CN115994684A - Enterprise risk assessment method, enterprise risk assessment device, computer equipment and medium - Google Patents
Enterprise risk assessment method, enterprise risk assessment device, computer equipment and medium Download PDFInfo
- Publication number
- CN115994684A CN115994684A CN202211345646.4A CN202211345646A CN115994684A CN 115994684 A CN115994684 A CN 115994684A CN 202211345646 A CN202211345646 A CN 202211345646A CN 115994684 A CN115994684 A CN 115994684A
- Authority
- CN
- China
- Prior art keywords
- data
- risk
- model
- enterprise
- parameters
- 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
- 238000012502 risk assessment Methods 0.000 title claims abstract description 80
- 238000000034 method Methods 0.000 title claims abstract description 61
- 238000012545 processing Methods 0.000 claims description 55
- 238000009826 distribution Methods 0.000 claims description 27
- 238000011156 evaluation Methods 0.000 claims description 25
- 238000012216 screening Methods 0.000 claims description 16
- 238000003860 storage Methods 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 13
- 238000013210 evaluation model Methods 0.000 claims description 12
- 238000007477 logistic regression Methods 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000000875 corresponding effect Effects 0.000 description 46
- 230000006870 function Effects 0.000 description 35
- 230000008569 process Effects 0.000 description 16
- 230000008859 change Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 8
- 230000004913 activation Effects 0.000 description 7
- 238000006243 chemical reaction Methods 0.000 description 6
- 230000003287 optical effect Effects 0.000 description 6
- 238000012217 deletion Methods 0.000 description 5
- 230000037430 deletion Effects 0.000 description 5
- 238000012905 input function Methods 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 5
- 238000013139 quantization Methods 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000012417 linear regression Methods 0.000 description 3
- 238000007476 Maximum Likelihood Methods 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000011144 upstream manufacturing Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000012854 evaluation process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- 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
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an enterprise risk assessment method, an enterprise risk assessment device, computer equipment and a medium, wherein the enterprise risk assessment method in one embodiment comprises the following steps: acquiring risk data of an enterprise to be evaluated according to the acquired industrial and commercial data of the enterprise to be evaluated, wherein the risk data comprises a plurality of input features; determining model parameters of a trained risk assessment model based on the risk data, thereby obtaining a risk assessment model corresponding to the risk data; the risk data is input into a risk assessment model for assessment, wherein the model parameters of the risk assessment model are determined. According to the assessment method provided by the embodiment of the invention, the risk data is generated by acquiring and analyzing the business data of the enterprise, and different risk assessment models are determined according to different risk data, so that the assessment model corresponding to the risk data is utilized for assessment, the whole method can fully mine public information data of the enterprise, and the accuracy of risk assessment is improved.
Description
Technical Field
The present invention relates to the field of risk assessment. And more particularly, to an enterprise risk assessment method, apparatus, computer device, and medium.
Background
The supply chain is an important factor in the development of relational manufacturing enterprises. If there is a risk of inadequacy in the business upstream and downstream of the supply chain, this is a significant potential hazard for manufacturing businesses. The risk situation of enterprises at the upstream and downstream of the supply chain is mastered, and countermeasures are timely made, so that the risk is reduced by manufacturing enterprises. However, in the related art customer risk evaluation process, risk evaluation is often performed based on the customer public credit data, and the sufficiency of the utilization of the customer public credit data can be further improved.
Disclosure of Invention
The invention aims to provide an enterprise risk assessment method, an enterprise risk assessment device, computer equipment and a medium, so as to solve at least one of the problems existing in the prior art.
In order to achieve the above purpose, the invention adopts the following technical scheme:
the first aspect of the invention provides an enterprise risk assessment method, which comprises the following steps:
acquiring risk data of an enterprise to be evaluated according to the acquired industrial and commercial data of the enterprise to be evaluated, wherein the risk data comprises a plurality of input features;
determining model parameters of a trained risk assessment model based on the risk data, thereby obtaining a risk assessment model corresponding to the risk data;
The risk data is input into a risk assessment model for assessment, wherein the model parameters of the risk assessment model are determined.
Further, determining model parameters of a trained risk assessment model based on the risk data, thereby obtaining a risk assessment model corresponding to the risk data, comprising:
obtaining main body parameters of the risk assessment model based on the risk data, wherein the main body parameters comprise compensation parameters, scale parameters and weight coefficients corresponding to all risk data;
determining a data type of the risk data and obtaining specific parameters corresponding to the data type, wherein the specific parameters comprise linear model parameters corresponding to a first model and nonlinear model parameters corresponding to a second model;
a risk assessment model corresponding to the risk data is determined based on the subject parameters and the specific parameters.
Further, obtaining the main parameters of the risk assessment model based on the risk data includes:
determining the negative probability of the enterprise to be evaluated by using the risk data;
obtaining a weight formula related to the weight coefficient and the negative probability by using a logistic regression model, wherein the weight formula is as follows:
Wherein P is a negative probability, w m For the mth weight coefficient corresponding to the mth input feature in the n risk data, m is E [1, n]B is a threshold constant;
and determining a compensation parameter and a scale parameter according to the negative probability, wherein the following formula is satisfied:
compensation parameter
Scale parameter
PDO is a preset doubling fraction; odds 0 Score for presetting negative-positive probability ratio 0 Is a preset basic score.
Further, determining a data type of the risk data and obtaining a specific parameter corresponding to the data type includes:
and judging the distribution type of the risk data, and if the risk data is in linear distribution, the linear model parameters comprise a negative-positive probability ratio obtained by the negative probability.
Further, determining a risk assessment model corresponding to the risk data based on the subject parameters and the specific parameters, comprising:
determining the first model from the subject parameters and the linear model parameters,
the first model is as follows: score1 = a-B x ln (Odds);
wherein Score1 is the evaluation Score of the enterprise to be evaluated, A is the compensation parameter, and B is the scale parameterOdds is the negative-to-positive probability ratio derived from the negative probabilities, 
Further, determining a data type of the risk data, and obtaining a specific parameter corresponding to the data type, further includes:
judging the distribution type of the risk data, and if the risk data is nonlinear distribution, judging that nonlinear model parameters are nonlinear coding parameters, wherein the nonlinear model parameters meet the following formula:
wherein WOE is as follows i Coding evidence weight corresponding to the ith sub-box where a certain input feature is located, and Bad i Bad for the number of negative going labels of the ith bin T Good for the number of negative tags of the total risk data i Good for the number of forward labels of the ith bin T The number of forward tags that are the total risk data.
Further, determining a risk assessment model corresponding to the risk data based on the subject parameters and the specific parameters, comprising:
determining a second model based on the weight coefficients and the nonlinear encoding parameters,
the second model is
Wherein w is m For the mth weight coefficient corresponding to the mth input feature in the risk data, m is E [1, n]N is the number of input features, A is the compensation parameter, B is the scale parameter,
further, obtaining risk data of the enterprise to be evaluated according to the obtained business data of the enterprise to be evaluated includes:
Acquiring the main body condition of the business data to obtain a data processing scheme;
and carrying out data processing on the industrial and commercial data according to the data processing scheme so as to output risk data, wherein if the industrial and commercial data are linear distribution data, the data processing scheme comprises preprocessing, normalization processing or feature screening processing.
Further, the data processing method performs data processing on the industrial and commercial data according to the data processing scheme so as to output risk data, and further includes:
if the business data are nonlinear distribution data, the data processing scheme further comprises feature binning and evidence weight processing before normalization processing, so that bins comprising a plurality of risk data are output, wherein the risk data comprise positive labels for identifying positive directions of enterprises to be evaluated and negative labels for identifying negative directions of the enterprises to be evaluated.
Further, the method further comprises:
and training the risk assessment model by taking the industrial and commercial data of the assessed enterprise as a test set and a training set, wherein the test set comprises test features and assessment scores of the assessed enterprise, and the training set comprises training features and assessment scores of the assessed enterprise.
A second aspect of the present invention provides an enterprise risk assessment apparatus, including:
the risk data generation unit is used for obtaining risk data of the enterprise to be evaluated according to the acquired business data of the enterprise to be evaluated, wherein the risk data comprises a plurality of input features;
an evaluation model generation unit for determining model parameters of a trained risk evaluation model based on the risk data, thereby obtaining a risk evaluation model corresponding to the risk data;
and the risk assessment unit is used for inputting the risk data into a risk assessment model with the determined model parameters for assessment.
A third aspect of the invention provides a computer readable storage medium having stored thereon a computer program, characterized in that the program when executed by a processor implements a method according to the first aspect of the invention.
A fourth aspect of the invention provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method according to the first aspect when executing the program.
The beneficial effects of the invention are as follows:
according to the assessment method provided by the embodiment of the invention, the risk data is generated by acquiring and analyzing the business data of the enterprise, and different risk assessment models are determined according to different risk data, so that the assessment model corresponding to the risk data is utilized for assessment, the whole method can fully mine public information data of the enterprise, and the accuracy of risk assessment is improved.
Drawings
The following describes the embodiments of the present invention in further detail with reference to the drawings.
FIG. 1 shows a flow diagram of an evaluation method of one embodiment of the invention;
FIG. 2 shows a schematic flow chart of step S1 of an embodiment of the present invention;
FIG. 3 is a schematic flow chart of step S3 according to an embodiment of the present invention;
fig. 4 shows a schematic flow chart of step S31 according to an embodiment of the present invention;
FIG. 5 shows a schematic diagram of negative probabilities, negative-to-positive probability ratios, and assessment scores for an embodiment of the present invention;
FIG. 6 shows a schematic structural diagram of a logistic regression model according to an embodiment of the present invention;
FIG. 7 is a schematic diagram showing the relationship between the estimated value and the cost function for the case of different input eigenvalues;
FIG. 8a shows a data schematic of evidence weighting processing according to an embodiment of the present invention;
FIG. 8b is a schematic diagram of weighting coefficients and threshold constants corresponding to different input features of an embodiment of the present invention;
FIG. 8c shows a schematic diagram of feature binning corresponding to different input features in an embodiment of the present invention;
FIG. 9 is a schematic view showing a structural framework of an evaluation apparatus according to another embodiment of the present invention;
fig. 10 shows a schematic structural diagram of a computer device according to another embodiment of the present invention.
Detailed Description
In order to more clearly illustrate the present invention, the present invention will be further described with reference to examples and drawings. Like parts in the drawings are denoted by the same reference numerals. It is to be understood by persons skilled in the art that the following detailed description is illustrative and not restrictive, and that this invention is not limited to the details given herein.
One embodiment of the present invention proposes a method for evaluating enterprise risk, as shown in fig. 1, the method includes:
s1, acquiring risk data of an enterprise to be evaluated according to acquired business data of the enterprise to be evaluated, wherein the risk data comprises a plurality of input features;
s3, determining model parameters of a trained risk assessment model based on the risk data, so as to obtain a risk assessment model corresponding to the risk data;
s5, inputting the risk data into a risk assessment model with the determined model parameters for assessment.
According to the assessment method provided by the embodiment of the invention, the risk data is generated by acquiring and analyzing the business data of the enterprise, and different risk assessment models are determined according to different risk data, so that the assessment model corresponding to the risk data is utilized for assessment, the whole method can fully mine public information data of the enterprise, and the accuracy of risk assessment is improved.
The method of the embodiment of the invention will now be described with reference to specific examples:
s1, acquiring risk data of an enterprise to be evaluated according to acquired business data of the enterprise to be evaluated, wherein the risk data comprise a plurality of input features.
