CN113850483A - Enterprise credit risk rating system - Google Patents
Enterprise credit risk rating system Download PDFInfo
- Publication number
- CN113850483A CN113850483A CN202111062689.7A CN202111062689A CN113850483A CN 113850483 A CN113850483 A CN 113850483A CN 202111062689 A CN202111062689 A CN 202111062689A CN 113850483 A CN113850483 A CN 113850483A
- Authority
- CN
- China
- Prior art keywords
- model
- enterprise
- risk rating
- credit risk
- data
- 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.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/25—Integrating or interfacing systems involving database management systems
- G06F16/254—Extract, transform and load [ETL] procedures, e.g. ETL data flows in data warehouses
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an enterprise credit risk rating system, which comprises a user side and a server side, wherein the user side is connected with the server side through a network; the user side is used for initiating a credit risk rating request to the server side, and the server side is used for generating a credit risk rating result according to the information risk rating request and returning the credit risk rating result to the user side; the server side comprises a data acquisition module, a data processing module and a model prediction module; the data acquisition module is used for acquiring the characteristic data of the target enterprise according to the credit risk rating request; the data processing module is used for preprocessing the collected characteristic data to obtain characteristic image information of the target enterprise; and the model prediction module is used for predicting the credit risk rating result of the target enterprise according to the characteristic portrait information of the target enterprise. The method and the system can quickly grade the credit risk of the enterprise, and are convenient and quick.
Description
Technical Field
The invention belongs to the technical field of artificial intelligence, and particularly relates to an enterprise credit risk rating system.
Background
At present, small and micro enterprises are different from personal loans, modeling samples of the small and micro enterprises are few, enterprise types are complex, banks generally rate enterprise credit risks through enterprise management or financial data, at present, the enterprise credit risks are mainly rated manually, and manual rating has great uncertainty and is easily influenced by administrative factors. And the rating efficiency is low.
Disclosure of Invention
The technical problem to be solved by the invention is to provide an enterprise credit risk rating system aiming at the defects in the prior art, which can quickly rate the credit risk of an enterprise and is convenient and quick.
In order to solve the technical problems, the invention adopts the technical scheme that: an enterprise credit risk rating system comprises a user side and a server side; the user side is used for initiating a credit risk rating request to the server side, and the server side is used for generating a credit risk rating result according to the information risk rating request and returning the credit risk rating result to the user side;
the server side comprises a data acquisition module, a data processing module and a model prediction module;
the data acquisition module is used for acquiring the characteristic data of the target enterprise according to the credit risk rating request;
the data processing module is used for preprocessing the collected characteristic data to obtain characteristic image information of the target enterprise;
and the model prediction module is used for predicting the credit risk rating result of the target enterprise according to the characteristic portrait information of the target enterprise.
In the enterprise credit risk rating system, the credit risk rating request includes an enterprise name and an enterprise identification code, and the target enterprise characteristic data acquired by the data acquisition module includes enterprise revenue amount, enterprise expenditure amount, enterprise loan data, and enterprise revenue amounts, enterprise expenditure amounts, and enterprise loan data of a plurality of associated enterprises having business traffic with the target enterprise;
the data processing module is used for preprocessing the acquired feature data and marking earning labels of target enterprises and related enterprises according to enterprise earning amount and enterprise expenditure amount, wherein the earning labels are profit enterprises or loss enterprises; marking liability labels of the target enterprise and the related enterprises according to the enterprise lending data, wherein the liability labels are high liability enterprises or low liability enterprises; the characteristic image information of the target enterprise comprises a revenue label and a liability label of the target enterprise, and a revenue label and a liability label for the related enterprise.
In the enterprise credit risk rating system, when the model prediction module predicts the credit risk rating result of the target enterprise, the feature portrait information of the target enterprise is input into a preset risk rating model to obtain the credit risk rating result, and the risk rating model is a machine learning model.
In the enterprise credit risk rating system, the server further includes a model updating module, and the model updating module is configured to update the risk rating model every T time intervals.
