Personnel credit assessment and management and control method and system based on big data

Abstract

The invention discloses a personnel credit assessment and management and control method and system based on big data, and the method comprises the steps: collecting personnel credit data, and extracting a target feature reflecting a credit risk, and obtaining an initial feature vector; obtaining a personnel credit feature vector based on the initial feature vector; calculating a target weight of each target feature in the personnel credit feature vector based on the personnel credit feature vector, and obtaining a scoring index of the credit risk; performing clustering analysis on the personnel credit data based on the scoring indexes to obtain personnel risk groups; performing risk feature distribution analysis on the personnel of each group based on the personnel risk group, and determining the credit risk level of the personnel; and determining a differentiated management and control strategy based on the credit risk level, and completing personnel credit management and control. According to the method, target features reflecting credit risks are extracted from mass personnel credit data, and a feature weight system is constructed; and meanwhile, different credit crowds are accurately divided according to the multi-dimensional feature similarity of the personnel.

Description

Technical Field

[0001] The present invention relates to the field of information processing, and in particular to a personnel credit assessment and management method and system based on big data. Background Art

[0002] In the process of personnel credit assessment and management based on big data, how to effectively classify and aggregate personnel is a key technical issue. Traditional personnel classification methods are usually based on a single dimension, such as credit rating or risk level, which makes it difficult to fully reflect the credit characteristics of personnel. Simply aggregating personnel according to the similarity of a certain dimension often leads to overly rough classification and inability to accurately identify the credit risks of different personnel. In addition, in actual credit management scenarios, different businesses have different tolerances for personnel credit risks, and the granularity requirements for classification and aggregation are also different. Therefore, there is an urgent need for a personnel classification and aggregation method that can comprehensively consider multiple credit characteristics. Summary of the invention

[0003] The present invention provides a personnel credit assessment and management method and system based on big data, which extracts target features reflecting credit risk from massive personnel credit data and constructs a feature weight system; at the same time, it accurately divides different credit groups according to the similarity of multi-dimensional features of personnel.

[0004] The technical solution adopted by the present invention is:

[0005] A personnel credit assessment and control method based on big data includes the following specific steps:

[0006] S1. Collect personnel credit data, extract target features reflecting credit risk and obtain initial feature vectors;

[0007] S2. Obtaining a personnel credit feature vector based on the initial feature vector;

[0008] S3. Calculate the target weight of each target feature in the personnel credit feature vector based on the personnel credit feature vector, and obtain the credit risk scoring index;

[0009] S4. Performing cluster analysis on personnel credit data based on the scoring index to obtain personnel risk grouping;

[0010] S5. Based on the personnel risk grouping, risk characteristic distribution analysis is performed on the personnel in each group to determine the credit risk level of the personnel;

[0011] S6. Determine a differentiated management and control strategy based on the credit risk level to complete personnel credit management and control.

[0012] Furthermore, the steps S1 to S2 specifically include:

[0013] 1) Collecting personnel credit data and preprocessing the personnel credit data to obtain a credit data set;

[0014] 2) Based on the requirements of credit risk assessment, determine the target characteristic dimensions of credit assessment;

[0015] 3) extracting multi-dimensional target features from the credit data set to obtain an initial feature vector;

[0016] 4) Performing dimensionality reduction processing on the initial feature vector to obtain the personnel credit feature vector.

[0017] Furthermore, the step S3 specifically includes:

[0018] 1) Calculate the initial weight of the target feature using the hierarchical analysis method;

[0019] 2) obtaining the experience and knowledge of experts in the field of credit assessment, and adjusting the initial weights to obtain the target weights;

[0020] 3) Sorting the target weights to obtain the scoring index.

[0021] Furthermore, the step S4 specifically includes:

[0022] 1) Building a clustering model and obtaining scoring index data, and inputting the scoring index data into the clustering model to calculate the credit similarity of the personnel;

[0023] 2) Based on the credit similarity, the personnel are divided into risk groups.

[0024] Furthermore, the step S5 specifically includes:

[0025] 1) Grouping personnel based on personnel risk groups to obtain grouping results;

[0026] 2) Count the grouping results and analyze the risk characteristic distribution of different groups to obtain the risk characteristic distribution map;

[0027] 3) Using a clustering algorithm to perform risk clustering on the risk feature distribution map to obtain a clustering result;

[0028] 4) Setting a risk threshold, dividing personnel based on the clustering results and the risk threshold to obtain the personnel's credit risk level; wherein the credit risk level includes: an acceptable risk level and a high risk level.

[0029] Furthermore, the step S6 specifically includes:

[0030] 1) For personnel with acceptable risk levels, the credit limit range is set by adopting a classification regression model to complete differentiated credit limit management for personnel with acceptable risk levels;

[0031] 2) For personnel with high risk levels, rules are extracted based on historical risk events, and the personnel with high risk levels are managed and controlled based on the extraction results;

[0032] Furthermore, the step S6 also includes: real-time monitoring of the credit data of personnel with different credit risks, and dynamically adjusting the management and control methods for different personnel.

[0033] Another technical solution adopted by the present invention is:

[0034] A personnel credit assessment and management system based on big data, including:

[0035] The data processing module is used to collect personnel credit data, extract target features reflecting credit risk to obtain an initial feature vector; and obtain a personnel credit feature vector based on the initial feature vector;

[0036] A weight calculation module is used to calculate the target weight of each target feature in the personnel credit feature vector based on the personnel credit feature vector, and obtain the credit risk scoring index;

[0037] A clustering module, which performs cluster analysis on personnel based on the scoring indicators to obtain personnel risk groups;

[0038] The grade classification module is used to analyze the risk characteristics of each person in each group based on the personnel risk grouping to determine the credit risk level of the person;

[0039] The control module is used to determine differentiated control strategies based on the credit risk level to complete personnel credit control.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1) By constructing a multi-dimensional personnel credit feature vector, the feature weights are determined by combining the analytic hierarchy process with expert experience to obtain a scoring index that comprehensively reflects credit risk;

[0042] 2) Use density clustering algorithm to automatically group people, and dynamically adjust clustering parameters according to the needs of different business scenarios to generate differentiated credit population segmentation results;

[0043] 3) Based on the classification results, analyze the risk characteristics of each group, set differentiated credit limits and control strategies, and achieve refined management;

[0044] 4) By continuously collecting credit behavior data, real-time verification and dynamic updating of the model, the adaptability and robustness of the model are improved, ensuring the accuracy and timeliness of credit population segmentation, and providing effective technical support for credit risk management. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments:

[0046] Figure 1 It is a flow chart of a personnel credit assessment and control method based on big data of the present invention;

[0047] Figure 2 It is a structural schematic diagram of a personnel credit evaluation and management system based on big data in the present invention. DETAILED DESCRIPTION

[0048] Example 1

[0049] like Figure 1 As shown, a personnel credit assessment and control method based on big data specifically includes the following steps:

[0050] S1. Collect personnel credit data, extract target features reflecting credit risk and obtain initial feature vectors;

[0051] S2. Obtaining a personnel credit feature vector based on the initial feature vector;

[0052] Specifically, 1) collecting personnel credit data and preprocessing the personnel credit data to obtain a credit data set;

[0053] 2) Based on the requirements of credit risk assessment, determine the target characteristic dimensions of credit assessment;

[0054] 3) extracting multi-dimensional target features from the credit data set to obtain an initial feature vector;

[0055] 4) Performing dimensionality reduction processing on the initial feature vector to obtain the personnel credit feature vector;

[0056] Among them, personal credit data includes bank transaction records, e-commerce platform consumption behavior and operator call records; due to the huge amount of data and various forms, it is necessary to pre-process the above collected data. In this embodiment, the pre-processing process includes using data cleaning technology, through field mapping and anomaly detection methods, to remove 5% of dirty data and noise data, and reasonably fill in missing values ​​to ensure the integrity and consistency of the data; the target feature dimensions include: personal basic information, credit history and behavioral preferences;

[0057] In this embodiment, feature engineering technology is used to start from the target feature dimension, and 100 target features reflecting credit risk are selected to form an initial feature vector; correlation analysis and variance analysis feature selection algorithms are used to select 30 target features related to credit risk in order to obtain a personnel credit feature vector;

[0058] S3. Calculate the target weight of each target feature in the personnel credit feature vector based on the personnel credit feature vector, and obtain the credit risk scoring index;

[0059] Specifically, 1) the initial weight of the target feature is calculated using the hierarchical analysis method;

[0060] 2) obtaining the experience and knowledge of experts in the field of credit assessment, and adjusting the initial weights to obtain the target weights;

[0061] 3) Sorting the target weights to obtain the scoring index;

