CN110956273A - Credit scoring method and system integrating multiple machine learning models - Google Patents
Credit scoring method and system integrating multiple machine learning models Download PDFInfo
- Publication number
- CN110956273A CN110956273A CN201911081115.7A CN201911081115A CN110956273A CN 110956273 A CN110956273 A CN 110956273A CN 201911081115 A CN201911081115 A CN 201911081115A CN 110956273 A CN110956273 A CN 110956273A
- Authority
- CN
- China
- Prior art keywords
- model
- machine learning
- credit
- sub
- models
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Technology Law (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Development Economics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention discloses a credit investigation scoring method and system fusing multiple machine learning models, wherein the method comprises the following steps: step 1, cleaning and converting customer data, processing missing values, combining and serializing characteristics, and deriving indexes. And 2, establishing identity traits, performance ability, behavior preference, credit history and social attribute models based on a machine learning algorithm. And 3, training and predicting the established model. And 4, fusing and weighting the sub-models, establishing a credit investigation comprehensive scoring model, and training and predicting. And 5, mapping the final model result scores, and dividing the customers into seven evaluation grades of extremely poor, common, medium, good, excellent and excellent. The invention analyzes the past actual performance of the user by quantifying the data characteristics of different dimensions of the financial user, thereby realizing the comprehensive evaluation of credit investigation of the user.
Description
Technical Field
The invention relates to the technical field of machine learning algorithms and big data processing, in particular to a credit investigation scoring method and system integrating multiple machine learning models.
Background
The Gradient Boosting Decision Tree (GBDT) is a Boosting Tree model, and each time the Tree model is built, the Gradient descending direction of the model loss function is built in the past.
Logistic Regression (LR) is a generalized linear Regression analysis model, and its algorithm is simple and efficient, and is often used to solve the problem of binary classification.
The standard scoring card model generally adopts a logistic regression algorithm, woe conversion is adopted for characteristics, model training is carried out on the basis of stepwise regression to select variables, and the number of the variables entering the model is controlled.
Random Forest (RF) is a classifier that trains and predicts samples using a number of trees, the class of which is determined by the mode of the class output by the individual trees.
The credit rating card model based on logistic regression is a prediction method widely applied to the fields of credit risk assessment and financial risk control, and the principle of the credit rating card model is that a model variable WOE coding mode is discretized and then an LR model is applied to carry out a generalized linear model of two classification variables.
The RFM model is an important tool and means for measuring the value of a user and the profit creating capability of the client, and the value condition of the client is described by 3 indexes of recent purchasing behavior, total purchasing frequency and purchasing amount of the user.
The SAS is a large application software system, which is modularized and integrated, and is composed of dozens of special modules, and functions of the SAS include data access, data storage and management, application development, graphic processing, data analysis, report compilation, an operation research method, measurement economics and the like.
The pearson correlation coefficient is a quotient used to measure the covariance and the standard deviation between two variables X and Y, and has a value between-1 and 1.
0-1 normalization, also known as dispersion normalization, is a linear transformation of the raw data to bring the result to the [0,1] interval. The formula is (X-min)/(max-min).
The prior art has certain defects:
1. the Logistic regression model is sensitive to multiple and collinear independent variables in the model, a large number of various features or variables cannot be well processed, and the requirement of training data is linear divisible.
2. Random forms do not necessarily work well on small and low-dimensional data sets and are easily over-fitted in the presence of large amounts of noise.
3. The credit score card model processes the features in a WOE coding mode and uses a Logistic regression model for training and prediction, and the feature processing and the model are single.
4. SAS has a strong computational power in statistical analysis, but is inferior to hadoop + spark in big data machine learning.
Disclosure of Invention
The traditional credit scoring model is based on statistical analysis software like SAS, and the like, utilizes user credit historical data and uses a Logistic regression and other single model methods to construct an evaluation model, but mass financial data cannot be effectively utilized, and valuable data resources are wasted; secondly, the model algorithm is single, and the ultimate effect is not particularly good due to the limitation of the model algorithm; furthermore, the model evaluation dimension is one-sided, and the comprehensive evaluation capability for the client is lacked.
Aiming at overcoming the defects in the prior art, the method adopts a mass of financial data sources, integrates a plurality of machine learning algorithms, realizes fine-grained and high-precision customer evaluation, including evaluation of customer value and customer risk, realizes the separation of stock customer groups, and forms a credit scoring system for customers. Therefore, mass financial data are efficiently processed, and the utilization rate of the data is improved; the model effect is obviously improved by organically combining a plurality of model algorithms; the evaluation of the client from multiple dimensions is realized, the performance of the client is evaluated in a three-dimensional manner, and the evaluation of the client is more comprehensive and accurate.
The technical scheme of the invention aims to quantitatively evaluate the risk performance and the value performance of the user, quantifies the actual performance of the user in the past year by an expert scoring method, predicts the expected performance of the user in the future year by a machine learning algorithm, and realizes the accurate evaluation of the user by the combination mode of the actual performance and the expected performance.
Based on data characteristics of different dimensions, firstly, the scoring dimension is disassembled, the five aspects including identity traits, performance ability, behavior preference, credit history and social attributes are included, and submodels are respectively constructed according to the data characteristics of the five dimensions, so that the scoring accuracy and interpretability are improved.
The credit investigation scoring system machine learning model realized by the invention comprises the following whole processes:
s1: cleaning and converting client data based on hadoop, performing missing value processing, characteristic combination, serialization and derivation indexes;
s2: establishing an identity trait model and a performance capability model for training and predicting based on a machine learning algorithm;
s3: constructing a behavior preference prediction model based on the combined application of a GBDT + logistic regression algorithm, and completing the construction of a behavior evaluation model by adopting a combined partitioning thought of an RFM model for training and prediction;
s4: establishing a social attribute value model based on a user relationship network, and training and predicting;
s5: establishing a credit history model based on a standard scoring card algorithm, and training and predicting;
s6: fusing and weighting the five sub-models, establishing a credit investigation comprehensive scoring model, and training and predicting;
s7: and mapping the final model result scores, and dividing the customers into seven evaluation grades of extremely poor, normal, medium, good, excellent and super.
The following five-dimensional modeling ideas are described in detail:
1. identity trait model: respectively defining a high-value client and a low-risk client as targets, taking the characteristics of the model from client attributes such as age, gender, occupation and the like, discretizing and serializing the characteristics, and constructing an identity characteristic model by adopting a Logistic regression model.
2. Performance capacity value model: respectively defining a high-value client and a low-risk client as targets, taking model characteristics from relevant indexes of client repayment capacity such as annual salaries, amount and the like, and performing discretization and serialization processing on the characteristics. And the model adopts a random forest model algorithm to complete the construction of the performance capability model.
3. Behavior preference model: respectively defining a high-value client and a low-risk client as targets, firstly establishing a client behavior preference prediction model, taking the characteristics of the model from client behavior expressions such as credit card transaction behaviors, staged product handling and the like, discretizing and serializing the characteristics, then deriving new characteristics by using a GBDT model algorithm in a mode of intercepting a GBDT tree branch, adopting a Logistic regression model for the model, and adopting L2 regularization for regularization; and meanwhile, according to the three actual expressed dimensions of the current value, the liveness and the risk level of the client, a user evaluation model is constructed by an expert scoring method. And finally, the construction of the behavior preference model is completed through the combination of the user evaluation model and the user prediction model.
4. Credit history model: defining overdue and bad risk customers as targets according to different weights, constructing a behavior evaluation model based on the credit history characteristics of the customers, predicting the credit history performance of the customers, and completing the construction of the credit history model, wherein the algorithm is constructed in a standard scoring card model mode.
5. The social attribute model is as follows: and defining different relation weights according to the evaluation result derivative characteristics of the identity trait model, the performance capability model, the behavior preference model and the credit history model of the customer relation circle, constructing a customer relation network, obtaining the user relation network score, and completing the construction of the social attribute model.
Preferably, in step 1, a hadoop distributed system infrastructure is used for data processing.
Preferably, in step 3, the identity trait model is established, and the fulfillment capability model adopts a machine learning algorithm.
Preferably, in step 3, a preference model is established, a machine learning algorithm is adopted, and combination is performed based on an RFM algorithm.
Preferably, in step 3, a social attribute model is established, and a user relationship network is adopted.
Preferably, in step 3, a credit history model is established, and a standard scoring card algorithm is adopted. And finally, defining high-value low-risk customers as target customer groups based on five-dimensional submodels, quantitatively evaluating the weights of the identity trait model, the performance capability model, the behavior preference model, the social attribute model and the credit history model, and performing combined weighting to obtain final credit investigation scores.
The invention also discloses a credit investigation scoring system fusing various machine learning models, which comprises the following components:
the data cleaning module is used for processing missing data, combining and serializing characteristics and segmenting according to different data characteristics;
the model target definition module is used for completing the definition of the model target based on the comprehensive performance of the user in dimensions such as income, operation cost, risk cost, transaction performance, product purchase performance and the like;
the sub-model creation module is used for creating five sub-models by training five sub-model data based on model targets by utilizing sample data, wherein the sub-models comprise identity traits, performance capability, behavior preference, credit history and social attributes;
the submodel training module is used for carrying out weighted fusion on the basis of the pearson correlation coefficients of the submodels and the model target, establishing a credit investigation comprehensive scoring model for the five created submodels, and carrying out training and prediction;
and the standardization processing module is used for carrying out mapping processing on the final model result scores and dividing the customers into seven evaluation grades of extremely poor, general, medium, good, excellent and excellent.
The technical scheme provided by the embodiment of the invention constructs a comprehensive evaluation model of user value, risk and performance for a financial institution, realizes the most comprehensive and accurate evaluation of the user from multiple dimensions, provides a set of unified evaluation standard for business operation of various scenes, is efficiently and widely applied to various links of user credit granting, product marketing and equity gifting, and achieves remarkable effect.
Drawings
Fig. 1 is a flow chart of a credit investigation scoring method fusing multiple machine learning models according to the present invention.
Fig. 2 is a schematic diagram of a credit assessment system fusing multiple machine learning models according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to specific examples, but the embodiments of the present invention are not limited thereto.
As shown in fig. 1, an embodiment of the present invention provides a credit assessment method that integrates multiple machine learning models.
Unless defined otherwise, technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention belongs. The use of "first," "second," and similar terms in the description and claims of this patent does not denote any order, quantity, or importance, but rather the terms are used to distinguish one element from another. Also, the use of the terms "a" or "an" and the like do not denote a limitation of quantity, but rather denote the presence of at least one.
The above description is only exemplary embodiments of the present invention and should not be taken as limiting the invention, and any modifications, equivalents, improvements and the like that are within the spirit and principle of the present invention should be included in the scope of the present invention.
The technical scheme provided by the embodiment of the invention constructs a comprehensive evaluation model of user value, risk and performance for a financial institution, realizes the most comprehensive and accurate evaluation of the user from multiple dimensions, provides a set of unified evaluation standard for business operation of various scenes, is efficiently and widely applied to various links of user credit granting, product marketing and equity gifting, and achieves remarkable effect.
The credit investigation scoring system machine learning model realized by the invention comprises the following whole processes:
s1: cleaning and converting client data based on hadoop, performing missing value processing, characteristic combination, serialization and derivation indexes; step 1 can adopt hadoop distributed system infrastructure to process data.
S2: establishing an identity trait model and a performance capability model for training and predicting based on a machine learning algorithm;
for the identity characteristic model, a high-value client and a low-risk client are respectively defined as targets, the characteristics of the model are obtained from client attributes such as age, gender, occupation and the like, discretization and serialization are firstly carried out on the characteristics, the model adopts a Logistic regression model, regularization adopts L1 regularization, and the identity characteristic model is constructed.
For the performance ability value model, a high-value client and a low-risk client are respectively defined as targets, model characteristics are taken from relevant indexes of client repayment ability, such as annual pay, amount and the like, and discretization and serialization processing are carried out on the characteristics. The model adopts a RandomForests model algorithm to complete the construction of the performance capability model.
S3: a machine learning algorithm is adopted, and combination realization is carried out based on an RFM algorithm, a behavior preference model is constructed, and training and prediction are carried out;
behavior preference model: respectively defining a high-value client and a low-risk client as targets, firstly establishing a client behavior preference prediction model, taking the characteristics of the model from client behavior expressions such as credit card transaction behaviors, staged product handling and the like, discretizing and serializing the characteristics, then deriving new characteristics by using a GBDT model algorithm in a mode of intercepting a GBDT tree branch, adopting a Logistic regression model for the model, and adopting L2 regularization for regularization; and simultaneously, according to the three actual expressed dimensions of the current value, the liveness and the risk level of the client, the combination is realized through an RFM algorithm, and the construction of a behavior preference model is completed.
S4: establishing a social attribute value model based on a user relationship network, and training and predicting;
the social attribute model is as follows: and defining different relationship weights according to the evaluation result derivative characteristics of the identity trait model, the performance capability model, the behavior preference model and the credit history model of the customer relationship circle, constructing a customer relationship network according to the first-degree friends and the second-degree friends of the friend circle of the user, obtaining the user relationship network score, and completing the construction of the social attribute model.
S5: establishing a credit history model based on a standard scoring card algorithm, and training and predicting;
credit history model: and defining overdue and bad risk customers as targets according to different weights, constructing a behavior evaluation model based on the credit history characteristics of the customers, predicting the credit history performance of the users, and completing the construction of the credit history model. Unlike the first three models, the credit history model is a behavior scoring card model.
S6: fusing and weighting the five sub-models, establishing a credit investigation comprehensive scoring model, and training and predicting;
s7: standardizing the final model result, and dividing evaluation grades for each client;
and finally, defining high-value low-risk customers as target customer groups based on five-dimensional submodels, quantitatively evaluating the weights of the identity trait model, the performance capability model, the behavior preference model, the social attribute model and the credit history model, and performing combined weighting to obtain final credit investigation scores.
As shown in fig. 2, the present invention also discloses a credit assessment system 20 integrating multiple machine learning models, which includes:
the data cleaning module 201 is used for processing missing data values, combining and serializing features, and segmenting according to different data characteristics;
the model target definition module 202 is used for completing the definition of the model target based on the comprehensive performance of the user in dimensions such as income, operation cost, risk cost, transaction performance, product purchase performance and the like;
the sub-model creating module 203 is used for creating five sub-models by training five sub-model data based on model targets by utilizing sample data, wherein the sub-models comprise identity traits, performance capability, behavior preference, credit history and social attributes;
the submodel training module 204 is used for carrying out weighted fusion on the basis of the pearson correlation coefficients of the submodels and the model targets, establishing a credit investigation comprehensive scoring model for the five created submodels, and carrying out training and prediction;
and the standardization processing module 205 is used for standardizing the final model result 0-1 and dividing each customer evaluation grade.
Claims (8)
1. A credit investigation scoring method fusing multiple machine learning models is characterized by comprising the following steps:
step 1, cleaning client data, processing missing data values, combining and serializing features, and segmenting according to different data characteristics to serve as samples of five sub-models;
step 2, completing model target definition based on comprehensive expressions of the user in dimensions such as income, operation cost, risk cost, transaction expression, product purchase expression and the like;
step 3, training five sub-model data based on the model target defined in the step 2 by using the sample data in the step 1 to obtain five sub-models, wherein the five sub-models comprise: identity traits, performance capabilities, behavioral preferences, credit history, and social attributes;
step 4, carrying out weighted fusion on the five submodels in the step 3 based on the pearson correlation coefficients of the submodels and the model targets, establishing a credit investigation comprehensive scoring model, and carrying out training and prediction;
and 5, standardizing the final model result 0-1, and grading each client according to evaluation grades.
2. The credit investigation scoring method fusing various machine learning models according to claim 1, characterized in that step 1 adopts hadoop distributed system infrastructure for data processing.
3. The credit investigation scoring method fusing various machine learning models according to claim 1, characterized in that in the step 3, an identity characteristic model is established, and a logistic regression algorithm is adopted for model construction.
4. The credit investigation scoring method fusing multiple machine learning models according to claim 1, wherein in the step 3, a performance capability model is established, and a random forest algorithm is adopted for model construction.
5. The credit investigation scoring method fusing various machine learning models as claimed in claim 1, wherein in the step 3, a behavior preference model is established, a GBDT algorithm is adopted for specific derivation, a logistic regression algorithm is adopted for model construction, and the model construction is combined and separated from the transaction performance, the product purchase performance and the activity of the user based on the implementation idea of the RFM model, so that the behavior preference model construction is completed.
6. The credit investigation scoring method fusing various machine learning models as claimed in claim 1, characterized in that in the step 3, a social attribute model is established, and a relationship network algorithm is adopted, so that scoring performances of friends of the user who circle a first degree and friends of the user who circle a second degree on the four sub-models in the step 3 are taken as scoring of the user social attribute sub-model.
7. The credit assessment method fusing multiple machine learning models according to claim 1, wherein in the step 3, a credit history model is established and implemented by adopting a standard scoring card model.
8. A credit assessment system fusing multiple machine learning models is characterized by comprising:
the data cleaning module is used for processing missing data, combining and serializing characteristics and segmenting according to different data characteristics;
the model target definition module is used for completing the definition of the model target based on the comprehensive performance of the user in dimensions such as income, operation cost, risk cost, transaction performance, product purchase performance and the like;
the sub-model creation module is used for creating five sub-models by training five sub-model data based on model targets by utilizing sample data, wherein the sub-models comprise identity traits, performance capability, behavior preference, credit history and social attributes;
the submodel training module is used for carrying out weighted fusion on the basis of the pearson correlation coefficients of the submodels and the model target, establishing a credit investigation comprehensive scoring model for the five created submodels, and carrying out training and prediction; and
and the standardization processing module is used for mapping the final model result scores and dividing the customers into a plurality of evaluation grades.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911081115.7A CN110956273A (en) | 2019-11-07 | 2019-11-07 | Credit scoring method and system integrating multiple machine learning models |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911081115.7A CN110956273A (en) | 2019-11-07 | 2019-11-07 | Credit scoring method and system integrating multiple machine learning models |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110956273A true CN110956273A (en) | 2020-04-03 |
Family
ID=69976680
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911081115.7A Pending CN110956273A (en) | 2019-11-07 | 2019-11-07 | Credit scoring method and system integrating multiple machine learning models |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110956273A (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111415247A (en) * | 2020-04-25 | 2020-07-14 | 中信银行股份有限公司 | Post-credit risk evaluation method and device, storage medium and electronic equipment |
| CN111798303A (en) * | 2020-07-06 | 2020-10-20 | 浙江公共安全技术研究院有限公司 | Method for assessing fulfillment ability of court executives |
| CN111813399A (en) * | 2020-07-23 | 2020-10-23 | 平安医疗健康管理股份有限公司 | Machine learning-based auditing rule processing method and device and computer equipment |
| CN111967521A (en) * | 2020-08-18 | 2020-11-20 | 中国银行股份有限公司 | Cross-border active user identification method and device |
| CN112102074A (en) * | 2020-10-14 | 2020-12-18 | 深圳前海弘犀智能科技有限公司 | Grading card modeling method |
| CN112258312A (en) * | 2020-10-16 | 2021-01-22 | 银联商务股份有限公司 | Personal credit scoring method and system, electronic device and storage medium |
| CN112017040B (en) * | 2020-10-16 | 2021-01-29 | 银联商务股份有限公司 | Credit scoring model training method, scoring system, equipment and medium |
| CN112434065A (en) * | 2020-11-20 | 2021-03-02 | 上海孚厘科技有限公司 | Prediction processing method based on bank flow interval data |
| CN112465624A (en) * | 2020-11-02 | 2021-03-09 | 中国测绘科学研究院 | Personal credit evaluation method based on social media state |
| CN112561320A (en) * | 2020-12-14 | 2021-03-26 | 中国建设银行股份有限公司 | Training method of mechanism risk prediction model, mechanism risk prediction method and device |
| CN112862593A (en) * | 2021-01-28 | 2021-05-28 | 深圳前海微众银行股份有限公司 | Credit scoring card model training method, device, system and computer storage medium |
| CN113537666A (en) * | 2020-04-16 | 2021-10-22 | 马上消费金融股份有限公司 | Evaluation model training method, evaluation and business auditing method, device and equipment |
| CN113538131A (en) * | 2021-07-23 | 2021-10-22 | 中信银行股份有限公司 | Method and device for modeling modular scoring card, storage medium and electronic equipment |
| CN113657993A (en) * | 2021-08-19 | 2021-11-16 | 中国平安财产保险股份有限公司 | Credit risk identification method, device, equipment and storage medium |
| CN113724061A (en) * | 2021-08-18 | 2021-11-30 | 杭州信雅达泛泰科技有限公司 | Consumer financial product credit scoring method and device based on customer grouping |
| CN116416056A (en) * | 2023-04-04 | 2023-07-11 | 深圳征信服务有限公司 | Credit data processing method and system based on machine learning |
| CN112862593B (en) * | 2021-01-28 | 2024-05-03 | 深圳前海微众银行股份有限公司 | Credit scoring card model training method, device and system and computer storage medium |
-
2019
- 2019-11-07 CN CN201911081115.7A patent/CN110956273A/en active Pending
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113537666B (en) * | 2020-04-16 | 2024-05-03 | 马上消费金融股份有限公司 | Evaluation model training method, evaluation and business auditing method, device and equipment |
| CN113537666A (en) * | 2020-04-16 | 2021-10-22 | 马上消费金融股份有限公司 | Evaluation model training method, evaluation and business auditing method, device and equipment |
| CN111415247A (en) * | 2020-04-25 | 2020-07-14 | 中信银行股份有限公司 | Post-credit risk evaluation method and device, storage medium and electronic equipment |
| CN111798303A (en) * | 2020-07-06 | 2020-10-20 | 浙江公共安全技术研究院有限公司 | Method for assessing fulfillment ability of court executives |
| CN111813399A (en) * | 2020-07-23 | 2020-10-23 | 平安医疗健康管理股份有限公司 | Machine learning-based auditing rule processing method and device and computer equipment |
| CN111967521A (en) * | 2020-08-18 | 2020-11-20 | 中国银行股份有限公司 | Cross-border active user identification method and device |
| CN111967521B (en) * | 2020-08-18 | 2023-09-19 | 中国银行股份有限公司 | Cross-border active user identification method and device |
| CN112102074A (en) * | 2020-10-14 | 2020-12-18 | 深圳前海弘犀智能科技有限公司 | Grading card modeling method |
| CN112102074B (en) * | 2020-10-14 | 2024-01-30 | 深圳前海弘犀智能科技有限公司 | Score card modeling method |
| CN112017040B (en) * | 2020-10-16 | 2021-01-29 | 银联商务股份有限公司 | Credit scoring model training method, scoring system, equipment and medium |
| CN112258312A (en) * | 2020-10-16 | 2021-01-22 | 银联商务股份有限公司 | Personal credit scoring method and system, electronic device and storage medium |
| CN112465624A (en) * | 2020-11-02 | 2021-03-09 | 中国测绘科学研究院 | Personal credit evaluation method based on social media state |
| CN112434065A (en) * | 2020-11-20 | 2021-03-02 | 上海孚厘科技有限公司 | Prediction processing method based on bank flow interval data |
| CN112561320A (en) * | 2020-12-14 | 2021-03-26 | 中国建设银行股份有限公司 | Training method of mechanism risk prediction model, mechanism risk prediction method and device |
| CN112862593A (en) * | 2021-01-28 | 2021-05-28 | 深圳前海微众银行股份有限公司 | Credit scoring card model training method, device, system and computer storage medium |
| CN112862593B (en) * | 2021-01-28 | 2024-05-03 | 深圳前海微众银行股份有限公司 | Credit scoring card model training method, device and system and computer storage medium |
| CN113538131A (en) * | 2021-07-23 | 2021-10-22 | 中信银行股份有限公司 | Method and device for modeling modular scoring card, storage medium and electronic equipment |
| CN113724061A (en) * | 2021-08-18 | 2021-11-30 | 杭州信雅达泛泰科技有限公司 | Consumer financial product credit scoring method and device based on customer grouping |
| CN113657993A (en) * | 2021-08-19 | 2021-11-16 | 中国平安财产保险股份有限公司 | Credit risk identification method, device, equipment and storage medium |
| CN116416056A (en) * | 2023-04-04 | 2023-07-11 | 深圳征信服务有限公司 | Credit data processing method and system based on machine learning |
| CN116416056B (en) * | 2023-04-04 | 2023-10-03 | 深圳征信服务有限公司 | Credit data processing method and system based on machine learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110956273A (en) | Credit scoring method and system integrating multiple machine learning models | |
| Dahooie et al. | A novel dynamic credit risk evaluation method using data envelopment analysis with common weights and combination of multi-attribute decision-making methods | |
| US7930242B2 (en) | Methods and systems for multi-credit reporting agency data modeling | |
| Chen et al. | Predicting default risk on peer-to-peer lending imbalanced datasets | |
| CN111861698B (en) | Pre-loan approval early warning method and system based on loan multi-head data | |
| AU2020101475A4 (en) | A Financial Data Analysis Method Based on Machine Learning Models | |
| CN112561598A (en) | Customer loss prediction and retrieval method and system based on customer portrait | |
| CN107609771A (en) | A kind of supplier's value assessment method | |
| Eddy et al. | Credit scoring models: Techniques and issues | |
| Mehta et al. | Employee attrition system using tree based ensemble method | |
| CN114997916A (en) | Prediction method, system, electronic device and storage medium of potential user | |
| CN112949954A (en) | Method for establishing financial fraud recognition model based on recognition learning | |
| Yuping et al. | New methods of customer segmentation and individual credit evaluation based on machine learning | |
| Zhu et al. | Loan default prediction based on convolutional neural network and LightGBM | |
| Najadat et al. | Performance evaluation of industrial firms using DEA and DECORATE ensemble method. | |
| Wu et al. | Customer churn prediction for commercial banks using customer-value-weighted machine learning models | |
| CN115204457A (en) | Loan default risk prediction method based on graph attention network | |
| CN115187312A (en) | Customer loss prediction method and system based on deep learning | |
| CN114529063A (en) | Financial field data prediction method, device and medium based on machine learning | |
| Dattachaudhuri et al. | Transparent neural based expert system for credit risk (TNESCR): an automated credit risk evaluation system | |
| US11004156B2 (en) | Method and system for predicting and indexing probability of financial stress | |
| WO1992017853A2 (en) | Direct data base analysis, forecasting and diagnosis method | |
| CN113034264A (en) | Method and device for establishing customer loss early warning model, terminal equipment and medium | |
| Lee et al. | Application of machine learning in credit risk scorecard | |
| Setnes et al. | Fuzzy target selection in direct marketing |
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 |