In the above enterprise credit risk rating system, the risk rating model is a Catboost model, and the updating module includes the following steps when updating the risk rating model:
step1, marking the latest sample data with risk levels to generate model learning set data, wherein the risk levels comprise normal, concern, secondary, suspicious and loss; the risk level is obtained through loan data, the normal represents that the history is not overdue, the attention represents that the maximum overdue of the history is less than 30 days, the secondary represents that the maximum overdue of the history is more than 30 days and less than 60 days, the suspicious represents that the maximum overdue of the history is more than 60 days and less than 90 days, and the loss represents that the maximum overdue of the history is more than 90 days; each piece of model learning set data comprises characteristic portrait information and risk level of the sample enterprise;
step2, training a Catboost model through K-fold cross validation by using model learning set data, wherein K is a positive integer greater than 1;
and 3, deploying the trained Catboost model on line.
In the above enterprise credit risk rating system, when the Catboost model is trained in step2, an output value of the Catboost model is a probability value, where the probability value is normal at a risk level of 0-0.1, the probability value is concerned at a risk level of 0.11-0.2, the probability value is subordinate at a risk level of 0.21-0.3, the probability value is suspicious at a risk level of 0.31-0.4, and the probability value is lost at a risk level of 0.41-1.0.
In the above enterprise credit risk rating system, when the Catboost model is trained in step2, the oblivious decision tree is used as a base model, and the index of each leaf node in the oblivious tree is coded as a binary vector with a length equal to the tree depth.
In the enterprise credit risk rating system, when the Catboost model is trained in the step2, the used model evaluation index is the K-S statistical value, and when the K-S statistical value is larger than 0.2, the Catboost model is output.
In the above enterprise credit risk rating system, the model prediction module is further configured to output a risk score, where the risk score is a 100-plus-100-probability value, and the probability value is a value predicted by the risk rating model.
Compared with the prior art, the invention has the following advantages: according to the method, the model prediction module with the artificial intelligence model is adopted to predict the characteristic portrait information of the target enterprise, so that the credit risk rating result of the target enterprise can be directly obtained, and the method is convenient and rapid.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
FIG. 1 is a system architecture diagram of the present invention.
Detailed Description
As shown in fig. 1, an enterprise credit risk rating system includes a user terminal 100 and a server terminal 200; the user side 100 is configured to initiate a credit risk rating request to the server side 200, and the server side 200 is configured to generate a credit risk rating result according to the information risk rating request and return the credit risk rating result to the user side 100;
the server 200 comprises a data acquisition module 201, a data processing module 202 and a model prediction module 203;
the data acquisition module 201 is configured to acquire feature data of a target enterprise according to the credit risk rating request;
the data processing module 202 is configured to pre-process the acquired feature data to obtain feature image information of the target enterprise;
the model prediction module 203 is configured to predict a credit risk rating result of the target enterprise according to the feature profile information of the target enterprise.
In this embodiment, the credit risk rating request includes an enterprise name and an enterprise identification code, and the target enterprise characteristic data acquired by the data acquisition module 201 includes enterprise revenue amount, enterprise expenditure amount, and enterprise loan data, and enterprise revenue amounts, enterprise expenditure amounts, and enterprise loan data of a plurality of associated enterprises that have business traffic with the target enterprise;
the data processing module 202 preprocesses the collected feature data, including marking earning tags of the target enterprise and the associated enterprises according to the earning amount and the expenditure amount of the enterprise, wherein the earning tags are profit enterprises or loss enterprises; marking liability labels of the target enterprise and the related enterprises according to the enterprise lending data, wherein the liability labels are high liability enterprises or low liability enterprises; the characteristic image information of the target enterprise comprises a revenue label and a liability label of the target enterprise, and a revenue label and a liability label for the related enterprise.
It should be noted that the target enterprise and the related enterprise refer to two companies with business and capital going, and in practice, the risk rating of one enterprise is greatly influenced by the related enterprise, for example, the liability risk of the dealer is easily transferred to the manufacturer, and the liability of the supplier also influences the production efficiency of the manufacturer. The invention can better evaluate the credit risk level of the target enterprise by bringing the earnings and the liabilities of the related enterprises into the characteristics of the target enterprise. In actual operation, the list of the associated enterprise may be directly input through the user terminal 100, and then the revenue information and the liability information may be extracted from an external financial system, a third-party credit system (such as websites of a national enterprise and business network, an enterprise survey, a heaven-eye survey, a credit initiator, an enterprise survey, etc.) according to the list of the associated enterprise.
In this embodiment, when the model prediction module 203 predicts the credit risk rating result of the target enterprise, the feature portrait information of the target enterprise is input into a preset risk rating model to obtain the credit risk rating result, and the risk rating model is a machine learning model.
In this embodiment, the server 200 further includes a model updating module 204, and the model updating module 204 is configured to update the risk rating model every T time interval. And T is 1 day. Through the continuous updating of the risk rating module, the capability of the risk rating module can be kept to be matched with the current enterprise development change, and the risk rating of the enterprise can be accurately given in time
In this embodiment, the risk rating model is a Catboost model, and the updating module 204 includes the following steps when updating the risk rating model:
step1, marking the latest sample data with risk levels to generate model learning set data, wherein the risk levels comprise normal, concern, secondary, suspicious and loss; the risk level is obtained through loan data, the normal represents that the history is not overdue, the attention represents that the maximum overdue of the history is less than 30 days, the secondary represents that the maximum overdue of the history is more than 30 days and less than 60 days, the suspicious represents that the maximum overdue of the history is more than 60 days and less than 90 days, and the loss represents that the maximum overdue of the history is more than 90 days; each piece of model learning set data comprises characteristic portrait information and risk level of the sample enterprise;
step2, training a Catboost model through K-fold cross validation by using model learning set data, wherein K is a positive integer greater than 1; specifically, K-fold cross validation is a model performance verification method, wherein K is a positive integer larger than 1, original data is divided into K equal parts, K-1 parts are randomly selected as a training set, the rest 1 parts are used as a validation set, a classifier is trained by the training set to generate a model, the model obtained by training is tested by the validation set, performance indexes are returned, in order to reduce sampling errors, all combinations of the training set and the validation set need to be traversed, and finally the average value of all generated indexes is taken as the final evaluation result of the model. Through K-fold verification, the model effect is verified, a reasonable basis is provided for the application of the model under the real service environment, and the default K value is selected to be K5.
And 3, deploying the trained Catboost model on line.
It should be noted that the castboost is a short form of a probabilistic + Boosting (a novel machine learning algorithm based on a Gradient enhanced decision tree (GBDT) algorithm, and is a novel machine learning algorithm based on a GBDT algorithm, and the castboost is based on a symmetric decision tree as a base learner, and by embedding an algorithm for automatically processing a class feature into a numerical feature, the method has an accurate and efficient processing effect on the class feature, can well process a classification feature problem and can effectively reduce an overfitting problem, and is greatly improved in many aspects compared with other early GBDT algorithms such as XGBoost and LightGBM, and the castboost is particularly suitable for processing a large amount of data and Features.
In a traditional gradient lifting algorithm, the overfitting problem caused by biased gradient estimation is difficult to avoid. The Catboost is improved by adopting unbiased estimation of gradient step length in the first stage of building the tree, and the second stage is the same as the traditional GBDT.
Catboost uses a way to improve GreedyTS to achieve statistics of the target variables. The model reduces data noise and generates fewer class characteristics by adding a prior distribution term p on the basis of taking the label average value as a node splitting standard. The calculation formula is as follows:
wherein, a represents a weight coefficient larger than 0, and p represents an increased prior value, which is helpful for reducing noise of low-frequency category, thereby effectively avoiding the overfitting phenomenon. For features with a smaller number of classes, the addition of the prior term facilitates the reduction of noise data.
