2017-02-24 | 2017-06-27 | 上海易贷网金融信息服务有限公司 | A kind of hybrid machine learning credit scoring model construction method |
| WO2018090657A1 (en) * | 2016-11-18 | 2018-05-24 | 同济大学 | Bp_adaboost model-based method and system for predicting credit card user default |
| CN109460795A (en) * | 2018-12-17 | 2019-03-12 | 北京三快在线科技有限公司 | Classifier training method, apparatus, electronic equipment and computer-readable medium |
| WO2019061989A1 (en) * | 2017-09-30 | 2019-04-04 | 平安科技（深圳）有限公司 | Loan risk control method, electronic device and readable storage medium |
| CN109670940A (en) * | 2018-11-12 | 2019-04-23 | 深圳壹账通智能科技有限公司 | Credit Risk Assessment Model generation method and relevant device based on machine learning |
| CN110009479A (en) * | 2019-03-01 | 2019-07-12 | 百融金融信息服务股份有限公司 | Credit assessment method and device, storage medium, computer equipment |
| CN110232405A (en) * | 2019-05-24 | 2019-09-13 | 东方银谷(北京)科技发展有限公司 | Method and device for personal credit file |
| CN110570300A (en) * | 2019-09-10 | 2019-12-13 | 武汉科技大学 | XGboost-driven Haishani conversion financial decision processing system |
-
2019
- 2019-12-21 CN CN201911331410.3A patent/CN111062444B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130346284A1 (en) * | 2012-06-26 | 2013-12-26 | Lauren Nicole Cozart Stubbs | Novel systems and processes for enhanced microlending |
| WO2018090657A1 (en) * | 2016-11-18 | 2018-05-24 | 同济大学 | Bp_adaboost model-based method and system for predicting credit card user default |
| CN106897918A (en) * | 2017-02-24 | 2017-06-27 | 上海易贷网金融信息服务有限公司 | A kind of hybrid machine learning credit scoring model construction method |
| WO2019061989A1 (en) * | 2017-09-30 | 2019-04-04 | 平安科技（深圳）有限公司 | Loan risk control method, electronic device and readable storage medium |
| CN109670940A (en) * | 2018-11-12 | 2019-04-23 | 深圳壹账通智能科技有限公司 | Credit Risk Assessment Model generation method and relevant device based on machine learning |
| CN109460795A (en) * | 2018-12-17 | 2019-03-12 | 北京三快在线科技有限公司 | Classifier training method, apparatus, electronic equipment and computer-readable medium |
| CN110009479A (en) * | 2019-03-01 | 2019-07-12 | 百融金融信息服务股份有限公司 | Credit assessment method and device, storage medium, computer equipment |
| CN110232405A (en) * | 2019-05-24 | 2019-09-13 | 东方银谷(北京)科技发展有限公司 | Method and device for personal credit file |
| CN110570300A (en) * | 2019-09-10 | 2019-12-13 | 武汉科技大学 | XGboost-driven Haishani conversion financial decision processing system |
Non-Patent Citations (2)
| Title |
|---|
| HUAWEI NIU等: "An endogenous structural credit risk model incorporating with moral hazard and rollover risk" * |
| 王嘉琪: "基于数据挖掘技术的P2P借贷违约风险识别模型研究" * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111582565A (en) * | 2020-04-26 | 2020-08-25 | 支付宝(杭州)信息技术有限公司 | Data fusion method and device and electronic equipment |
| CN111695989A (en) * | 2020-06-18 | 2020-09-22 | 新分享科技服务(深圳)有限公司 | Modeling method and platform of wind-control credit model |
| CN111798047A (en) * | 2020-06-30 | 2020-10-20 | 平安普惠企业管理有限公司 | Wind control prediction method and device, electronic equipment and storage medium |
| CN113744049A (en) * | 2021-09-18 | 2021-12-03 | 中国银行股份有限公司 | Personal credit investigation score obtaining method, device, server and medium |
| CN113744049B (en) * | 2021-09-18 | 2024-05-24 | 中国银行股份有限公司 | Personal credit score acquisition method, device, server and medium |
| CN116228139A (en) * | 2023-01-10 | 2023-06-06 | 南京田雨企业咨询管理有限公司 | Automobile financial risk auditing method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111062444B (en) | 2023-12-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111062444B (en) | Credit risk prediction method, credit risk prediction system, credit risk prediction terminal and storage medium | |
| CN111553759A (en) | Product information pushing method, device, equipment and storage medium | |
| CN113011889B (en) | Account anomaly identification method, system, device, equipment and medium | |
| CN111368926B (en) | Image screening method, device and computer readable storage medium | |
| CN110288459A (en) | Loan prediction technique, device, equipment and storage medium | |
| CN111639690A (en) | Fraud analysis method, system, medium, and apparatus based on relational graph learning | |
| CN112487284A (en) | Bank customer portrait generation method, equipment, storage medium and device | |
| Eddy et al. | Credit scoring models: Techniques and issues | |
| CN112990989B (en) | Value prediction model input data generation method, device, equipment and medium | |
| CN110765110A (en) | Generalization capability processing method, device, equipment and storage medium | |
| CN111352926A (en) | Data processing method, device, equipment and readable storage medium | |
| CN108629381A (en) | Crowd's screening technique based on big data and terminal device | |
| US11921821B2 (en) | System and method for labelling data for trigger identification | |
| CN107402984B (en) | A kind of classification method and device based on theme | |
| CN112364136B (en) | Keyword generation method, device, equipment and storage medium | |
| Fermanian et al. | Fair learning with bagging | |
| CN114170000A (en) | Credit card user risk category identification method, device, computer equipment and medium | |
| Nureni et al. | Loan approval prediction based on machine learning approach | |
| CN111752985A (en) | Method, device and storage medium for generating main portrait | |
| US11983162B2 (en) | Change management process for identifying potential regulatory violations for improved processing efficiency | |
| US11989777B2 (en) | Pairing and grouping user profiles accessed from pre-current systems | |
| US20230334307A1 (en) | Training an artificial intelligence engine to predict a user likelihood of attrition | |
| US20230342351A1 (en) | Change management process for identifying inconsistencies for improved processing efficiency | |
| US20240193581A1 (en) | Card customization | |
| US12020092B1 (en) | Generating an encoded document |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |