CN115018627A - Credit risk evaluation method and device, storage medium and electronic equipment - Google Patents
Credit risk evaluation method and device, storage medium and electronic equipment Download PDFInfo
- Publication number
- CN115018627A CN115018627A CN202210606396.9A CN202210606396A CN115018627A CN 115018627 A CN115018627 A CN 115018627A CN 202210606396 A CN202210606396 A CN 202210606396A CN 115018627 A CN115018627 A CN 115018627A
- Authority
- CN
- China
- Prior art keywords
- credit
- risk evaluation
- credit risk
- user
- evaluation model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention provides a credit risk evaluation method and device, a storage medium and electronic equipment, which can be applied to the financial field or other fields, wherein the method comprises the following steps: responding to a credit risk evaluation instruction, and determining a user corresponding to the credit risk evaluation instruction; acquiring credit behavior data of the user; performing feature extraction on credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model to obtain first credit behavior features of the credit behavior data; performing feature extraction on the credit behavior data of the user through a second feature extraction module in the credit risk evaluation model to obtain a second credit behavior feature of the credit behavior data; obtaining, by an output module in the credit risk evaluation model, a credit risk evaluation result of the user based on the first credit behavior feature and the second credit behavior feature. The evaluation accuracy and the evaluation efficiency of the credit risk can be improved.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a credit risk evaluation method and device, a storage medium and electronic equipment.
Background
Currently, with the development of economic society, the internet financial products of banks are developing at a high speed. While providing credit service on the internet, banks need to evaluate the credit risk of information of credit customers so as to reduce the default risk of customers.
The existing credit risk evaluation mode is characterized in that the evaluation is carried out through the characteristics of an expert design applicant, the evaluation subjectivity is strong, batch duplication of experience cannot be achieved, and the workload of model development is large, so that the accuracy and the efficiency of credit risk evaluation are low.
Disclosure of Invention
The invention aims to provide a credit risk evaluation method which can improve the evaluation accuracy and the evaluation efficiency of credit risk.
The invention also provides a credit risk evaluation device which is used for ensuring the realization and the application of the method in practice.
A credit risk assessment method, comprising:
responding to a credit risk evaluation instruction, and determining a user corresponding to the credit risk evaluation instruction;
acquiring credit behavior data of the user;
performing feature extraction on credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model to obtain first credit behavior features of the credit behavior data;
performing feature extraction on the credit behavior data of the user through a second feature extraction module in the credit risk evaluation model to obtain a second credit behavior feature of the credit behavior data;
and obtaining a credit risk evaluation result of the user based on the first credit behavior characteristic and the second credit behavior characteristic through an output module in the credit risk evaluation model.
Optionally, the obtaining, by an output module in the credit risk assessment model, a credit risk assessment result of the user based on the first credit behavior feature and the second credit behavior feature includes:
performing feature fusion on the first credit behavior feature and the second credit behavior feature through a feature fusion layer of an output module in the credit risk evaluation model to obtain fusion features;
determining a probability value between the user and each preset candidate tag based on the fusion characteristics through an output layer of an output module in the credit risk evaluation model; and taking the candidate label to which the target probability value with the maximum value in the probability values belongs as the credit risk evaluation result of the user.
The above method, optionally, is a process of constructing the credit risk assessment model, including:
acquiring a training data set and an initial credit risk evaluation model; the training data set comprises a plurality of training samples, and each training sample comprises historical credit behavior data of labeled sample labels of a historical user; the specimen label represents a credit risk evaluation result of the historical user;
training the initial credit risk evaluation model by using each training sample in the training data set until the initial credit risk evaluation model meets a preset training condition;
and determining the initial credit risk evaluation model meeting the training conditions as a credit risk evaluation model.
Optionally, in the method, the feature extraction is performed on the credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model to obtain a first credit behavior feature of the credit behavior data, and the method includes:
performing feature extraction on the credit behavior data through a convolutional neural network in a first feature extraction module of the credit risk evaluation model to obtain a first credit behavior initial feature;
and processing the first credit behavior initial feature through a multi-head attention mechanism layer in a first feature extraction module of the credit risk evaluation model to obtain a first credit behavior feature of the credit behavior data.
The method described above, optionally, after obtaining the credit risk evaluation result of the user, further includes:
and outputting credit risk warning information aiming at the user under the condition that the credit risk evaluation result of the user meets a preset credit risk warning condition.
A credit risk assessment device, comprising:
the determining unit is used for responding to a credit risk evaluation instruction and determining a user corresponding to the credit risk evaluation instruction;
the acquisition unit is used for acquiring credit behavior data of the user;
the first feature extraction unit is used for performing feature extraction on credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model to obtain first credit behavior features of the credit behavior data;
the second feature extraction unit is used for performing feature extraction on the credit behavior data of the user through a second feature extraction module in the credit risk evaluation model to obtain a second credit behavior feature of the credit behavior data;
and the output unit is used for obtaining a credit risk evaluation result of the user based on the first credit behavior characteristic and the second credit behavior characteristic through an output module in the credit risk evaluation model.
The above apparatus, optionally, the output unit includes:
the characteristic fusion subunit is used for performing characteristic fusion on the first credit behavior characteristic and the second credit behavior characteristic through a characteristic fusion layer of an output module in the credit risk evaluation model to obtain a fusion characteristic;
the output subunit is used for determining a probability value between the user and each preset candidate label based on the fusion feature through an output layer of an output module in the credit risk evaluation model; and taking the candidate label to which the target probability value with the maximum value in the probability values belongs as the credit risk evaluation result of the user.
The above apparatus, optionally, further comprises: a model training unit; the model training unit is used for:
acquiring a training data set and an initial credit risk evaluation model; the training data set comprises a plurality of training samples, and each training sample comprises historical credit behavior data of labeled sample labels of a historical user; the specimen label represents a credit risk evaluation result of the historical user;
sequentially training the initial credit risk evaluation model by utilizing each training sample in the training data set until the initial credit risk evaluation model meets a preset training condition;
and determining the initial credit risk evaluation model meeting the training conditions as a credit risk evaluation model.
The above apparatus, optionally, the first feature extraction unit is configured to:
performing feature extraction on the credit behavior data through a convolutional neural network in a first feature extraction module of the credit risk evaluation model to obtain a first credit behavior initial feature;
and processing the first credit behavior initial characteristic through a multi-head attention mechanism layer in a first characteristic extraction module of the credit risk evaluation model to obtain a first credit behavior characteristic of the credit behavior data.
The above apparatus, optionally, the credit risk assessment apparatus, further includes: an alarm unit;
and the alarm unit is used for outputting credit risk alarm information aiming at the user under the condition that the credit risk evaluation result of the user meets a preset credit risk alarm condition.
A storage medium comprising stored instructions, wherein the instructions, when executed, control a device on which the storage medium is located to perform a credit risk assessment method as described above.
An electronic device comprising a memory, and one or more instructions, wherein the one or more instructions are stored in the memory and configured to be executed by one or more processors to perform the credit risk assessment method as described above.
Compared with the prior art, the invention has the following advantages:
the invention provides a credit risk evaluation method and device, a storage medium and electronic equipment, wherein the method comprises the following steps: responding to a credit risk evaluation instruction, and determining a user corresponding to the credit risk evaluation instruction; acquiring credit behavior data of the user; performing feature extraction on credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model to obtain first credit behavior features of the credit behavior data; performing feature extraction on credit behavior data of the user through a second feature extraction module in the credit risk evaluation model to obtain second credit behavior features of the credit behavior data; and obtaining a credit risk evaluation result of the user based on the first credit behavior characteristic and the second credit behavior characteristic through an output module in the credit risk evaluation model. By applying the method provided by the embodiment of the invention, the credit risk evaluation of the user is carried out by extracting the characteristics of different levels of the credit behavior data, so that the evaluation accuracy and the evaluation efficiency of the credit risk can be improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a flowchart of a credit risk assessment method according to the present invention;
FIG. 2 is a flowchart of a process for obtaining a credit risk assessment result of the user according to the present invention;
FIG. 3 is a flow chart of a process for constructing a credit risk assessment model according to the present invention;
FIG. 4 is an exemplary diagram of a process for obtaining a first credit behavior feature of credit behavior data provided by the invention;
FIG. 5 is a schematic structural diagram of a credit risk assessment model according to the present invention;
FIG. 6 is a schematic structural diagram of a credit risk assessment apparatus according to the present invention;
fig. 7 is a schematic structural diagram of an electronic device according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In this application, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The embodiment of the invention provides a credit risk evaluation method, which can be applied to electronic equipment, wherein the method has a flow chart as shown in fig. 1, and specifically comprises the following steps:
s101: and responding to the credit risk evaluation instruction, and determining a user corresponding to the credit risk evaluation instruction.
In this embodiment, the credit risk assessment instruction may be an instruction triggered when credit risk assessment needs to be performed on the user, for example, the instruction may be triggered when the user transacts a loan service, a credit lease service, or the like, or may be automatically triggered when the user satisfies a preset credit assessment condition.
S102: and acquiring credit behavior data of the user.
In this embodiment, new user behavior data of the user may be acquired in a preset database based on the user identifier, or credit behavior data of the user may be acquired from instruction information of the credit risk evaluation instruction.
Optionally, the credit behavior data may include performance behavior information of the user, transaction behavior information, and other user behavior information that affects credit.
S103: and performing feature extraction on the credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model to obtain a first credit behavior feature of the credit behavior data.
In this embodiment, the first feature extraction module may extract a local feature that is important for the credit risk evaluation of the user, where the local feature is the first credit behavior feature and may be obtained by performing weighting processing on a global feature of the credit behavior data.
S104: and performing feature extraction on the credit behavior data of the user through a second feature extraction module in the credit risk evaluation model to obtain a second credit behavior feature of the credit behavior data.
In this embodiment, the second credit behaviour signature may be a global signature of the credit behaviour data.
S105: obtaining, by an output module in the credit risk evaluation model, a credit risk evaluation result of the user based on the first credit behavior feature and the second credit behavior feature.
In this embodiment, the output module may fuse the first credit behavior feature and the second credit behavior feature to obtain a fused feature, and obtain a credit risk evaluation result of the user according to the fused feature.
Alternatively, the credit risk assessment result may be a credit risk rating, or a credit risk score.
The embodiment of the invention provides a credit risk evaluation method, which comprises the following steps: responding to a credit risk evaluation instruction, and determining a user corresponding to the credit risk evaluation instruction; acquiring credit behavior data of the user; performing feature extraction on credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model to obtain first credit behavior features of the credit behavior data; performing feature extraction on credit behavior data of the user through a second feature extraction module in the credit risk evaluation model to obtain second credit behavior features of the credit behavior data; and obtaining a credit risk evaluation result of the user based on the first credit behavior characteristic and the second credit behavior characteristic through an output module in the credit risk evaluation model. By applying the method provided by the embodiment of the invention, the credit risk evaluation of the user is carried out by extracting the characteristics of different levels of the credit behavior data, so that the evaluation accuracy and the evaluation efficiency of the credit risk can be improved.
In an embodiment provided by the present invention, based on the foregoing implementation process, optionally, the obtaining, by an output module in the credit risk assessment model, a credit risk assessment result of the user based on the first credit behavior feature and the second credit behavior feature, as shown in fig. 2, includes:
s201: and performing feature fusion on the first credit behavior feature and the second credit behavior feature through a feature fusion layer of an output module in the credit risk evaluation model to obtain a fusion feature.
In this embodiment, the first credit behavior feature and the second credit behavior feature may be spliced to obtain a fusion feature, the splicing manner may be that the first credit behavior feature is spliced after the second credit behavior feature, or the second credit behavior feature is spliced after the first credit behavior feature, and the fusion feature is obtained by fusing the first credit behavior feature and the second credit behavior feature, so that a feature having an important influence on credit evaluation can be obtained while loss of global information is also reduced.
S202: determining a probability value between the user and each preset candidate tag based on the fusion characteristics through an output layer of an output module in the credit risk evaluation model; and taking the candidate label to which the target probability value with the maximum value in the probability values belongs as the credit risk evaluation result of the user.
In this embodiment, the candidate tags may represent risk classifications of the user, the risk classifications of the user represented by different candidate tags are different, the output layer may determine, according to the fusion feature, a probability value that the user belongs to each candidate tag, and determine, according to the probability value that the user belongs to each candidate tag, a credit risk evaluation result of the user, which may specifically be the candidate tag with the highest probability value as the credit risk evaluation result of the user.
In an embodiment provided by the present invention, based on the implementation process, optionally, the process of constructing the credit risk evaluation model is as shown in fig. 3, and includes:
s301: acquiring a training data set and an initial credit risk evaluation model; the training data set comprises a plurality of training samples, and each training sample comprises historical credit behavior data of a historical user and a sample label; the specimen label represents the credit risk evaluation result of the historical user.
In this embodiment, the sample label, that is, the candidate label, may label the credit risk evaluation result of the user history.
S302: and training the initial credit risk evaluation model by using each training sample in the training data set until the initial credit risk evaluation model meets a preset training condition.
In this embodiment, a residual error module may be provided for the first feature extraction module of the initial credit risk evaluation model, the initial credit risk evaluation model provided with the residual error module is trained through training samples in training data, and specifically, each training sample may be input into the initial credit risk evaluation model to obtain an evaluation result corresponding to the training sample output by the initial credit risk evaluation model; calculating a loss function value according to the evaluation result and the sample label of the training sample; and adjusting the network parameters of the initial credit risk evaluation model according to the loss function values.
Optionally, the training condition may be that the training frequency of the initial credit risk evaluation model is greater than a preset training frequency threshold, or that the prediction accuracy of the initial credit risk evaluation model is greater than a preset accuracy threshold, or that the loss function of the initial credit risk evaluation model converges.
S303: and determining the initial credit risk evaluation model meeting the training conditions as a credit risk evaluation model.
The initial credit risk evaluation model which meets the training condition and does not carry the residual error module can be determined as the credit risk evaluation model.
In an embodiment provided by the present invention, based on the implementation process, optionally, feature extraction is performed on credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model to obtain a first credit behavior feature of the credit behavior data, as shown in fig. 4, where the method includes:
s401: and performing feature extraction on the credit behavior data through a convolutional neural network in a first feature extraction module of the credit risk evaluation model to obtain a first credit behavior initial feature.
In this embodiment, the convolutional neural network in the first feature extraction module includes a plurality of convolutional layers, and the credit behavior data is processed through each convolutional layer to obtain the first credit behavior initial feature.
S402: and processing the first credit behavior initial feature through a multi-head attention mechanism layer in a first feature extraction module of the credit risk evaluation model to obtain a first credit behavior feature of the credit behavior data.
In this embodiment, the multi-head attention mechanism layer includes a plurality of parallel attention mechanism layers, and the first credit behavior initial characteristic is processed by the multi-head attention mechanism layer to obtain the first credit behavior characteristic.
In an embodiment provided by the present invention, based on the foregoing implementation process, optionally, after obtaining the credit risk evaluation result of the user, the method further includes:
and outputting credit risk warning information aiming at the user under the condition that the credit risk evaluation result of the user meets the preset credit risk warning condition.
In this embodiment, the credit risk evaluation result may be a credit risk score or a credit risk rating; under the condition that the credit risk evaluation result is the credit risk score, if the credit risk score is larger than a preset risk threshold, determining that the risk evaluation result of the user meets a credit risk alarm condition; and under the condition that the credit risk evaluation result is the credit risk classification, if the credit risk classification is larger than the preset alarm classification, determining that the risk evaluation result of the user meets the credit risk alarm condition.
Referring to fig. 5, a schematic structural diagram of a credit risk evaluation model provided in an embodiment of the present invention includes that the credit risk evaluation model includes a first feature extraction module, a second feature extraction module, and an output module, where the first feature extraction module may include a convolutional neural network and a multi-head attention mechanism layer; the second feature extraction module may include a convolutional neural network having at least one convolutional layer; the output module can comprise a feature fusion layer and an output layer, and an intermediate layer can be arranged between the feature fusion layer and the output layer.
The multi-head attention mechanism layer in this embodiment may focus attention on the emphasis of the application scenario, ignoring some non-essential factors. It can be seen as a combinatorial function that highlights the effect of a key input on the output by computing the probability distribution of attention.
Note that the mechanism can be described as mapping a query and a set of corresponding key-value pairs to an output, where the query is Q, the key is K, the value is V, and the output is output, and these several elements are vectors. In general, the key K and the value V are equal. The relationship between them can be represented by the following formula:
wherein, d k Is the dimension of K, will affect the size of the dot product. softmax is an activation function that can be used to normalize weights, where Z is a K-dimensional vector, Z j Representing one element of the K-dimensional vector. With softmax we can normalize the elements in the vector between 0 and 1 and let the sum of these elements be 1.
Specifically, the key sum value is all features in the credit behavior data extracted by the convolutional neural network, and a weight matrix which needs to be learned by the corresponding attention module is queried. The output is a weighted sum of values (characteristic of credit activity data) where the weight assigned to each value is calculated by querying the compatibility function with the corresponding key. Through the attention mechanism, the deep learning model can be more focused on information which is more important for the precipitation prediction task.
A multi-head attention mechanism is made up of several parallel attention mechanism layers with different training parameters, each head performing a linear transformation before attention manipulation to project the three inputs into the lower dimension. The operation of each attention mechanism is performed independently and the final result is then obtained by concatenating the outputs of each head. Specifically, the input to the multi-head attention layer is three sequences of vectors: q, K and V. For the ith head, the formula for the attention mechanism is as follows:
in this connection, it is possible to use,
is used to map the three inputs to have a lower latitude d p The parameters of the three matrices are required for model learning. The output of the multi-head attention mechanism is then a combination of the above results for each head, and the formula of the multi-head attention mechanism is as follows:
Multihead(Q,K,V)＝Concat(head 1 ,head 2 ,...,head h )W O
where h is the number of heads,
and is also a weight matrix that requires the model to learn.
The advantage of multi-headed attention is that it can learn relevant information in different subspaces. However, some minor features may be lost due to the salient features, which may lead to the incomplete global information.
The learning expression effect of the network cannot be increased due to the depth of the simple stacked network, because gradient divergence and information loss can be caused as the depth of the network is increased. Therefore, by introducing the residual module, the problem of gradient disappearance caused by the increase of the network depth can be solved. Furthermore, residual concatenation avoids losing global features to ensure the integrity of the original information. Therefore, in this embodiment, a multi-head attention mechanism is combined with residual connection to avoid information loss caused by the attention mechanism, and the improved formula is as follows:
REAT(f,X)＝X+f(X)
wherein, f (X) ═ MultiHead (X, X). In the credit risk evaluation model, X is a feature of credit behavior data extracted by the convolutional neural network.
The multi-headed attention mechanism employed in this embodiment is a self-multi-headed attention mechanism in which the key K, the value V and the query Q are all the same tensor sequence X. In this way, each row vector in the feature matrix is a dot product of all column vectors, and the first feature extraction module provided by the embodiment of the present invention can extract more comprehensive features, that is, blend global features and local features.
The goal of the credit evaluation model is to extract high-level features of personal credits through a deep network to enable risk assessment prediction. To better capture the characteristics of personal credits, two channels are included in the model: the device comprises a first feature extraction module and a second feature extraction module. The first feature extraction module is used for extracting high-level semantics aiming at personal credit behaviors, the second feature extraction module is used for acquiring deeper feature expressions from obtained non-semantic features, self-multiple attention and residual connection are utilized in a channel of the first feature extraction module, and loss of global information is reduced while factors having important influence on credit evaluation are particularly concerned. And finally, splicing the features extracted by the two channels, and sending the spliced features into a full connection layer to obtain credit evaluation output.
Corresponding to the method shown in fig. 1, an embodiment of the present invention further provides a credit risk evaluation apparatus, which is used for specifically implementing the method shown in fig. 1, where the credit risk evaluation apparatus provided in the embodiment of the present invention may be applied to an electronic device, and a schematic structural diagram of the credit risk evaluation apparatus is shown in fig. 6, and specifically includes:
the determining unit 601 is configured to determine, in response to a credit risk evaluation instruction, a user corresponding to the credit risk evaluation instruction;
an obtaining unit 602, configured to obtain credit behavior data of the user;
a first feature extraction unit 603, configured to perform feature extraction on credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model, to obtain a first credit behavior feature of the credit behavior data;
a second feature extraction unit 604, configured to perform feature extraction on credit behavior data of the user through a second feature extraction module in the credit risk evaluation model, to obtain a second credit behavior feature of the credit behavior data;
an output unit 605, configured to obtain a credit risk evaluation result of the user based on the first credit behavior feature and the second credit behavior feature through an output module in the credit risk evaluation model.
By applying the device provided by the embodiment of the invention, the credit risk evaluation of the user is carried out by extracting the characteristics of different levels of the credit behavior data, so that the evaluation accuracy and the evaluation efficiency of the credit risk can be improved.
In an embodiment provided by the present invention, based on the above scheme, optionally, the output unit 605 includes:
the characteristic fusion subunit is used for performing characteristic fusion on the first credit behavior characteristic and the second credit behavior characteristic through a characteristic fusion layer of an output module in the credit risk evaluation model to obtain a fusion characteristic;
the output subunit is used for determining a probability value between the user and each preset candidate label based on the fusion feature through an output layer of an output module in the credit risk evaluation model; and taking the candidate label to which the target probability value with the maximum value in the probability values belongs as the credit risk evaluation result of the user.
In an embodiment provided by the present invention, based on the above scheme, optionally, the method further includes: a model training unit; the model training unit is used for:
acquiring a training data set and an initial credit risk evaluation model; the training data set comprises a plurality of training samples, and each training sample comprises historical credit behavior data of labeled sample labels of a historical user; the specimen label represents a credit risk evaluation result of the historical user;
sequentially training the initial credit risk evaluation model by utilizing each training sample in the training data set until the initial credit risk evaluation model meets a preset training condition;
and determining the initial credit risk evaluation model meeting the training conditions as a credit risk evaluation model.
In an embodiment provided by the present invention, based on the above scheme, optionally, the first feature extraction unit 603 is configured to:
performing feature extraction on the credit behavior data through a convolutional neural network in a first feature extraction module of the credit risk evaluation model to obtain a first credit behavior initial feature;
and processing the first credit behavior initial feature through a multi-head attention mechanism layer in a first feature extraction module of the credit risk evaluation model to obtain a first credit behavior feature of the credit behavior data.
In an embodiment provided by the present invention, based on the above scheme, optionally, the credit risk evaluation apparatus further includes: an alarm unit;
and the alarm unit is used for outputting credit risk alarm information aiming at the user under the condition that the credit risk evaluation result of the user meets a preset credit risk alarm condition.
The specific principle and the implementation process of each unit and each module in the credit risk evaluation device disclosed in the embodiment of the present invention are the same as those of the credit risk evaluation method disclosed in the embodiment of the present invention, and reference may be made to corresponding parts in the credit risk evaluation method provided in the embodiment of the present invention, which are not described herein again.
The embodiment of the invention also provides a storage medium, which comprises a stored instruction, wherein when the instruction runs, the equipment where the storage medium is located is controlled to execute the credit risk evaluation method.
An electronic device is provided in an embodiment of the present invention, and its structural diagram is shown in fig. 7, which specifically includes a memory 701 and one or more instructions 702, where the one or more instructions 702 are stored in the memory 701, and are configured to be executed by one or more processors 703 to perform the following operations according to the one or more instructions 702:
responding to a credit risk evaluation instruction, and determining a user corresponding to the credit risk evaluation instruction;
acquiring credit behavior data of the user;
performing feature extraction on credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model to obtain first credit behavior features of the credit behavior data;
performing feature extraction on the credit behavior data of the user through a second feature extraction module in the credit risk evaluation model to obtain a second credit behavior feature of the credit behavior data;
and obtaining a credit risk evaluation result of the user based on the first credit behavior characteristic and the second credit behavior characteristic through an output module in the credit risk evaluation model.
It should be noted that the credit risk evaluation method and apparatus, the storage medium, and the electronic device provided by the present invention can be used in the fields of artificial intelligence, block chaining, distributed, cloud computing, big data, internet of things, mobile internet, network security, chip, virtual reality, augmented reality, holography, quantum computing, quantum communication, quantum measurement, digital twin, or finance. The above description is only an example, and does not limit the application fields of the credit risk evaluation method and apparatus, the storage medium, and the electronic device provided by the present invention.
It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. For the device-like embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in the process, method, article, or apparatus that comprises the element.
For convenience of description, the above devices are described as being divided into various units by function, and are described separately. Of course, the functions of the units may be implemented in the same software and/or hardware or in a plurality of software and/or hardware when implementing the invention.
From the above description of the embodiments, it is clear to those skilled in the art that the present invention can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present invention may be embodied in the form of software products, which may be stored in a storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and include instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the embodiments or some parts of the embodiments.
The credit risk evaluation method provided by the invention is described in detail above, and the principle and the implementation mode of the invention are explained by applying specific examples, and the description of the examples is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (10)
1. A credit risk assessment method, comprising:
responding to a credit risk evaluation instruction, and determining a user corresponding to the credit risk evaluation instruction;
acquiring credit behavior data of the user;
performing feature extraction on credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model to obtain first credit behavior features of the credit behavior data;
performing feature extraction on the credit behavior data of the user through a second feature extraction module in the credit risk evaluation model to obtain a second credit behavior feature of the credit behavior data;
and obtaining a credit risk evaluation result of the user based on the first credit behavior characteristic and the second credit behavior characteristic through an output module in the credit risk evaluation model.
2. The method of claim 1, wherein obtaining, by an output module in the credit risk assessment model, a credit risk assessment result of the user based on the first credit behavior feature and the second credit behavior feature comprises:
performing feature fusion on the first credit behavior feature and the second credit behavior feature through a feature fusion layer of an output module in the credit risk evaluation model to obtain fusion features;
determining a probability value between the user and each preset candidate tag based on the fusion characteristics through an output layer of an output module in the credit risk evaluation model; and taking the candidate label to which the target probability value with the maximum value in the probability values belongs as the credit risk evaluation result of the user.
3. The method of claim 1, wherein the process of constructing the credit risk assessment model comprises:
acquiring a training data set and an initial credit risk evaluation model; the training data set comprises a plurality of training samples, and each training sample comprises historical credit behavior data of labeled sample labels of a historical user; the specimen label represents a credit risk evaluation result of the historical user;
training the initial credit risk evaluation model by using each training sample in the training data set until the initial credit risk evaluation model meets a preset training condition;
and determining the initial credit risk evaluation model meeting the training conditions as a credit risk evaluation model.
4. The method according to claim 1, wherein the obtaining of the first credit behavior feature of the credit behavior data through feature extraction of the credit behavior data of the user by a first feature extraction module in a pre-constructed credit risk evaluation model comprises:
performing feature extraction on the credit behavior data through a convolutional neural network in a first feature extraction module of the credit risk evaluation model to obtain a first credit behavior initial feature;
and processing the first credit behavior initial feature through a multi-head attention mechanism layer in a first feature extraction module of the credit risk evaluation model to obtain a first credit behavior feature of the credit behavior data.
5. The method of claim 1, wherein after obtaining the credit risk assessment result of the user, the method further comprises:
and outputting credit risk warning information aiming at the user under the condition that the credit risk evaluation result of the user meets a preset credit risk warning condition.
6. A credit risk assessment device, comprising:
the determining unit is used for responding to a credit risk evaluation instruction and determining a user corresponding to the credit risk evaluation instruction;
the acquisition unit is used for acquiring credit behavior data of the user;
the first feature extraction unit is used for performing feature extraction on credit behavior data of the user through a first feature extraction module in a pre-constructed credit risk evaluation model to obtain first credit behavior features of the credit behavior data;
the second feature extraction unit is used for performing feature extraction on the credit behavior data of the user through a second feature extraction module in the credit risk evaluation model to obtain a second credit behavior feature of the credit behavior data;
and the output unit is used for obtaining a credit risk evaluation result of the user based on the first credit behavior characteristic and the second credit behavior characteristic through an output module in the credit risk evaluation model.
7. The apparatus of claim 6, wherein the output unit comprises:
the characteristic fusion subunit is used for performing characteristic fusion on the first credit behavior characteristic and the second credit behavior characteristic through a characteristic fusion layer of an output module in the credit risk evaluation model to obtain a fusion characteristic;
the output subunit is used for determining a probability value between the user and each preset candidate label based on the fusion feature through an output layer of an output module in the credit risk evaluation model; and taking the candidate label to which the target probability value with the maximum value in the probability values belongs as the credit risk evaluation result of the user.
8. The apparatus of claim 6, further comprising: a model training unit; the model training unit is used for:
acquiring a training data set and an initial credit risk evaluation model; the training data set comprises a plurality of training samples, and each training sample comprises historical credit behavior data of labeled sample labels of a historical user; the specimen label represents a credit risk evaluation result of the historical user;
sequentially training the initial credit risk evaluation model by utilizing each training sample in the training data set until the initial credit risk evaluation model meets a preset training condition;
and determining the initial credit risk evaluation model meeting the training condition as a credit risk evaluation model.
9. A storage medium, characterized in that the storage medium comprises stored instructions, wherein when the instructions are executed, a device on which the storage medium is located is controlled to execute the credit risk assessment method according to any one of claims 1 to 5.
10. An electronic device comprising a memory and one or more instructions, wherein the one or more instructions are stored in the memory and configured to be executed by the one or more processors to perform the credit risk assessment method of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210606396.9A CN115018627A (en) | 2022-05-31 | 2022-05-31 | Credit risk evaluation method and device, storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210606396.9A CN115018627A (en) | 2022-05-31 | 2022-05-31 | Credit risk evaluation method and device, storage medium and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115018627A true CN115018627A (en) | 2022-09-06 |
Family
ID=83070658
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210606396.9A Pending CN115018627A (en) | 2022-05-31 | 2022-05-31 | Credit risk evaluation method and device, storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115018627A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117474004A (en) * | 2023-10-17 | 2024-01-30 | 中投国信(北京)科技发展有限公司 | User credit recovery evaluation method, device and storage medium |
-
2022
- 2022-05-31 CN CN202210606396.9A patent/CN115018627A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117474004A (en) * | 2023-10-17 | 2024-01-30 | 中投国信(北京)科技发展有限公司 | User credit recovery evaluation method, device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110852881B (en) | Risk account identification method and device, electronic equipment and medium | |
| CN109829629A (en) | Generation method, device, computer equipment and the storage medium of risk analysis reports | |
| CN111339260A (en) | BERT and QA thought-based fine-grained emotion analysis method | |
| CN112785005B (en) | Multi-objective task assistant decision-making method and device, computer equipment and medium | |
| CN112819024B (en) | Model processing method, user data processing method and device and computer equipment | |
| CN112561320A (en) | Training method of mechanism risk prediction model, mechanism risk prediction method and device | |
| CN116432184A (en) | Malicious software detection method based on semantic analysis and bidirectional coding characterization | |
| Wimmer et al. | Leveraging vision-language models for granular market change prediction | |
| CN111709225A (en) | Event cause and effect relationship judging method and device and computer readable storage medium | |
| Zhang | The Evaluation on the Credit Risk of Enterprises with the CNN‐LSTM‐ATT Model | |
| CN115018627A (en) | Credit risk evaluation method and device, storage medium and electronic equipment | |
| CN114637850A (en) | Abnormal behavior recognition and model training method, device, equipment and storage medium | |
| CN117217807B (en) | Bad asset estimation method based on multi-mode high-dimensional characteristics | |
| KR102546328B1 (en) | Method, device and system for providing content information monitoring and content planning automation solution for online marketing | |
| US20230281477A1 (en) | Framework system for improving performance of knowledge graph embedding model and method for learning thereof | |
| CN114119191A (en) | Wind control method, overdue prediction method, model training method and related equipment | |
| CN113761184A (en) | Text data classification method, equipment and storage medium | |
| CN115600226B (en) | Warehouse bill mortgage data processing method and device | |
| Xiong et al. | L-RBF: A customer churn prediction model based on lasso+ RBF | |
| CN116611911A (en) | Credit risk prediction method and device based on support vector machine | |
| Sun et al. | DeepMC: DNN test sample optimization method jointly guided by misclassification and coverage | |
| CN112884028A (en) | System resource adjusting method, device and equipment | |
| CN114529191B (en) | Method and device for risk identification | |
| CN118260683B (en) | Big data-based anti-fraud model training method and system | |
| KR102714099B1 (en) | Time series data-based financial anomaly detection system |
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 |