Catboost uses the order principle to rely on the set of samples that have been observed so far for the calculation of TS values. Selecting a tree structure based on a greedy algorithm, finding out all possible segmentation modes, calculating a penalty function of each mode, selecting the smallest mode, distributing the result to leaf nodes, repeating the process by the subsequent leaf nodes, carrying out random rearrangement before constructing a new tree, constructing the new tree according to a gradient descending direction, and using different arrangements by the Catboost in different gradient lifting steps. The Catboost algorithm calculates the importance of the feature variables using the following formula:
wherein, c1，c2Number of documents in leaf node, v1，v2The value of the calculation formula in the leaf node.
The importance of the characteristic variables is calculated, and the first 20 characteristic variables are extracted according to the importance degree of the characteristic variables to reduce the dimension of high-dimensional data, so that the information redundancy is reduced, and the time complexity of the model is reduced.
In this embodiment, when the Catboost model is trained in step2, an output value of the Catboost model is a probability value, where the probability value is normal at a risk level of 0 to 0.1, the probability value is concerned at a risk level of 0.11 to 0.2, the probability value is subordinate at a risk level of 0.21 to 0.3, the probability value is suspicious at a risk level of 0.31 to 0.4, and the probability value is lost at a risk level of 0.41 to 1.0.
In this embodiment, when the castboost model is trained in step2, the oblivious decision tree is used as a base model, and the index of each leaf node in the oblivious tree is coded as a binary vector with a length equal to the tree depth. The node value calculation method avoids the problem of direct over-fitting calculation, and the same segmentation criteria are used on the whole level of the oblivious tree, so that the oblivious tree can reach balance and is not easy to over-fit. The features are stored in a discretization mode, and the histogram calculation method does not depend on atomic operation and is more efficient to realize.
It should be noted that, in actual operation, perfect hash is used to perform bitwise compression on the class characteristics. And a distributed learning mode is adopted, a plurality of data learning sets are calculated side by side, and the acceleration of a plurality of GPUs can be realized.
In this embodiment, when the Catboost model is trained in step2, the model evaluation index used is the K-S statistic value, and when the K-S statistic value is greater than 0.2, the Catboost model is output.
It should be noted that the model effect mainly lies in being able to distinguish each risk level, the commonly used evaluation index Kolmogorov-Smirnov (K-S) statistic value is used to measure the prediction result, the higher K-S indicates the stronger distinguishing capability of the model for positive and negative samples, and the calculation method is as follows:
assuming that f (s | P) is the cumulative distribution function of the predicted values of the positive samples and f (s | N) is the cumulative distribution function of the negative samples on the predicted values, then:
the KS value is the maximum value of the difference between the cumulative bad occupancy curve and the cumulative good occupancy curve. The KS value represents the ability of the model to distinguish between positive and negative examples. The larger the KS value, the better the prediction accuracy of the model. The positive and negative samples are defined herein as: the normal and concern risk levels are low, positive, and the secondary, suspicious, loss risk levels are high, negative.
In the verification process, KS is more than 0.2 ideally, when KS is less than 0.2, the verification set is input again for training the model deviation sample of the real-time risk level and the model prediction result, parameters are adjusted continuously, and finally the optimization of the credit rating model is achieved.
In this embodiment, the model prediction module 203 is further configured to output a risk score, where the risk score is a 100-. And the final credit rating is 90 points or more corresponding to normal risk enterprise tags, 80-89 points corresponding to attention risk enterprise tags, 70-79 points corresponding to secondary risk enterprise tags, 60-69 points corresponding to suspicious risk enterprise tags, and 60 points or less corresponding to loss enterprise tags, and the credit risk level corresponding to the threshold interval of the prediction rating result is determined.
In this embodiment, the deploying, uploading and monitoring of the Catboost model specifically include the following conditions:
step1, model online
Deploying a Catboost model; the enterprise with normal risk level predicts extremely low overdue probability and extremely high enterprise credit, triggers automatic response and directly applies for the credit; enterprise prediction small probability concerning the risk level is overdue, a suggested manual quick passing response is triggered, manual quick passing of the audit is reminded, and the risk point of the user is synchronously transmitted to serve as a manual reference; enterprises with secondary risk levels have a certain probability of default after loan, trigger a suggested manual routine review response, remind manual routine verification, and synchronously transmit the risk points of the users as manual references; enterprise forecast of suspicious risk levels has a high probability of generating a default after loan, a suggested manual prudent examination response is triggered, manual prudent examination is reminded, and risk points of users are synchronously transmitted to serve as manual references; enterprise prediction of loss risk level will have overdue violations after loan with great probability, trigger automatic rejection response, and directly reject its credit application.
Step2, model monitoring
And (3) regularly following the performance condition of the rated enterprises, and evaluating the attenuation of the prediction capability of the model by monitoring indexes such as score distribution, passing rate, overdue rate and stability of the customer group.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and all simple modifications, changes and equivalent structural changes made to the above embodiment according to the technical spirit of the present invention still fall within the protection scope of the technical solution of the present invention.
Claims (9)
1. An enterprise credit risk rating system is characterized by comprising a user side and a server side; the user side is used for initiating a credit risk rating request to the server side, and the server side is used for generating a credit risk rating result according to the information risk rating request and returning the credit risk rating result to the user side;
the server side comprises a data acquisition module, a data processing module and a model prediction module;
the data acquisition module is used for acquiring the characteristic data of the target enterprise according to the credit risk rating request;
the data processing module is used for preprocessing the collected characteristic data to obtain characteristic image information of the target enterprise;
and the model prediction module is used for predicting the credit risk rating result of the target enterprise according to the characteristic portrait information of the target enterprise.
2. The system of claim 1, wherein the credit risk rating request includes a business name and a business identification code, and the target business characteristic data collected by the data collection module includes a business revenue amount, a business expenditure amount, and business loan data of a plurality of related businesses that have business transactions with the target business;
the data processing module is used for preprocessing the acquired feature data and marking earning labels of target enterprises and related enterprises according to enterprise earning amount and enterprise expenditure amount, wherein the earning labels are profit enterprises or loss enterprises; marking liability labels of the target enterprise and the related enterprises according to the enterprise lending data, wherein the liability labels are high liability enterprises or low liability enterprises; the characteristic image information of the target enterprise comprises a revenue label and a liability label of the target enterprise, and a revenue label and a liability label for the related enterprise.
3. The enterprise credit risk rating system of claim 2, wherein the model prediction module inputs the feature profile information of the target enterprise into a preset risk rating model to obtain the credit risk rating result when predicting the credit risk rating result of the target enterprise, and the risk rating model is a machine learning model.
4. The enterprise credit risk rating system of claim 3, wherein the server further comprises a model update module for updating the risk rating model every T times.
5. The enterprise credit risk rating system of claim 4, wherein the risk rating model is a Catboost model, and wherein the model update module updating the risk rating model comprises:
step1, marking the latest sample data with risk levels to generate model learning set data, wherein the risk levels comprise normal, concern, secondary, suspicious and loss; the risk level is obtained through loan data, the normal represents that the history is not overdue, the attention represents that the maximum overdue of the history is less than 30 days, the secondary represents that the maximum overdue of the history is more than 30 days and less than 60 days, the suspicious represents that the maximum overdue of the history is more than 60 days and less than 90 days, and the loss represents that the maximum overdue of the history is more than 90 days; each piece of model learning set data comprises characteristic portrait information and risk level of the sample enterprise;
step2, training a Catboost model through K-fold cross validation by using model learning set data, wherein K is a positive integer greater than 1;
and 3, deploying the trained Catboost model on line.
6. The enterprise credit risk rating system of claim 5, wherein in training the Catboost model in step2, the output of the Catboost model is a probability value, wherein the probability value is normal for a risk level of 0-0.1, the probability value is focus for a risk level of 0.11-0.2, the probability value is secondary for a risk level of 0.21-0.3, the probability value is suspicious for a risk level of 0.31-0.4, and the probability value is loss for a risk level of 0.41-1.0.
7. The enterprise credit risk rating system of claim 5, wherein the Catboost model in step2 is trained using an oblivious decision tree as a base model, and the index of each leaf node in the oblivious tree is encoded as a binary vector having a length equal to the tree depth.
8. The business credit risk rating system of claim 5, wherein the model evaluation index used in training the Catboost model in step2 is a K-S statistic value, and the Catboost model is output when the K-S statistic value is greater than 0.2.
9. The enterprise credit risk rating system of claim 1, wherein the model prediction module is further configured to output a risk score, wherein the risk score is a 100-by-100 probability value, and the probability value is a value predicted by the risk rating model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111062689.7A CN113850483A (en) | 2021-09-10 | 2021-09-10 | Enterprise credit risk rating system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111062689.7A CN113850483A (en) | 2021-09-10 | 2021-09-10 | Enterprise credit risk rating system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113850483A true CN113850483A (en) | 2021-12-28 |
Family
ID=78973598
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111062689.7A Withdrawn CN113850483A (en) | 2021-09-10 | 2021-09-10 | Enterprise credit risk rating system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113850483A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115713249A (en) * | 2022-10-10 | 2023-02-24 | 重庆移通学院 | Government affair satisfaction evaluation system and method based on data security and privacy protection |
-
2021
- 2021-09-10 CN CN202111062689.7A patent/CN113850483A/en not_active Withdrawn
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115713249A (en) * | 2022-10-10 | 2023-02-24 | 重庆移通学院 | Government affair satisfaction evaluation system and method based on data security and privacy protection |
| CN115713249B (en) * | 2022-10-10 | 2023-06-13 | 重庆移通学院 | Government satisfaction evaluation system and method based on data security and privacy protection |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111291816B (en) | Method and device for carrying out feature processing aiming at user classification model | |
| US11348012B2 (en) | System and method for forming predictions using event-based sentiment analysis | |
| CN110852881B (en) | Risk account identification method and device, electronic equipment and medium | |
| CN104321794A (en) | A system and method using multi-dimensional rating to determine an entity's future commercial viability | |
| CN113095927A (en) | Method and device for identifying suspicious transactions of anti-money laundering | |
| CN117041017B (en) | Intelligent operation and maintenance management method and system for data center | |
| CN111160959A (en) | User click conversion estimation method and device | |
| CN112836750A (en) | System resource allocation method, device and equipment | |
| Sun et al. | Financial distress prediction based on similarity weighted voting CBR | |
| CN113706291A (en) | Fraud risk prediction method, device, equipment and storage medium | |
| CN113962160A (en) | Internet card user loss prediction method and system based on user portrait | |
| CN110310012B (en) | Data analysis method, device, equipment and computer readable storage medium | |
| CN114187125A (en) | Claims case distribution method, device, equipment and storage medium | |
| CN113850483A (en) | Enterprise credit risk rating system | |
| CN113742495B (en) | Rating feature weight determining method and device based on prediction model and electronic equipment | |
| CN113571198B (en) | Conversion rate prediction method, conversion rate prediction device, conversion rate prediction equipment and storage medium | |
| CN114741515A (en) | Social network user attribute prediction method and system based on graph generation | |
| CN112712270B (en) | Information processing method, device, equipment and storage medium | |
| CN115936104A (en) | Method and apparatus for training machine learning models | |
| CN109308565B (en) | Crowd performance grade identification method and device, storage medium and computer equipment | |
| CN112613920A (en) | Loss probability prediction method and device | |
| CN112308294A (en) | Default probability prediction method and device | |
| CN116050539A (en) | Machine learning classification model interpretation method based on greedy algorithm | |
| Xue et al. | Calibrated data-dependent constraints with exact satisfaction guarantees | |
| CN118094365A (en) | Transaction data processing method and related 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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20211228 |