In the embodiment of the invention, the business data comprises a plurality of business change data such as enterprise names, addresses, responsible person names, setup or business dates, economic properties, production and management ranges, production and management modes, total funds, personnel numbers, practitioner numbers, enterprise address change frequencies, enterprise legal change frequencies, punishment frequencies, complaint frequencies and the like, and the risk hiding information of the enterprise is hidden in the data, so that the embodiment of the invention mines a plurality of business data in the step to obtain the risk data of the enterprise to be evaluated.
In an alternative embodiment, as shown in fig. 2, step S1 "obtaining risk data of the enterprise to be evaluated according to the obtained business data of the enterprise to be evaluated" includes:
s11, acquiring the main situation of the industrial and commercial data to obtain a data processing scheme. In an alternative embodiment, the business data is often in the form of a database or file. Illustratively, the database includes formats in the form of Oracle, mysql, mySql Server, etc., and the file includes formats in the form of CSV, etc.
Exemplary, the embodiment of the invention extracts the business data related to the enterprise risk assessment by means of keyword character extraction, for example, data information including enterprise address change frequency, enterprise legal change frequency, punishment frequency, complaint frequency, annual/quaternary/monthly change trend of change frequency and the like.
In an alternative embodiment, the acquiring the subject matter of the business data includes: and (3) setting a reasonable data preprocessing scheme for missing value conditions, abnormal value conditions, average values, median, maximum values, minimum values, distribution conditions and the like of each field in the industrial and commercial data.
And S13, carrying out data processing on the business data according to the data processing scheme so as to output risk data, wherein if the business data are linear distribution data, the data processing scheme comprises preprocessing, normalization processing or feature screening processing.
In an alternative embodiment, the data preprocessing includes: data cleaning, data format conversion and the like.
Illustratively, dirty data, missing values, and outliers in business data are processed during the data cleansing process. In a specific example, the deletion value and the abnormal value are processed to delete the business data whose deletion rate exceeds a certain deletion rate threshold value, and the deletion value and the abnormal value in the remaining business data are set as risk. In this example, the deletion rate threshold may be designed according to practical applications, for example, 30%,50%,90%, etc., and will not be described herein. In another example, in the data format conversion process, the business data of different formats are mainly unified, for example, the format of time data is unified, and dimension units are unified.
In an alternative embodiment, the normalization process is formulated as:
In an alternative embodiment, feature screening includes screening of individual features and correlation-based feature screening. Illustratively, the screening of individual features is a variable predictive capability-based screening, including IV value-based variable screening, stepwise-based variable screening, model (e.g., RF, GBDT, etc.) feature importance-based variable screening, LASSO regularization-based variable screening, and the like. In another example, the correlation-based feature screening includes a pairwise correlation analysis of the independent variables, a multiple co-linearity analysis of the independent variables, and correlation of the independent variables with the dependent variables.
In an alternative embodiment, the number of risk data after feature screening is less than or equal to 30, so as to improve the feasibility of the risk assessment model in actual use.
Through the feature screening process, the following effects can be achieved: (1) culling features that are less relevant to the target variable. (2) avoiding feature redundancy. (3) And the burden of later verification, deployment and monitoring risks is reduced. (4) guarantee the interpretability of the variable.
In an alternative embodiment, step S13 "data processing the business data according to the data processing scheme to output risk data" further includes:
if the business data are nonlinear distribution data, the data processing scheme further comprises feature binning and evidence weight processing before normalization processing, so that bins comprising a plurality of risk data are output, wherein the risk data comprise positive labels for identifying positive directions of enterprises to be evaluated and negative labels for identifying negative directions of the enterprises to be evaluated.
In this embodiment, considering diversity of the business data, some types of business data can be directly used as risk data for subsequent risk assessment, for example, the business data and the risk assessment are linearly distributed, but some business data cannot be directly used as risk data for subsequent risk assessment, for example, the risk is not linear between the age of the customer and the risk of default, the greater the age is, the lower the risk is, and the risk of the age is higher than that of the other age. Therefore, in order to solve the nonlinear distribution between the business data and the risk assessment, the nonlinear business data needs to be further processed, so as to improve the assessment accuracy.
In an alternative embodiment, the feature binning process is used to transform numeric business data into category type business data, and to transform continuous business data into discretized business data, enabling segmentation of risk data. The definition of the characteristic sub-boxes is as follows: the continuous variable is segmented and discretized, and the multi-state discrete variables are combined, so that the state number of the discrete variable is reduced.
In an alternative embodiment, feature binning may be divided into two binning methods, supervised and unsupervised.
The unsupervised feature binning mainly comprises the following categories: 1) Equal frequency division box: the independent variables are arranged in order from small to large, and are equally divided into k parts according to the number of the independent variables, and each part is used as a sub-box. 2) Equidistant box separation: the independent variables are arranged in order from small to large, the value range of the independent variables is divided into k equidistant intervals, and each interval is used as a sub-box. 3) Clustering and box division: independent variables are clustered into k classes by a k-means clustering method, but the order of the bins needs to be ensured in the clustering process.
The supervised binning includes Split binning and Merge binning. The Split binning is a top-down (i.e., split-based) data segmentation method. The selection indexes of the Split boxes mainly include entopy, gini index, IV value and the like. Merge binning is a bottom-up (i.e., merge-based) data discretization method, and the binning algorithm employed is Chimerger binning.
Adopts characteristic box division treatment and has the following advantages: 1) The feature binning can effectively process missing values and outliers in the features. 2) After feature binning, the data and model will be more stable. 3) The feature binning can simplify the logistic regression model, reduce the risk of model overfitting and improve the generalization capability of the model. 4) All features are uniformly transformed into category type variables. 5) The method can be applied to a standard grading card model after feature box division treatment, and grades different segments, so that the method has applicability.
In an alternative embodiment, evidence weight processing (WOE encoding) is used to convert category type business data into numeric type business data, i.e., discretized business data into continuous business data.
Based on the foregoing discussion, the numerical variable has been converted into the class variable in the feature binning process, but for the logistic regression model, only the numerical variable can be input as the feature, so the data after the feature binning process cannot be input as the feature, and therefore, the class variable after the feature binning needs to be converted.
In the related art, a method of converting a class variable into a numerical variable is commonly used in one-hot coding (one-hot vector coding), but for logical regression, a matrix output by one-hot coding is sparse, and cannot make a logistic regression model of the embodiment of the invention have a better effect, so that the embodiment of the invention uses evidence weight processing (WOE coding), and the numerical value output after the evidence weight processing has: the larger the score, the larger the contribution degree of the variable to the good label (positive direction) or the bad label (negative direction), namely that the linear relation exists between the score and the tendency of the variable, so that the evaluation based on the logistic regression model can have better effect.
Therefore, according to the embodiment of the invention, through the combined scheme of the characteristic box division processing and the evidence weight processing, the continuous industrial and commercial data are converted into the category data through the box division processing, and are converted into the continuous industrial and commercial data through the evidence weight processing, so that the nonlinear data characteristics are converted into the linear data characteristics, and the nonlinear data characteristics can be applied to a logistic regression model. The WOE code value output after the combination processing can represent not only the classification in the feature bin but also the weight of the bin classification. Also, the processing based on this combination scheme is not sensitive to fluctuations in abnormal data, for example, an individual age of 20, a 200-fold carelessly pressed keyboard, and a 10-fold fluctuation does not occur. Compared with independent thermal vector coding, the combination scheme can ensure the integrity of variables, avoid sparse matrix and dimension disasters, and improve the evaluation effect.
Based on the processing of the business data in the step S1, the embodiment of the invention can design different data processing schemes according to different conditions of the business data, can provide more dimensional information for enterprise risk evaluation, and improves evaluation accuracy.
And S3, determining model parameters of the trained risk assessment model based on the risk data, so as to obtain a risk assessment model corresponding to the risk data.
Based on the discussion of the foregoing step 1, since the types of the business data are different, and there are different types of the input as the risk assessment model, the present invention proposes a risk assessment model applicable to different types of risk data
In an alternative embodiment, as shown in fig. 3, step S3 "the risk data determines model parameters of a trained risk assessment model, thereby obtaining a risk assessment model corresponding to the risk data", including:
and S31, obtaining main body parameters of the risk assessment model based on the risk data, wherein the main body parameters comprise compensation parameters, scale parameters and weight coefficients corresponding to all the risk data.
In the embodiment of the invention, the first model and the second model are designed according to different types of risk data, but the first model and the second model have common parameters, so that in the step, the main body parameters common to the first model and the second model can be determined.
In an alternative embodiment, as shown in fig. 4, step S31 "the risk data obtains the subject parameters of the risk assessment model" includes:
s311, determining the negative probability P of the enterprise to be evaluated by using the risk data.
In one specific example, if the total risk data associated with the enterprise under evaluation includes 100 pieces, where there are 20 records associated with risk, the penalty includes 10 pieces, and the business litigation between the design and other enterprises 10 pieces, then the negative probability P is the ratio of the number of negative risk data to the number of total risk data. In another example, the negative probability P also means the rate of breach of the business, such as the probability of breach in the total queriable business records.
The ratio of the negative probability to the positive probability, odds, can be obtained from the negative probability P, satisfying the following formula:
in this embodiment, instead of directly evaluating the negative probability, odds mapping obtained according to the negative probability is used as a score, so that the influence of the negative probability can be further amplified, thereby improving the evaluation accuracy.
In a specific example, as shown in fig. 5, if the middle Odds and the negative probabilities on the right are taken as risk assessment criteria, when Odds change, for example, odds drops from 5% to 1.25%, the negative probability p is indicated to drop from 4.8% to 1.2%, but the change is not intuitive, and based on the negative probabilities, the evaluation score is mapped, so that, in the equivalent case, the evaluation score is increased from 50 to 80, and the change of the risk is more intuitively reflected.
S313, obtaining a weight formula related to the weight coefficient and the negative probability by using a logistic regression model, wherein the weight formula is as follows:
wherein w is m For the mth weight coefficient corresponding to the mth input feature in the n risk data, m is E [1, n]B is a threshold constant.
In an alternative embodiment, as shown in fig. 6, the logistic regression model includes: a characteristic input end, an input function layer, an activation function layer, a difference comparison layer, a quantization function layer and a result output end,
the feature input terminal is used for inputting risk data as input features.
The input function layer is used for multiplying each input feature by the pre-estimated weight coefficient corresponding to the input feature, summing the products and outputting product superposition data. In the embodiment of the invention, the input function layer adopts a linear regression function, and the model formula of the linear regression model is as follows: predicted value z=w T x+b, therefore, the predicted value z=w based on the linear regression model equation, taking the risk data as an input feature 1 x e +w 2 x 2 +w m x m +...+w n x n +b。
The activation function layer is connected with the input function layer and is used for calculating the product superposition data and outputting an estimated value corresponding to the current estimated weight coefficient. In the embodiment of the invention, a monotonically reducible Sigmoid function is adopted 
As an activation function, the formulas applied to the embodiment of the invention, in which the negative probability P and the input characteristics are applicable to the activation function, are as follows：
In one specific example, the value of the activation function layer output is greater than or equal to 0 and less than or equal to 1, which satisfies the a posteriori probability estimate p of category 1 (y= 1|x), namely: if there is a test sample, the result of the Sigmoid function can be used as the probability that the sample belongs to category 1.
And one end of the residual error comparison layer is positioned between the activation function layer and the quantization function layer, the other end of the residual error comparison layer is positioned at the input end and is used for comparing an estimated value with a preset threshold value, if the comparison result does not meet the preset threshold value, the estimated weight coefficient is updated, the estimated value is output until the comparison result meets the preset threshold value, and if the comparison result meets the preset threshold value, the estimated value is directly output.
In the embodiment of the invention, because the estimated weight coefficient in the input function layer may not be the optimal weight coefficient, the embodiment of the invention performs multiple iterations and comparisons through the residual comparison layer, thereby obtaining the optimal estimated weight coefficient.
In a specific example, a maximum likelihood estimation method is first used to obtain the cost function of the optimal estimated weight coefficient w, i.e. the objective function J (w). The specific process is as follows:
where p (y=1|x; w) represents the probability that given w, the result y of the input feature x belongs to class 1.
According to the definition:
and obtaining a pre-estimated weight parameter w according to the input characteristic x by using a maximum likelihood estimation method.
To simplify the operation, a logarithm is taken on both sides of this equation.
After the conversion, w is found that maximizes l (w), and the negative minimum is found, i.e., the objective function J (w).
For a better understanding of the cost function, an exemplary case when the number of input features n=1
Namely:
the function graph is shown in fig. 7:
as can be seen from the above graph, when the value y of the input feature is 1, the estimated value is
The closer to 1 the less the cost is paid, and vice versa; similarly, if the value y of the sample is 0, the value is estimated +.>
The closer to 0, the smaller the cost, and vice versa. Therefore, one cost function value corresponds to one optimal estimated weight coefficient, and the uniqueness of the estimated weight coefficient is ensured. This is also the embodiment of the invention that uses least squares instead of linear regressionThe cost function is defined by adopting a least square method, and then the cost function is defined, and the function input to the activated function layer is a non-convex function, which means that the cost function has a plurality of local minimum values and is unfavorable for solving the estimated weight coefficient. / >
Further, after the cost function is defined, a gradient descent algorithm is used for solving the convergence value, so that the optimal estimated weight coefficient is obtained.
And (3) utilizing the monotonic and minutely characteristic of the Sigmoid function, adopting a random gradient descent method, disturbing input characteristics during each iteration, continuously updating the estimated weight coefficient by the following method, comparing the estimated value with a preset threshold value, and outputting the estimated value until the comparison result meets the preset threshold value, thereby obtaining the optimal estimated weight coefficient.
The quantization function layer is connected with the activation function layer and is used for classifying the estimated value and outputting a class result. Based on the foregoing discussion, the estimation value of the Sigmoid function output is located at [0,1]Therefore, the quantization function is adopted, and the Sigmoid function result is classified into a category 1 with the value of 0.5 or more and a category 0 with the value of less than 0.5. The formula of the quantization function is:
the result output end is used for outputting the input characteristics, the category results, the estimated values and the estimated weight coefficients corresponding to the estimated values, wherein the estimated weight coefficients are used as weight coefficients of a weight formula.
Based on the above process, obtaining a predicted weight coefficient corresponding to the input feature, and taking the predicted weight coefficient as a weight coefficient in a weight formula for constructing the negative probability, thereby obtaining a weight formula with known weight:
Wherein w is m For n risk numbersAccording to the mth weight coefficient corresponding to the mth input feature, m is E [1, n]B is a threshold constant.
S315, determining compensation parameters and scale parameters according to the negative probability, and satisfying the following formula:
compensation parameter
Scale parameter
PDO is a preset doubling fraction; odds 0 Score for presetting negative-positive probability ratio 0 Is a preset basic score.
Based on the weight formula of step S313:
based on the scoring formula of the scoring card,
score=a+b× (y), a being the compensation parameter, B being the scale parameter, y being the estimated value;
the method can obtain:
from the doubling PDO definition it follows that: score+pdo=a+b×ln (2 Odds) (equation 2);
doubling fraction PDO (Point of Double Odds) represents how many fractions Odds double by a factor q (typically taken as 2), i.e. every time Odds decreases by a factor q, the fraction increases by D. Illustratively, odds=50, pdo=20, q=2, then it is expressed when the ratio of the negative number to the positive number is 50: at 1, odds is 50,
for each 2-fold decrease in Odds (e.g., from 50:1 to 25:1), the PDO score was further increased by 20 minutes on an as-received basis.
The equation 3 can be used to determine:
bringing equation 4 into equation 1 yields: 
From equation 5:
in one specific example, the Score in equation 6 employs a preset base Score 0 Odds employs a preset negative-to-positive probability ratio Odds 0 Thereby obtaining the compensation parameter
Scale parameter
Are known parameters.
The compensation parameter a and the scale parameter B can be used as the main parameters of the first model and the second model.
S33, judging the data type of the risk data, and obtaining specific parameters corresponding to the data type, wherein the specific parameters comprise linear model parameters corresponding to a first model and nonlinear model parameters corresponding to a second model;
in an alternative embodiment, step S33 "determine the data type of the risk data and obtain a specific parameter corresponding to the data type" includes:
s331, judging the distribution type of the risk data, and if the risk data are in linear distribution, the linear model parameters comprise negative-positive probability ratios obtained according to the negative-positive probabilities.
In this embodiment, for the first model, risk data is linearly distributed, and after preprocessing, normalization and feature screening, the enterprise to be evaluated can be directly evaluated by using the first model.
In an alternative embodiment, step S35 "determining a risk assessment model corresponding to the risk data based on the subject parameters and the specific parameters" includes:
s351, determining the first model according to the main body parameters and the linear model parameters,
the first model is as follows: score1 = a-B x ln (Odds);
wherein Score1 is the evaluation Score of the enterprise to be evaluated, A is a compensation parameter, B is a scale parameter, and the negative-positive probability ratio obtained by negative probability is the negative-positive probability ratio
In one specific example, the base Score is preset 0 600, the basic Odds is 50, pdo is 20, and the compensation parameter a and the scale parameter B are determined according to the foregoing steps to be:
scale parameter
Compensation parameter a=600-28.9 x ln (50))= 487.1,
the first model is Score 1=487.1+28.9×ln (Odds).
In the embodiment of the invention, the negative-positive probability ratio Odds is a linear model parameter, the negative-positive probability ratio Odds can be directly obtained according to the negative probability of the enterprise to be evaluated, and the first model can be applied to risk data in linear distribution, so that the evaluation score of the enterprise to be evaluated is obtained.
Based on the foregoing discussion, the risk data is not necessarily standard in a linear distribution, so in another alternative embodiment, step S33 "determines a data type of the risk data, and deriving a specific parameter corresponding to the data type" further includes:
S331, judging the distribution type of the risk data, and if the risk data is nonlinear distribution, the nonlinear model parameters are nonlinear coding parameters, so that the following formula is satisfied:
wherein WOE is as follows i Coding evidence weight corresponding to the ith sub-box where a certain input feature is located, and Bad i Bad for the number of negative going labels of the ith bin T Good for the number of negative tags of the total risk data i Good for the number of forward labels of the ith bin T The number of forward tags that are the total risk data.
Through WOE conversion, the input features of all risk data are positively correlated with the negative probability P, so the coefficients should be positive numbers. Fig. 8a shows an example of calculation of nonlinear coding parameters (WOE coding) under different observation samples, and the method for WOE conversion of risk data according to the embodiment of the present invention may also refer to the embodiment shown in fig. 8a for calculation, which is not described herein.
Further, WOE is based on nonlinear coding parameters i The relationship between the proportion of negative labels in the risk data and the WOE value is as follows: the greater the proportion of negative going labels, the greater the WOE. Thus, the nonlinear coding parameter WOE obtained after feature binning i Not only represents the classification of a feature bin, but also represents the weight of the bin class, and the relationship between the evaluation value and the risk data can be accurately and effectively constructed.
Further, according to the nonlinear encoding parameters determined in the above steps, in an alternative embodiment, S35, determining a risk assessment model corresponding to the risk data based on the subject parameters and the specific parameters includes:
determining a second model based on the weight coefficients and the nonlinear encoding parameters,
the second model is
Wherein w is m For the mth weight coefficient corresponding to the mth input feature in the risk data, m is E [1, n]N is the number of input features, A is the compensation parameter, B is the scale parameter。
From the second model, the direct inputs of the second model include: a is compensation parameter, B is scale parameter, nonlinear coding parameter WOE obtained according to risk data i And the weight coefficient further comprises parameters such as input characteristics, a threshold constant B, negative probability P, negative-positive probability ratio and the like obtained according to the risk data in the process of determining the compensation parameter A and the scale parameter B based on the logistic regression model.
Based on the second model, the embodiment of the invention can score the risk data in nonlinear distribution, so as to obtain the score to be evaluated of the enterprise.
S5, inputting the risk data into a risk assessment model with the determined model parameters for assessment.
In one specific example, as shown in fig. 8b, the input features generated from the risk data include rev_grp, due3059_grp, due90_grp, and due6089_grp. For example, the enterprise address change frequency is expressed in rev_grp, the enterprise legal change frequency is expressed in dur 3059_grp, the penalty frequency is expressed in dur 90_grp, and the complaint frequency is expressed in d ue6089_grp. When the risk data is in linear distribution, the weight coefficient and the threshold constant b corresponding to the input characteristic can be obtained after the processing scheme of the risk data is selected, and after the parameters of the first model are determined, the first model can be determined, so that the risk value of the enterprise is evaluated by using the first model.
Further, when the risk data is in a nonlinear distribution, the processing of the risk data not only obtains the weight coefficient and the threshold constant b corresponding to the input feature as shown in fig. 8b, but also includes: the binned packets of each input feature, such as 0-5 shown in FIG. 8c, are subjected to binning processing, and the nonlinear encoding parameters WOE obtained by performing evidence weight conversion processing i To be evaluated using the second model.
Based on the steps, the embodiment of the invention can score the risk data of different data types, has a corresponding data processing scheme and a corresponding first model for the data with linear distribution, and also designs a characteristic box and evidence weight processing scheme and a corresponding second model for the data with nonlinear distribution, thereby having wide applicability.
After the risk data is input into the corresponding risk assessment model, credit scoring can be carried out on enterprises, whether credit is given or not and the credit limit and the interest rate of the credit are given or not are determined according to the credit scoring, and therefore transaction risks existing in financial transactions are identified and reduced.
In an alternative embodiment, the method further comprises:
and training the risk assessment model by taking the industrial and commercial data of the assessed enterprise as a test set and a training set, wherein the test set comprises test features and assessment scores of the assessed enterprise, and the training set comprises training features and assessment scores of the assessed enterprise.
The training process can be referred to the previous embodiments, and will not be described herein.
As shown in fig. 9, another embodiment of the present invention provides an enterprise risk assessment apparatus, including:
The risk data generation unit is used for obtaining risk data of the enterprise to be evaluated according to the acquired business data of the enterprise to be evaluated, wherein the risk data comprises a plurality of input features;
an evaluation model generation unit for determining model parameters of a trained risk evaluation model based on the risk data, thereby obtaining a risk evaluation model corresponding to the risk data;
and the risk assessment unit is used for inputting the risk data into a risk assessment model with the determined model parameters for assessment.
It should be noted that the functions and principles performed by the evaluation device are similar to those of the evaluation method of the above embodiment, and are not described herein.
Another embodiment of the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements: acquiring risk data of an enterprise to be evaluated according to the acquired industrial and commercial data of the enterprise to be evaluated, wherein the risk data comprises a plurality of input features; determining model parameters of a trained risk assessment model based on the risk data, thereby obtaining a risk assessment model corresponding to the risk data; the risk data is input into a risk assessment model for assessment, wherein the model parameters of the risk assessment model are determined.
In practical applications, the computer-readable storage medium may take the form of any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In this embodiment, a computer-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
The computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, either in baseband or as part of a carrier wave. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, smalltalk, C ++ and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider).
As shown in fig. 10, another embodiment of the present invention provides a schematic structural diagram of a computer device. The computer device 12 shown in fig. 10 is merely an example and should not be construed as limiting the functionality and scope of use of embodiments of the present invention.
As shown in FIG. 10, the computer device 12 is in the form of a general purpose computing device. Components of computer device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, a bus 18 that connects the various system components, including the system memory 28 and the processing units 16.
The system memory 28 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM) 30 and/or cache memory 32. The computer device 12 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 34 may be used to read from or write to non-removable, nonvolatile magnetic media (not shown in FIG. 10, commonly referred to as a "hard disk drive"). Although not shown in fig. 10, a magnetic disk drive for reading from and writing to a removable non-volatile magnetic disk (e.g., a "floppy disk"), and an optical disk drive for reading from or writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In such cases, each drive may be coupled to bus 18 through one or more data medium interfaces. Memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to carry out the functions of embodiments of the invention.
A program/utility 40 having a set (at least one) of program modules 42 may be stored in, for example, memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each or some combination of which may include an implementation of a network environment. Program modules 42 generally perform the functions and/or methods of the embodiments described herein.
The computer device 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), one or more devices that enable a user to interact with the computer device 12, and/or any devices (e.g., network card, modem, etc.) that enable the computer device 12 to communicate with one or more other computing devices. Such communication may occur through an input/output (I/O) interface 22. Moreover, computer device 12 may also communicate with one or more networks such as a Local Area Network (LAN), a Wide Area Network (WAN) and/or a public network, such as the Internet, through network adapter 20. As shown in fig. 10, the network adapter 20 communicates with other modules of the computer device 12 via the bus 18. It should be appreciated that although not shown in fig. 10, other hardware and/or software modules may be used in connection with computer device 12, including, but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, data backup storage systems, and the like.
The processor unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, for example, to implement an enterprise risk assessment method provided by embodiments of the present invention.
In the description of the present invention, 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. Moreover, 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 one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
It should be understood that the foregoing examples of the present invention are provided merely for clearly illustrating the present invention and are not intended to limit the embodiments of the present invention, and that various other changes and modifications may be made therein by one skilled in the art without departing from the spirit and scope of the present invention as defined by the appended claims.
Claims (13)
1. A method for assessing risk of an enterprise, the method comprising:
acquiring risk data of an enterprise to be evaluated according to the acquired industrial and commercial data of the enterprise to be evaluated, wherein the risk data comprises a plurality of input features;
determining model parameters of a trained risk assessment model based on the risk data, thereby obtaining a risk assessment model corresponding to the risk data;
the risk data is input into a risk assessment model for assessment, wherein the model parameters of the risk assessment model are determined.
2. The assessment method according to claim 1, wherein determining model parameters of a trained risk assessment model based on the risk data, thereby yielding a risk assessment model corresponding to the risk data, comprises:
obtaining main body parameters of the risk assessment model based on the risk data, wherein the main body parameters comprise compensation parameters, scale parameters and weight coefficients corresponding to all risk data;
determining a data type of the risk data and obtaining specific parameters corresponding to the data type, wherein the specific parameters comprise linear model parameters corresponding to a first model and nonlinear model parameters corresponding to a second model;
A risk assessment model corresponding to the risk data is determined based on the subject parameters and the specific parameters.
3. The evaluation method according to claim 2, wherein deriving the subject parameters of the risk evaluation model based on the risk data comprises:
determining the negative probability of the enterprise to be evaluated by using the risk data;
obtaining a weight formula related to the weight coefficient and the negative probability by using a logistic regression model, wherein the weight formula is as follows:
wherein P is a negative probability, w m For the mth weight coefficient corresponding to the mth input feature in the n risk data, m is E [1, n]B is a threshold constant;
and determining a compensation parameter and a scale parameter according to the negative probability, wherein the following formula is satisfied:
compensation parameter
Scale parameter
PDO is a preset doubling fraction, odds 0 Score for presetting negative-positive probability ratio 0 Is a preset basic score.
4. The evaluation method according to claim 3, wherein determining a data type of the risk data and obtaining a specific parameter corresponding to the data type, comprises:
and judging the distribution type of the risk data, and if the risk data is in linear distribution, the linear model parameters comprise a negative-positive probability ratio obtained by the negative probability.
5. The evaluation method according to claim 4, wherein determining a risk evaluation model corresponding to the risk data based on the subject parameter and the specific parameter includes:
determining the first model from the subject parameters and the linear model parameters,
the first model is as follows: score1 = a-B x ln (Odds);
wherein Score1 is the evaluation Score of the enterprise to be evaluated, A is a compensation parameter, B is a scale parameter, oddsFor a negative-to-positive probability ratio derived from negative probabilities,
6. the evaluation method according to claim 3, wherein determining a data type of the risk data and obtaining a specific parameter corresponding to the data type, further comprises:
judging the distribution type of the risk data, and if the risk data is nonlinear distribution, judging that nonlinear model parameters are nonlinear coding parameters, wherein the nonlinear model parameters meet the following formula:
wherein WOE is as follows i Coding evidence weight corresponding to the ith sub-box where a certain input feature is located, and Bad i Bad for the number of negative going labels of the ith bin T Good for the number of negative tags of the total risk data i Good for the number of forward labels of the ith bin T The number of forward tags that are the total risk data.
7. The evaluation method according to claim 6, wherein determining a risk evaluation model corresponding to the risk data based on the subject parameter and the specific parameter includes:
determining a second model based on the weight coefficients and the nonlinear encoding parameters,
the second model is
Wherein w is m For the mth weight coefficient corresponding to the mth input feature in the risk data, m is E [1, n]N is the number of input features, A is the compensation parameter, B is the scale parameter,
8. the evaluation method according to any one of claims 1 to 7, wherein obtaining risk data of an enterprise to be evaluated from the obtained business data of the enterprise to be evaluated, comprises:
acquiring the main body condition of the business data to obtain a data processing scheme;
and carrying out data processing on the industrial and commercial data according to the data processing scheme so as to output risk data, wherein if the industrial and commercial data are linear distribution data, the data processing scheme comprises preprocessing, normalization processing or feature screening processing.
9. The evaluation method according to claim 8, wherein the data processing is performed on the business data according to the data processing scheme to output risk data, further comprising:
If the business data are nonlinear distribution data, the data processing scheme further comprises feature binning and evidence weight processing before normalization processing, so that bins comprising a plurality of risk data are output, wherein the risk data comprise positive labels for identifying positive directions of enterprises to be evaluated and negative labels for identifying negative directions of the enterprises to be evaluated.
10. The method of evaluating according to claim 8, wherein the method further comprises:
and training the risk assessment model by taking the industrial and commercial data of the assessed enterprise as a test set and a training set, wherein the test set comprises test features and assessment scores of the assessed enterprise, and the training set comprises training features and assessment scores of the assessed enterprise.
11. An enterprise risk assessment apparatus, comprising:
the risk data generation unit is used for obtaining risk data of the enterprise to be evaluated according to the acquired business data of the enterprise to be evaluated, wherein the risk data comprises a plurality of input features;
an evaluation model generation unit for determining model parameters of a trained risk evaluation model based on the risk data, thereby obtaining a risk evaluation model corresponding to the risk data;
And the risk assessment unit is used for inputting the risk data into a risk assessment model with the determined model parameters for assessment.
12. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the method according to any one of claims 1-10.
13. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any of claims 1-10 when the program is executed by the processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211345646.4A CN115994684A (en) | 2022-10-31 | 2022-10-31 | Enterprise risk assessment method, enterprise risk assessment device, computer equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211345646.4A CN115994684A (en) | 2022-10-31 | 2022-10-31 | Enterprise risk assessment method, enterprise risk assessment device, computer equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115994684A true CN115994684A (en) | 2023-04-21 |
Family
ID=85989453
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211345646.4A Pending CN115994684A (en) | 2022-10-31 | 2022-10-31 | Enterprise risk assessment method, enterprise risk assessment device, computer equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115994684A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117010697A (en) * | 2023-09-25 | 2023-11-07 | 山东财经大学 | Visual enterprise risk assessment method based on artificial intelligence |
| CN117829586A (en) * | 2023-12-08 | 2024-04-05 | 深圳市南弯数字科技有限公司 | Control method, equipment and storage medium of enterprise risk assessment system |
-
2022
- 2022-10-31 CN CN202211345646.4A patent/CN115994684A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117010697A (en) * | 2023-09-25 | 2023-11-07 | 山东财经大学 | Visual enterprise risk assessment method based on artificial intelligence |
| CN117010697B (en) * | 2023-09-25 | 2023-12-19 | 山东财经大学 | Visual enterprise risk assessment method based on artificial intelligence |
| CN117829586A (en) * | 2023-12-08 | 2024-04-05 | 深圳市南弯数字科技有限公司 | Control method, equipment and storage medium of enterprise risk assessment system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cao et al. | A two‐stage Bayesian network model for corporate bankruptcy prediction | |
| US20220343433A1 (en) | System and method that rank businesses in environmental, social and governance (esg) | |
| CN115994684A (en) | Enterprise risk assessment method, enterprise risk assessment device, computer equipment and medium | |
| CN112116184A (en) | Factory risk estimation using historical inspection data | |
| CN111343161A (en) | Abnormal information processing node analysis method, abnormal information processing node analysis device, abnormal information processing node analysis medium and electronic equipment | |
| CN113344700B (en) | Multi-objective optimization-based wind control model construction method and device and electronic equipment | |
| CN111612040B (en) | Financial data anomaly detection method and related device based on isolated forest algorithm | |
| WO2022155740A1 (en) | Systems and methods for outlier detection of transactions | |
| CN112241805A (en) | Defect prediction using historical inspection data | |
| CN110781970A (en) | Method, device and equipment for generating classifier and storage medium | |
| CN112883990A (en) | Data classification method and device, computer storage medium and electronic equipment | |
| CN111626331B (en) | Automatic industry classification device and working method thereof | |
| CN111724241A (en) | Enterprise invoice virtual invoice detection method based on dynamic edge feature enhanced graph attention network | |
| CN111178687A (en) | Financial risk classification method and device and electronic equipment | |
| CN113379124A (en) | Personnel stability prediction method and device based on prediction model | |
| CN111815435A (en) | Visualization method, device, equipment and storage medium for group risk characteristics | |
| CN116720946A (en) | Credit risk prediction method, device and storage medium based on recurrent neural network | |
| CN118786449A (en) | Systems and methods for generating insight based upon regulatory reports and analysis | |
| CN113780675B (en) | Consumption prediction method and device, storage medium and electronic equipment | |
| CN110619024A (en) | Credit evaluation method, system and related device | |
| US20230186414A1 (en) | Methods and Systems for Consumer Harm Risk Assessment and Ranking Through Consumer Feedback Data | |
| US11935075B2 (en) | Card inactivity modeling | |
| CN116245630A (en) | Anti-fraud detection method and device, electronic equipment and medium | |
| CN113468203B (en) | Financial user image drawing method based on recurrent neural network and attention mechanism | |
| CN112328899B (en) | Information processing method, information processing apparatus, storage medium, and electronic device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |