CN116664274A - Risk assessment method and device, storage medium and electronic equipment - Google Patents
Risk assessment method and device, storage medium and electronic equipment Download PDFInfo
- Publication number
- CN116664274A CN116664274A CN202310614843.XA CN202310614843A CN116664274A CN 116664274 A CN116664274 A CN 116664274A CN 202310614843 A CN202310614843 A CN 202310614843A CN 116664274 A CN116664274 A CN 116664274A
- Authority
- CN
- China
- Prior art keywords
- evaluation
- model
- credit
- evaluation model
- target
- 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 73
- 238000012502 risk assessment Methods 0.000 title claims abstract description 42
- 238000013210 evaluation model Methods 0.000 claims abstract description 217
- 238000011156 evaluation Methods 0.000 claims abstract description 75
- 238000012795 verification Methods 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims description 106
- 238000002790 cross-validation Methods 0.000 claims description 50
- 230000006870 function Effects 0.000 claims description 33
- 238000004590 computer program Methods 0.000 claims description 6
- 238000012911 target assessment Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 4
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 26
- 238000010586 diagram Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 8
- 238000012545 processing Methods 0.000 description 8
- 238000010200 validation analysis Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000011835 investigation Methods 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- RWSOTUBLDIXVET-UHFFFAOYSA-N Dihydrogen sulfide Chemical compound S RWSOTUBLDIXVET-UHFFFAOYSA-N 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012954 risk control Methods 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
- 230000007306 turnover Effects 0.000 description 1
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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Business, Economics & Management (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Technology Law (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention discloses a risk assessment method, a risk assessment device, a storage medium and electronic equipment. Relates to the field of artificial intelligence, and the method comprises the following steps: obtaining target information of an object to be borrowed, wherein the target information at least comprises: information characterizing social relationships of objects to be borrowed; evaluating the target information through a target evaluation model to obtain a credit evaluation result, wherein model parameters of the target evaluation model are determined through multiple times of cross verification; and determining the credit rating of the object to be borrowed according to the credit evaluation result, and determining the borrowing risk of borrowing the object to be borrowed according to the credit rating of the object to be borrowed. The invention solves the technical problem that the borrowing risk cannot be accurately determined due to low evaluation accuracy of the credit of the user when evaluating the borrowing risk according to the credit of the borrowing user in the prior art.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to a risk assessment method, a risk assessment device, a storage medium and electronic equipment.
Background
With the rapid development of economy, financial institution-based extended borrowing services are more and more, and the extended borrowing services generally relate to funds turnover in consumption, life and other aspects, and have the characteristics of rapidness, convenience, flexibility, no mortgage and the like. However, since borrowers are mostly low-income groups, credit conditions are not stable, and the borrowing businesses have the problem of high risk at the same time, so that the risk control for the borrowers is particularly important.
At present, the traditional wind control model usually adopts modes such as manual screening and expert experience to carry out risk assessment on borrowers, so that the problems of strong subjectivity, difficulty in ensuring accuracy and the like of manual judgment exist.
In view of the above problems, no effective solution has been proposed at present.
Disclosure of Invention
The embodiment of the invention provides a risk assessment method, a risk assessment device, a storage medium and electronic equipment, which are used for at least solving the technical problem that the borrowing risk cannot be accurately determined due to low assessment accuracy of credit of a user when the borrowing risk is assessed according to the credit of the borrowing user in the prior art.
According to an aspect of an embodiment of the present invention, there is provided a risk assessment method, including: obtaining target information of an object to be borrowed, wherein the target information at least comprises: information characterizing social relationships of objects to be borrowed; evaluating the target information through a target evaluation model to obtain a credit evaluation result, wherein model parameters of the target evaluation model are determined through multiple times of cross verification; and determining the credit rating of the object to be borrowed according to the credit evaluation result, and determining the borrowing risk of borrowing the object to be borrowed according to the credit rating of the object to be borrowed.
Further, the risk assessment method further comprises: constructing N first evaluation models according to N groups of first model parameters and an initial model structure, wherein the N groups of first model parameters are parameters obtained by adjusting the same group of model parameters, and N is a positive integer greater than 1; for each first evaluation model, performing M times of cross validation on the first evaluation model, and determining the performance score of the first evaluation model according to the results of the M times of cross validation, wherein M is a positive integer greater than 1; and screening the model with the highest performance score from the N first evaluation models to obtain a third evaluation model, and training the third evaluation model by adopting a training sample set to obtain a target evaluation model, wherein the training sample set consists of sample target information of a sample object and a real credit label of the sample object.
Further, the risk assessment method further comprises: before N first evaluation models are constructed according to N groups of first model parameters and an initial model structure, constructing an evaluation model according to preset model parameters, and training the evaluation model by adopting a training sample set to obtain an initial evaluation model; and according to the parameter adjustment rule, adjusting the model parameters in the initial evaluation model for N times to obtain N groups of first model parameters.
Further, the risk assessment method further comprises: training a first evaluation model through M times of cross validation based on a training sample set to obtain P second evaluation models, and determining the performance score of each second evaluation model, wherein each time of cross validation divides the training sample set into k subsets, the subsets obtained by each time of division are different, M, P, k is a positive integer greater than 1, and P=M×k; and determining the performance scores of the first evaluation model according to the performance scores of the P second evaluation models.
Further, the risk assessment method further comprises: for each cross-validation, dividing the training sample set into k subsets; carrying out k times of training on the first evaluation model according to k subsets to obtain k second evaluation models, wherein each time of training in the k times of training adopts k-1 subsets in the k subsets, and the k-1 subsets used in each time of training are not completely identical; and inputting the target subset into the second evaluation model for each second evaluation model to obtain a credit evaluation result output by the second evaluation model, and evaluating the performance score of the second evaluation model according to the credit evaluation result output by the second evaluation model, wherein the target subset is a subset which is not used when the second evaluation model is trained in k subsets.
Further, the risk assessment method further comprises: inputting the training sample set into a third evaluation model to obtain a sample credit evaluation result output by the third evaluation model; acquiring a weight value matched with each real credit label, wherein different real credit labels are matched with different weight values; and calculating a loss function value according to the sample credit evaluation result, the real credit label and the weight value, and optimizing model parameters of the third evaluation model according to the loss function value until the loss function value is smaller than or equal to a preset value, so as to obtain a target evaluation model.
Further, the risk assessment method further comprises: acquiring initial sample target information of a sample object and a real credit label; determining the number of objects of the sample objects matched with each real credit label, and determining the real credit label with the least number of matched objects as a target real credit label; generating synthetic sample target information according to initial sample target information matched with the target real credit label; and taking the initial sample target information or the synthesized sample target information as sample target information, and forming a training sample set by the sample target information and the real credit label.
According to another aspect of the embodiment of the present invention, there is also provided a risk assessment apparatus, including: the first obtaining module is configured to obtain target information of an object to be borrowed, where the target information at least includes: information characterizing social relationships of objects to be borrowed; the evaluation module is used for evaluating the target information through a target evaluation model to obtain a credit evaluation result, wherein the model parameters of the target evaluation model are determined through multiple times of cross verification; the first determining module is used for determining the credit rating of the object to be borrowed according to the credit evaluation result, and determining the borrowing risk of borrowing the object to be borrowed according to the credit rating of the object to be borrowed.
According to another aspect of the embodiments of the present invention, there is also provided a computer readable storage medium having a computer program stored therein, wherein the computer program is configured to perform the risk assessment method described above when run.
According to another aspect of an embodiment of the present invention, there is also provided an electronic device including one or more processors; and a memory for storing one or more programs that, when executed by the one or more processors, cause the one or more processors to implement a method for running the program, wherein the program is configured to perform the risk assessment method described above when run.
In the embodiment of the invention, the target evaluation model is obtained through multiple times of cross verification, so that the borrowing risk of the object to be borrowed is determined according to the target evaluation model, the target information of the object to be borrowed is obtained, then the target information is evaluated through the target evaluation model, the credit evaluation result is obtained, the credit grade of the object to be borrowed is determined according to the credit evaluation result, and the borrowing risk of the object to be borrowed is determined according to the credit grade of the object to be borrowed. Wherein, the target information at least comprises: and (3) representing the information of the social relationship of the object to be borrowed, wherein the model parameters of the target evaluation model are determined through multiple times of cross verification.
It is easy to note that in the above process, since the training sample set for training the model is randomly divided in the process of cross-validation, by determining the model parameters of the target evaluation model based on multiple cross-validation, the influence of random division on the evaluation of the model parameters in the process of determining the model parameters by single cross-validation is avoided, the accuracy of determining the model parameters is low, and the evaluation effect of the target evaluation model is affected, so that the stability and accuracy of credit evaluation are improved, and the credit evaluation result realizes effective description of the credit of the object to be borrowed. Further, the borrowing risk of borrowing the object to be borrowed is determined according to the credit evaluation result, accurate and rapid determination of the borrowing risk is achieved, and the problem that evaluation accuracy is poor due to strong subjectivity in manual evaluation is avoided.
Therefore, the scheme provided by the application achieves the purpose of obtaining the target evaluation model through multiple times of cross verification, so that the borrowing risk of the object to be borrowed is determined according to the target evaluation model, the technical effect of accurately determining the borrowing risk is realized, and the technical problem that the borrowing risk cannot be accurately determined due to low evaluation accuracy of the credit of the user when the borrowing risk is evaluated according to the credit of the borrowing user in the prior art is solved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute a limitation on the application. In the drawings:
FIG. 1 is a schematic diagram of an alternative risk assessment method according to an embodiment of the present application;
FIG. 2 is a schematic diagram of an alternative generation target assessment model according to an embodiment of the present application;
FIG. 3 is a schematic diagram of an alternative risk assessment apparatus according to an embodiment of the present application;
fig. 4 is a schematic diagram of an alternative electronic device according to an embodiment of the application.
Detailed Description
In order that those skilled in the art will better understand the present application, a technical solution in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present application without making any inventive effort, shall fall within the scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the application described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or fully authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region, and provide corresponding operation entries for the user to select authorization or rejection.
Example 1
According to an embodiment of the present invention, there is provided an embodiment of a risk assessment method, it being noted that the steps shown in the flowchart of the drawings may be performed in a computer system such as a set of computer executable instructions, and, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in an order other than that shown or described herein.
FIG. 1 is a schematic diagram of an alternative risk assessment method according to an embodiment of the present invention, as shown in FIG. 1, the method comprising the steps of:
step S101, obtaining target information of an object to be borrowed, wherein the target information at least comprises: information characterizing social relationships of objects to be borrowed.
Alternatively, the electronic device, the application system, the server, and the like may be used as an execution subject of the risk assessment method, and in this embodiment, the risk assessment system is used as an execution subject to obtain the target information of the object to be borrowed. Optionally, the object to be borrowed is a user borrowing to a financial institution, the target information is obtained by performing data preprocessing on initial information of the object to be borrowed, and the initial information includes social relationship information of the object to be borrowed at least, and may further include user information, history borrowing information, credit investigation information and the like of the object to be borrowed. Optionally, the social relationship information may be determined according to the registration information of the user at the foregoing financial institution and the fund transfer information of the user, where the social relationship information includes but is not limited to a friend relationship, a relative relationship, and the like of the user, for example, the user a and the user B each transact a bank card at the foregoing financial institution, and if the number of transfers between the user a and the user B is greater than a preset number, it is determined that the user a and the user B are friends with each other. The user information may be submitted when the user applies for the borrowing, or the user is registered in a financial institution before, the user information includes but is not limited to information such as age, gender, education level, occupation, etc., the history borrowing information is history borrowing information of the user in the foregoing financial institution, the history borrowing information includes but is not limited to borrowing amount, borrowing period, borrowing purpose, repayment condition, etc., the credit investigation information may be recorded by the foregoing financial institution, and the credit investigation information includes but is not limited to credit score, credit card amount usage, repayment record, overdue record, liability condition, etc.
An optional process of determining target information from initial information is described. Optionally, after the initial information is obtained, the risk assessment system may perform data cleaning processing on the initial information to remove noise, abnormal values, missing values and the like in the initial information, so as to ensure accuracy and integrity of data. Optionally, the data cleansing process includes, but is not limited to:
(1) Removing abnormal values: taking the data exceeding the preset range and the preset threshold value in the data as abnormal values, and deleting or correcting the data;
(2) Filling up missing values: because some missing values possibly exist in the data acquisition process, some methods are needed to be adopted for filling, such as mean value, median, mode, interpolation and the like;
(3) Removing the duplicate values: if duplicate data exists in the data set, deduplication processing is required to ensure the uniqueness of the data.
Further, after the data cleansing process is performed, in order to improve the usability and effectiveness of the data, a data conversion process may be performed on the initial information to convert the initial information into a form suitable for modeling and analysis, thereby obtaining target information. Optionally, the data conversion process includes at least one of:
(1) And (5) digitizing: the non-numeric data is converted into numeric data for subsequent processing and analysis.
(2) Standardization: the data are converted into the form that the mean value is 0 and the standard deviation is 1, so that dimension and proportion differences of the data are eliminated, and the prediction accuracy of the model is improved.
(3) Regularization: the data is converted into a form in the range of 0-1, so that the amplitude and the bias of the data are eliminated, and the stability of the model is improved.
And step S102, evaluating the target information through a target evaluation model to obtain a credit evaluation result, wherein the model parameters of the target evaluation model are determined through multiple times of cross verification.
Further, after the target information is obtained, the target information may be input into a target evaluation model to evaluate the target information by the target evaluation model, where in this embodiment, the target evaluation model is an XGBoost (extreme gradient boost tree, eXtreme Gradient Boosting) model, and the target evaluation model processes the target information to output a probability value, that is, the aforementioned credit evaluation result, where the probability value is used to characterize the credit of the user.
The target evaluation model is a trained model, and model parameters are optimized through k-fold cross validation (namely the cross validation) in the training process of the target evaluation model. Alternatively, the risk assessment system may acquire a plurality of models generated from different parameter values, then perform multiple cross-verifications on each model, thereby selecting a relatively better model from the plurality of models according to the results of the multiple cross-verifications, and then determine the target assessment model based on the relatively better model.
Specifically, in single k-fold cross validation, the training sample set is divided into k non-overlapping subsets, one subset is used as a validation set each time, the rest k-1 subsets are used as training sets for training, so that k times of training and validation are performed, and finally, the performance score of the training model in the current k-fold cross validation is determined according to the results obtained by the k times of training and validation. The final performance score of the model can be further determined according to the performance scores of the model in multiple cross-validation, and the final performance score is used for representing the quality degree of the model.
It should be noted that, because the training sample set for training the model is randomly divided in the process of cross-validation, the model parameters of the target evaluation model are determined based on multiple times of cross-validation, so that the influence of random division on the evaluation of the model parameters in the process of determining the model parameters of the target evaluation model by adopting single time of cross-validation is avoided, the accuracy of determining the model parameters is low, the evaluation effect of the target evaluation model is affected, and the problem that the borrowing risk cannot be accurately determined is further caused, thereby improving the stability and accuracy of credit evaluation.
Step S103, determining the credit rating of the object to be borrowed according to the credit evaluation result, and determining the borrowing risk of borrowing the object to be borrowed according to the credit rating of the object to be borrowed.
Optionally, since the credit evaluation result is a probability value, the [0,1] interval may be divided, and a matching relationship between different intervals obtained by the division and different credit levels may be set, so that the credit level of the object to be borrowed is determined according to the interval to which the credit evaluation result belongs. For example, [0,0.5 ] is set to match the first credit rating, and (0.5, 1) is set to match the second credit rating, then when the credit evaluation result is 0.3, the credit rating of the object to be borrowed is determined to be the first credit rating.
Further, a matching relationship between the credit rating and the borrowing risk may be preset, for example, when the credit rating is the first credit rating, it is determined that the borrowing risk is high, and the borrowing is not allowed to be given to the object to be borrowed, whereas when the credit rating is the second credit rating, it is determined that the borrowing risk is low, and the borrowing is allowed to be given to the object to be borrowed. It should be noted that the credit level may be divided into more than two levels, and the borrowing risk may be divided into risk levels with finer granularity according to the number of levels of the credit level, so as to assist in determining the borrowing amount based on determining whether to borrow according to the risk levels.
It should be noted that, the credit evaluation result output by the target evaluation model realizes the effective description of the credit degree of the object to be borrowed, so that the borrowing risk determined according to the credit evaluation result is more accurate.
Based on the above-mentioned schemes defined in steps S101 to S103, it may be known that, in the embodiment of the present invention, a mode of obtaining the target evaluation model through multiple cross-verifications is adopted, so as to determine the borrowing risk of the object to be borrowed according to the target evaluation model, and the credit evaluation result is obtained by obtaining the target information of the object to be borrowed and then evaluating the target information through the target evaluation model, so that the credit rating of the object to be borrowed is determined according to the credit evaluation result, and the borrowing risk of borrowing to the object to be borrowed is determined according to the credit rating of the object to be borrowed. Wherein, the target information at least comprises: and (3) representing the information of the social relationship of the object to be borrowed, wherein the model parameters of the target evaluation model are determined through multiple times of cross verification.
It is easy to note that in the above process, since the training sample set for training the model is randomly divided in the process of cross-validation, by determining the model parameters of the target evaluation model based on multiple cross-validation, the influence of random division on the evaluation of the model parameters in the process of determining the model parameters by single cross-validation is avoided, the accuracy of determining the model parameters is low, and the evaluation effect of the target evaluation model is affected, so that the stability and accuracy of credit evaluation are improved, and the credit evaluation result realizes effective description of the credit of the object to be borrowed. Further, the borrowing risk of borrowing the object to be borrowed is determined according to the credit evaluation result, accurate and rapid determination of the borrowing risk is achieved, and the problem that evaluation accuracy is poor due to strong subjectivity in manual evaluation is avoided.
Therefore, the scheme provided by the application achieves the purpose of obtaining the target evaluation model through multiple times of cross verification, so that the borrowing risk of the object to be borrowed is determined according to the target evaluation model, the technical effect of accurately determining the borrowing risk is realized, and the technical problem that the borrowing risk cannot be accurately determined due to low evaluation accuracy of the credit of the user when the borrowing risk is evaluated according to the credit of the borrowing user in the prior art is solved.
In an alternative embodiment, fig. 2 is a schematic diagram of an alternative generation of a target assessment model according to an embodiment of the present application, and as shown in fig. 2, the risk assessment system may obtain the target assessment model by:
step S201, constructing N first evaluation models according to N groups of first model parameters and an initial model structure, wherein the N groups of first model parameters are parameters obtained by adjusting the same group of model parameters, and N is a positive integer greater than 1.
In this embodiment, the parameter types of the model parameters are not limited to the following types:
learning rate: learning rate, controlling step length of each update;
max_depth: the maximum depth of the tree, controlling the complexity of the tree;
min_child_weight: the minimum weight of the leaf nodes controls the number of the leaf nodes;
subsamples: the proportion of sampling during each iteration controls the overfitting;
colsample_byte: the number of columns randomly sampled by each tree, and controlling the overfitting;
num_boost_round: the iteration times control the complexity of the model.
Optionally, in the process of determining the target evaluation model, a set of basic parameters may be determined first, where the basic parameters may be preset manually or may be obtained through training. And then, the set of basic parameters can be adjusted for N times, so that N sets of first model parameters are obtained, and N first evaluation models can be constructed according to the N sets of first model parameters. Wherein a set of first model parameters corresponds to a first evaluation model.
Step S202, for each first evaluation model, performing M times of cross validation on the first evaluation model, and determining the performance score of the first evaluation model according to the results of the M times of cross validation, wherein M is a positive integer greater than 1.
Alternatively, M cross-validations may be performed for each first evaluation model. Each cross-validation obtains a sub-performance score matched with the first evaluation model, so that the performance score of the first evaluation model can be determined according to the sub-performance scores corresponding to the M times of cross-validation.
Step S203, a model with highest performance score is screened out from the N first evaluation models to obtain a third evaluation model, and a training sample set is adopted to train the third evaluation model to obtain a target evaluation model, wherein the training sample set consists of sample target information of a sample object and a real credit label of the sample object.
Optionally, after determining the performance score of each first evaluation model, a training sample set may be used to train the first evaluation model with the highest performance score, so as to obtain the target evaluation model. The training sample set is composed of sample target information of a sample object and real credit labels of the sample object, wherein the sample target information is obtained by carrying out data preprocessing on sample initial information of the sample object, and the sample initial information at least comprises social relation information of the sample object and can also comprise user information, history borrowing information, credit information and the like of the sample object. The real credit label of the sample object characterizes the credit rating of the sample object, e.g., the real credit label is the first credit rating or the second credit rating. The method for training XGBoost model in the prior art may be used to train the first evaluation model with the highest performance score, so that the description is omitted here.
It should be noted that, by evaluating the first evaluation model constructed based on different model parameters by using multiple cross-verifications, accurate determination of the model parameters to be used is achieved, so that the evaluation accuracy of the target evaluation model obtained based on model parameter training to be used is effectively improved.
In an alternative embodiment, before constructing N first evaluation models according to N sets of first model parameters and the initial model structure, the risk evaluation system may construct an evaluation model according to preset model parameters, and train the evaluation model by using a training sample set to obtain an initial evaluation model, so as to adjust model parameters in the initial evaluation model N times according to a parameter adjustment rule, to obtain N sets of first model parameters.
Alternatively, the risk assessment system may construct an assessment model according to preset model parameters and the initial model structure described above. The parameter adjustment rule may be preset, and the parameter adjustment rule is used to indicate the parameter that is allowed to be adjusted and the adjustment range corresponding to the parameter, and each random adjustment in the N times of random adjustment may adjust only one model parameter, or may adjust multiple model parameters.
It should be noted that, by performing parameter adjustment on the model parameters obtained by training, N groups of first model parameters are obtained, so that the initial quality of the first model parameters is effectively ensured, and the evaluation of the target evaluation model determined later is more accurate.
In an alternative embodiment, in the process of performing M times of cross-validation on the first evaluation model and determining the performance score of the first evaluation model according to the result of M times of cross-validation, the risk evaluation system may train the first evaluation model through M times of cross-validation based on the training sample set to obtain P second evaluation models, and determine the performance score of each second evaluation model, so as to determine the performance score of the first evaluation model according to the performance scores of the P second evaluation models. Wherein each cross-validation divides the training sample set into k subsets, and each division results in a different subset, M, P, k is a positive integer greater than 1, and p=m×k.
When the first evaluation model is trained through single k-fold cross validation, k second evaluation models can be obtained, and when the first evaluation model is trained through M k-fold cross validation, m×k=p second evaluation models are obtained.
In the k-fold cross verification process, the two stages of training and verification are divided, so that the verification result of the second evaluation model in the k-fold cross verification can be used as the performance score of the second evaluation model, the performance score in the same k-fold cross verification is averaged to obtain the performance score corresponding to the k-fold cross verification, and the sub-performance score corresponding to the first evaluation model is obtained. And then, averaging the performance scores corresponding to the M times of k-fold cross validation, thereby obtaining the performance score of the first evaluation model.
It should be noted that by evaluating the performance score of the first evaluation model based on the second evaluation model trained from the first evaluation model, an accurate determination of the performance score of the first evaluation model is achieved.
In an alternative embodiment, in the process of training the first evaluation model through M times of cross validation based on the training sample set to obtain P second evaluation models and determining the performance score of each second evaluation model, the risk evaluation system may divide the training sample set into k subsets for each cross validation, and then perform k times of training on the first evaluation model according to the k subsets to obtain k second evaluation models, so that for each second evaluation model, the target subset is input to the second evaluation model to obtain a credit evaluation result output by the second evaluation model, and evaluate the performance score of the second evaluation model according to the credit evaluation result output by the second evaluation model. Wherein each of the k exercises uses k-1 subsets of the k subsets, and the k-1 subsets used for each exercise are not exactly the same, and the target subset is a subset that is not used when the second evaluation model is trained in the k subsets.
Optionally, in a single k-fold cross validation, the training sample set is divided into k non-overlapping subsets, one subset is used as the validation set (i.e., the target subset described above) at a time, and the remaining k-1 subsets are used as the training set for training, thereby performing k times of training. And after the training is carried out for k times, obtaining k second evaluation models, inputting a verification set which is not used for training into the second evaluation models for each second evaluation model, and obtaining credit evaluation results output by the second evaluation models, thereby determining the performance scores of the second evaluation models according to the credit evaluation results and the real credit labels corresponding to the samples in the verification set.
It should be noted that, by performing k times of training on the first evaluation model by using k-fold cross validation, k second evaluation models are obtained, so that the performance score of the second evaluation model can effectively reflect the performance score of the first evaluation model.
In an alternative embodiment, in the process of training the third evaluation model by using the training sample set to obtain the target evaluation model, the risk evaluation system may input the training sample set to the third evaluation model to obtain a sample credit evaluation result output by the third evaluation model, and then obtain a weight value matched with each real credit label, so as to calculate a loss function value according to the sample credit evaluation result, the real credit label and the weight value, and optimize model parameters of the third evaluation model according to the loss function value until the loss function value is smaller than or equal to a preset value, so as to obtain the target evaluation model. Wherein different real credit tags match different weight values.
Optionally, in this embodiment, the method of training the XGBoost model in the prior art is used to train the third evaluation model, and only the loss function used to train the XGBoost model and the method of calculating the loss function value are optimized.
Specifically, a loss function used to train the XGBoost model is described. In this embodiment, the third evaluation model is set as a two-class model, and therefore, a logarithmic loss is selected as an initial loss function, and the logarithmic loss is used to measure the difference between the model prediction probability value and the actual class, and a smaller loss value indicates a better model prediction effect. The mathematical expression for the log loss is as follows:
L(y,p)＝-[y*log(p)+(1-y)*log(1-p)]
where L (y, p) represents the loss function value, y represents the true credit label, which may be 0 or 1,0 may represent the second credit level, 1 may represent the first credit level, and p represents the sample credit evaluation result.
Optionally, after determining the initial loss function, a regularization term, such as L1 regularization or L2 regularization, may be added to the initial loss function to obtain an updated loss function, so as to avoid over-fitting of the model.
Wherein, L1 regularization can generate thin fluffs by adding the sum of absolute values of all model parameters in an initial loss function as a penalty term, namely, partial model parameters are made to be 0, which helps to reduce model complexity and reduce overfitting. The form of L1 regularization is as follows:
L 1 (ω)＝L0(ω)+λΣ|ω i |
Wherein L is 1 (ω) represents the updated loss function, L0 (ω) represents the initial loss function, ω represents the model parameters of the third evaluation model, ω i And (3) representing an ith model parameter of the third evaluation model, wherein lambda is a regularization coefficient and is used for controlling the weight of the regularization term.
Alternatively, L2 regularization is performed by adding the sum of squares of the individual model parameters to the initial loss function as a penalty term. L2 regularization enables values of model parameters to be close to 0 but not 0, which helps smooth the model, reducing overfitting. The form of L2 regularization is as follows:
L 2 (ω)＝L0(ω)+λ*Σω i 2
wherein L is 2 And (ω) represents the updated loss function.
Further, for the two classification problems, because the number of positive and negative samples in the training sample set may be different greatly, the model tends to be more classified, and the prediction performance of few classes is reduced, so different weights can be set for the positive and negative samples when the loss function value is calculated according to the updated loss function, so that the model is more focused on the few classes. For example, in this embodiment, the sample with the real credit label being the first credit level is a negative sample, the sample with the real credit label being the second credit level is a positive sample, and the weight value of the real credit label matching is preset, which may be determined according to the ratio of the number of samples of the positive and negative samples, for example, if the ratio of the number of samples corresponding to the first credit level to the second credit level is 1: and 3, the weight value of the first credit level matching is 3/4, and the weight value of the second credit level matching is 1/4. Optionally, after obtaining the weight value matched with each real credit tag, the risk assessment system may assign a corresponding weight to each sample when calculating the loss function value according to the updated loss function, so as to obtain a more accurate loss function value. And further, a training process for the third evaluation model can be determined according to the loss function value, so that a target evaluation model is obtained.
Further, after model training is completed, i.e., after the target evaluation model is obtained, the target evaluation model may be evaluated using the test data set to learn about the performance of the target evaluation model. Optionally, in the testing process, the performance of the target evaluation model can be evaluated through indexes such as accuracy, recall rate and the like. The accuracy rate refers to the proportion of the number of correctly classified samples to the total number of samples, and the recall rate refers to the proportion of the number of correctly classified positive samples to all positive samples.
It should be noted that, by calculating the loss function value in combination with the weight value matched with the real credit label in the training process, the model can pay more attention to a few types of samples, so that the training effect of the model is improved, and the evaluation accuracy of the target evaluation model is further improved.
In an alternative embodiment, the risk assessment system may generate the training sample set by: acquiring initial sample target information of a sample object and a real credit label; determining the number of objects of the sample objects matched with each real credit label, and determining the real credit label with the least number of matched objects as a target real credit label; generating synthetic sample target information according to initial sample target information matched with the target real credit label; and taking the initial sample target information or the synthesized sample target information as sample target information, and forming a training sample set by the sample target information and the real credit label.
Alternatively, the number of positive and negative samples may be balanced, since the number of positive and negative samples in the training sample set may be significantly different, which may affect the prediction accuracy and generalization ability of the model. Specifically, initial sample target information and a real credit label of a sample object may be acquired first, where the initial sample target information is real information corresponding to the sample object. And then determining the object number of the sample objects matched with each real credit label according to the matching relation between the sample objects and the real credit labels, so that the real credit label with the least matched object number is determined as a target real credit label, and the initial sample target information matched with the target real credit label is determined as a minority sample, thereby realizing the determination of the minority sample.
Optionally, after determining the minority class samples, the data imbalance problem may be handled by an SMOTE (oversampling technique for synthesizing minority class samples, synthetic Minority Over-sampling Technique), which is an oversampling method based on generating synthesized samples, which generates new samples by interpolating between minority class samples (i.e. the aforementioned initial sample target information for true credit tag matching), so as to achieve the purpose of increasing the number of minority class samples and making the data set more balanced.
Specifically, the SMOTE method is as follows:
step 1: for each minority class sample, K samples nearest to each minority class sample are selected, wherein K is a super parameter and is usually 5;
step 2: randomly selecting one sample from K nearest neighbor samples, and calculating the difference value between the value of each information in a minority sample and the value of each information in the sample;
step 3: multiplying the difference value by a random number r (r is in the range of 0, 1), and adding the value of each information in the minority class of samples to generate a new synthesized sample, namely synthesized sample target information;
step 4: repeating step 2-3 to generate a specified number of new synthesized samples.
Further, the real initial sample target information or the generated synthesized sample target information may be used as sample target information, so that a training sample set is formed by the sample target information and the real credit label.
It should be noted that, by using the SMOTE method to generate the synthesized sample target information, the number of minority samples can be increased, so that the data set is more balanced, and the prediction accuracy of the model is improved. Meanwhile, because the generated synthesized sample target information is obtained by interpolation among a few types of samples, the condition of over fitting does not occur.
Therefore, the scheme provided by the application achieves the purpose of obtaining the target evaluation model through multiple times of cross verification, so that the borrowing risk of the object to be borrowed is determined according to the target evaluation model, the technical effect of accurately determining the borrowing risk is realized, and the technical problem that the borrowing risk cannot be accurately determined due to low evaluation accuracy of the credit of the user when the borrowing risk is evaluated according to the credit of the borrowing user in the prior art is solved.
Example 2
According to an embodiment of the present application, there is provided an embodiment of a risk assessment apparatus, wherein fig. 3 is a schematic diagram of an alternative risk assessment apparatus according to an embodiment of the present application, as shown in fig. 3, and the apparatus includes:
the first obtaining module 301 is configured to obtain target information of an object to be borrowed, where the target information at least includes: information characterizing social relationships of objects to be borrowed;
the evaluation module 302 is configured to evaluate the target information through a target evaluation model to obtain a credit evaluation result, where model parameters of the target evaluation model are determined through multiple cross-validation;
the first determining module 303 is configured to determine a credit rating of the object to be borrowed according to the credit evaluation result, and determine a borrowing risk of borrowing the object to be borrowed according to the credit rating of the object to be borrowed.
It should be noted that the first obtaining module 301, the evaluating module 302, and the first determining module 303 correspond to steps S101 to S103 in the above embodiment, and the three modules are the same as examples and application scenarios implemented by the corresponding steps, but are not limited to those disclosed in the above embodiment 1.
Optionally, the risk assessment device further includes: the construction module is used for constructing N first evaluation models according to N groups of first model parameters and an initial model structure, wherein the N groups of first model parameters are parameters obtained by adjusting the same group of model parameters, and N is a positive integer greater than 1; the second determining module is used for carrying out M times of cross validation on each first evaluating model, and determining the performance score of the first evaluating model according to the results of the M times of cross validation, wherein M is a positive integer greater than 1; the first training module is used for screening out the model with the highest performance score from the N first evaluation models to obtain a third evaluation model, and training the third evaluation model by adopting a training sample set to obtain a target evaluation model, wherein the training sample set consists of sample target information of a sample object and a real credit label of the sample object.
Optionally, the risk assessment device further includes: the second training module is used for constructing an evaluation model according to preset model parameters, and training the evaluation model by adopting a training sample set to obtain an initial evaluation model; and the adjusting module is used for adjusting the model parameters in the initial evaluation model for N times according to the parameter adjusting rule to obtain N groups of first model parameters.
Optionally, the second determining module further includes: the training sub-module is used for training the first evaluation model through M times of cross validation based on the training sample set to obtain P second evaluation models, and determining the performance score of each second evaluation model, wherein each time of cross validation divides the training sample set into k subsets, the subsets obtained by each time of division are different, M, P, k is a positive integer greater than 1, and P=M×k; and the determining submodule is used for determining the performance scores of the first evaluation model according to the performance scores of the P second evaluation models.
Optionally, the training sub-module further includes: a dividing unit for dividing the training sample set into k subsets for each cross-validation; the training unit is used for carrying out k times of training on the first evaluation model according to k subsets to obtain k second evaluation models, wherein each time of training in the k times of training adopts k-1 subsets in the k subsets, and the k-1 subsets used in each time of training are not identical; and the evaluation unit is used for inputting the target subset into the second evaluation model for each second evaluation model to obtain a credit evaluation result output by the second evaluation model, and evaluating the performance score of the second evaluation model according to the credit evaluation result output by the second evaluation model, wherein the target subset is a subset which is not used when the second evaluation model is trained in k subsets.
Optionally, the first training module further includes: the first processing sub-module is used for inputting the training sample set into the third evaluation model to obtain a sample credit evaluation result output by the third evaluation model; the acquisition sub-module is used for acquiring a weight value matched with each real credit label, wherein different real credit labels are matched with different weight values; and the second processing sub-module is used for calculating a loss function value according to the sample credit evaluation result, the real credit label and the weight value, optimizing the model parameters of the third evaluation model according to the loss function value until the loss function value is smaller than or equal to a preset value, and obtaining a target evaluation model.
Optionally, the risk assessment device further includes: the second acquisition module is used for acquiring initial sample target information of the sample object and a real credit label; a third determining module, configured to determine the number of objects of the sample objects matched with each real credit tag, and determine the real credit tag with the least number of matched objects as the target real credit tag; the generating module is used for generating synthetic sample target information according to the initial sample target information matched with the target real credit label; and the processing module is used for taking the initial sample target information or the synthesized sample target information as sample target information, and forming a training sample set by the sample target information and the real credit label.
Example 3
According to another aspect of the embodiments of the present application, there is also provided a computer readable storage medium having a computer program stored therein, wherein the computer program is configured to perform the risk assessment method described above when run.
Example 4
According to another aspect of the embodiments of the present application, there is also provided an electronic device, wherein fig. 4 is a schematic diagram of an alternative electronic device according to an embodiment of the present application, as shown in fig. 4, the electronic device including one or more processors; and a memory for storing one or more programs that, when executed by the one or more processors, cause the one or more processors to implement a method for running the program, wherein the program is configured to perform the risk assessment method described above when run.
The foregoing embodiment numbers of the present application are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
In the foregoing embodiments of the present application, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed technology may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and the division of units may be a logic function division, and there may be another division manner in actual implementation, for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in essence or a part contributing to the prior art or all or part of the technical solution in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method of the various embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.
Claims (10)
1. A risk assessment method, comprising:
obtaining target information of an object to be borrowed, wherein the target information at least comprises: information characterizing social relationships of the objects to be borrowed;
evaluating the target information through a target evaluation model to obtain a credit evaluation result, wherein model parameters of the target evaluation model are determined through multiple times of cross verification;
and determining the credit rating of the object to be borrowed according to the credit evaluation result, and determining the borrowing risk of borrowing the object to be borrowed according to the credit rating of the object to be borrowed.
2. The method according to claim 1, wherein the target assessment model is obtained by:
constructing N first evaluation models according to N groups of first model parameters and an initial model structure, wherein the N groups of first model parameters are parameters obtained by adjusting the same group of model parameters, and N is a positive integer greater than 1;
For each first evaluation model, performing M times of cross validation on the first evaluation model, and determining the performance score of the first evaluation model according to the M times of cross validation results, wherein M is a positive integer greater than 1;
and screening out the model with the highest performance score from the N first evaluation models to obtain a third evaluation model, and training the third evaluation model by adopting a training sample set to obtain the target evaluation model, wherein the training sample set consists of sample target information of a sample object and a real credit label of the sample object.
3. The method according to claim 2, wherein before constructing N first evaluation models from N sets of first model parameters and an initial model structure, the method comprises:
constructing an evaluation model according to preset model parameters, and training the evaluation model by adopting the training sample set to obtain an initial evaluation model;
and according to a parameter adjustment rule, adjusting the model parameters in the initial evaluation model for N times to obtain the N groups of first model parameters.
4. The method of claim 2, wherein M times of cross-validation are performed on the first evaluation model, and determining the performance score of the first evaluation model based on the M times of cross-validation results comprises:
Training the first evaluation model through the M times of cross validation based on the training sample set to obtain P second evaluation models, and determining the performance score of each second evaluation model, wherein each time of cross validation divides the training sample set into k subsets, the subsets obtained by each time of cross validation are different, M, P, k is a positive integer greater than 1, and P=M×k;
and determining the performance scores of the first evaluation model according to the performance scores of the P second evaluation models.
5. The method of claim 4, wherein training the first evaluation model through the M times of cross-validation based on the training sample set results in P second evaluation models and determines a performance score for each second evaluation model, comprising:
for each cross-validation, dividing the training sample set into the k subsets;
performing k times of training on the first evaluation model according to the k subsets to obtain k second evaluation models, wherein each time of training in the k times of training adopts k-1 subsets in the k subsets, and the k-1 subsets used in each time of training are not completely identical;
and for each second evaluation model, inputting a target subset into the second evaluation model to obtain a credit evaluation result output by the second evaluation model, and evaluating the performance score of the second evaluation model according to the credit evaluation result output by the second evaluation model, wherein the target subset is a subset which is not used when training the second evaluation model in the k subsets.
6. The method of claim 2, wherein training the third evaluation model with a training sample set to obtain the target evaluation model comprises:
inputting the training sample set into the third evaluation model to obtain a sample credit evaluation result output by the third evaluation model;
acquiring a weight value matched with each real credit label, wherein different real credit labels are matched with different weight values;
and calculating a loss function value according to the sample credit evaluation result, the real credit label and the weight value, and optimizing model parameters of the third evaluation model according to the loss function value until the loss function value is smaller than or equal to a preset value, so as to obtain the target evaluation model.
7. The method of claim 2, wherein the training sample set is generated by:
acquiring initial sample target information of a sample object and the real credit label;
determining the number of objects of the sample objects matched with each real credit label, and determining the real credit label with the least number of matched objects as a target real credit label;
generating synthetic sample target information according to the initial sample target information matched with the target real credit label;
And taking the initial sample target information or the synthesized sample target information as the sample target information, and forming the training sample set by the sample target information and the real credit label.
8. A risk assessment apparatus, comprising:
the first obtaining module is configured to obtain target information of an object to be borrowed, where the target information at least includes: information characterizing social relationships of the objects to be borrowed;
the evaluation module is used for evaluating the target information through a target evaluation model to obtain a credit evaluation result, wherein model parameters of the target evaluation model are determined through multiple times of cross verification;
and the first determining module is used for determining the credit grade of the object to be borrowed according to the credit evaluation result and determining the borrowing risk of borrowing the object to be borrowed according to the credit grade of the object to be borrowed.
9. A computer readable storage medium, characterized in that the computer readable storage medium has stored therein a computer program, wherein the computer program is arranged to execute the risk assessment method according to any of the claims 1 to 7 at run-time.
10. An electronic device, the electronic device comprising one or more processors; a memory for storing one or more programs that, when executed by the one or more processors, cause the one or more processors to implement a method for running a program, wherein the program is configured to perform the risk assessment method of any of claims 1 to 7 when run.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310614843.XA CN116664274A (en) | 2023-05-26 | 2023-05-26 | Risk assessment method and device, storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310614843.XA CN116664274A (en) | 2023-05-26 | 2023-05-26 | Risk assessment method and device, storage medium and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116664274A true CN116664274A (en) | 2023-08-29 |
Family
ID=87713088
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310614843.XA Pending CN116664274A (en) | 2023-05-26 | 2023-05-26 | Risk assessment method and device, storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116664274A (en) |
-
2023
- 2023-05-26 CN CN202310614843.XA patent/CN116664274A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109389494B (en) | Loan fraud detection model training method, loan fraud detection method and device | |
| Breinlich et al. | Machine learning in international trade research-evaluating the impact of trade agreements | |
| CN110866832A (en) | Risk control method, system, storage medium and computing device | |
| CN112256971A (en) | Sequence recommendation method and computer-readable storage medium | |
| CN116450951A (en) | Service recommendation method and device, storage medium and electronic equipment | |
| CN113256335B (en) | Data screening method, multimedia data delivery effect prediction method and device | |
| CN112434862B (en) | Method and device for predicting financial dilemma of marketing enterprises | |
| CN112508684B (en) | Collecting-accelerating risk rating method and system based on joint convolutional neural network | |
| CN112308623A (en) | High-quality client loss prediction method and device based on supervised learning and storage medium | |
| CN113591892A (en) | Training data processing method and device | |
| CN116664274A (en) | Risk assessment method and device, storage medium and electronic equipment | |
| CN114820160A (en) | Loan interest rate estimation method, device, equipment and readable storage medium | |
| CN113435713B (en) | Risk map compiling method and system based on GIS technology and two-model fusion | |
| CN113256404A (en) | Data processing method and device | |
| CN113095392A (en) | Abnormal service identification method and device | |
| CN117236900B (en) | Individual tax data processing method and system based on flow automation | |
| CN113052689B (en) | Product recommendation method and device based on decision tree | |
| CN117150321B (en) | Equipment trust evaluation method and device, service equipment and storage medium | |
| CN112862602B (en) | User request determining method, storage medium and electronic device | |
| CN115565611B (en) | Biological regression prediction method, device, equipment and storage medium | |
| CN117592859A (en) | LSTM-based customer financial risk assessment method, device and equipment | |
| Draghi et al. | Identifying and handling data bias within primary healthcare data using synthetic data generators | |
| CN118158276A (en) | Service matching method, device and equipment based on user analysis | |
| CN116976906A (en) | Transaction risk assessment method, device, electronic equipment and medium | |
| CN113554507A (en) | Personal credit evaluation method and system based on back propagation neural network |
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 |