[0062] In this embodiment, the hierarchical analysis method is used to calculate the feature weights of each target feature, and the feature vectors are calculated and consistency tests are performed by pairwise comparison of the judgment matrix to obtain the initial weights that pass the test; the experience and knowledge of 15 experts in the field of credit assessment are obtained, the initial weights are adjusted by the Delphi method, and expert opinions are collected and agreed upon through three rounds of questionnaires, and the initial weights are optimized to obtain the importance weights that comprehensively consider quantitative analysis and expert experience; the principal component analysis method is used to reduce the dimension of the target features, and through eigenvalue decomposition and importance sorting, the target features of the top N principal components with cumulative variance contribution rates reaching 95% are selected as scoring indicators;

[0063] S4. Performing cluster analysis on personnel credit data based on the scoring index to obtain personnel risk grouping;

[0064] Specifically, 1) constructing a clustering model and obtaining scoring index data, and inputting the scoring index data into the clustering model to calculate the credit similarity of the personnel;

[0065] 2) Based on the credit similarity, the personnel are divided into risk groups;

[0066] In this embodiment, personnel are divided into three risk groups according to credit similarity, namely high credit risk group, medium credit risk group and low credit risk group; each group is analyzed, and it is found that the average number of overdue payments in the high credit risk group is 3 times, and the average debt ratio is 50%; the average number of overdue payments in the medium credit risk group is 1 time, and the average debt ratio is 30%; the average number of overdue payments in the low credit risk group is 0 times, and the average debt ratio is 10%. Combining the clustering results with the personnel attribute data, it is found that the proportion of people under 30 years old in the high credit risk group is relatively high, and credit education and risk warning services can be carried out for this group; the proportion of highly educated people in the low credit risk group is relatively high, and personalized services such as financial product recommendations can be carried out for this group; in order to continuously track the changes in the credit status of personnel, the clustering model and clustering results are updated once a quarter, and the ARI index is used to evaluate the stability of the clustering results. When the ARI index is lower than 8, the model parameters are adjusted in time to ensure the accuracy and timeliness of the grouping.

[0067] S5. Based on the personnel risk grouping, risk characteristic distribution analysis is performed on the personnel in each group to determine the credit risk level of the personnel;

[0068] Specifically, 1) grouping personnel based on personnel risk grouping to obtain grouping results;

[0069] 2) Count the grouping results and analyze the risk characteristic distribution of different groups to obtain the risk characteristic distribution map;

[0070] 3) Using a clustering algorithm to perform risk clustering on the risk feature distribution map to obtain a clustering result;

[0071] 4) Setting a risk threshold, dividing personnel based on the clustering results and the risk threshold to obtain the personnel's credit risk level; wherein the credit risk level includes: an acceptable risk level and a high risk level;

[0072] In this embodiment, for people in different groups, their credit risk characteristic distribution is statistically analyzed, and a population risk characteristic distribution map is generated using data visualization technology, including overdue rate and fraud rate; based on the generated risk characteristic distribution map, the K-means clustering algorithm is used to perform risk clustering on the population, and by calculating the Euclidean distance from each sample point to the cluster center, the sample points closest to the cluster center are divided into corresponding clusters to obtain clustering results of three types of people: low risk, medium risk, and high risk; the risk tolerance threshold set by the business department is obtained as the risk threshold, such as if the overdue rate does not exceed 5%, the population risk clustering result is compared with the threshold, and the people below the threshold are at an acceptable risk level, and the people above the threshold are at a high risk level.

[0073] S6. Determine a differentiated management and control strategy based on the credit risk level to complete personnel credit management and control.

[0074] Specifically, 1) for personnel with acceptable risk levels, a classification regression model is used to set a credit limit range, thereby completing differentiated credit limit management for personnel with acceptable risk levels;

[0075] 2) For personnel with high risk levels, rules are extracted based on historical risk events, and the personnel with high risk levels are managed and controlled based on the extraction results;

[0076] In addition, in order to dynamically adjust the management and control methods for different personnel, the credit data of personnel with different credit risks are monitored in real time.

[0077] In this embodiment, the classification regression model is trained to generate a credit limit decision tree based on the historical credit performance data of personnel at that level, including income, assets and liabilities variables, with credit limit as the target variable, so as to achieve differentiated credit limit management for people of different risk levels. For example, the credit limit range of the low-risk group is 50,000 to 100,000 yuan, and the credit limit range of the medium-risk group is 10,000 to 50,000 yuan, thereby completing differentiated credit limit management for people with acceptable risk levels; a series of management and control strategies are set through the rule engine, and historical risk events are summarized and refined into rule conditions. For example, if the number of overdue payments exceeds 3 times and the amount exceeds 10,000 yuan, the credit limit freezing rule is triggered, and then the corresponding management and control measures are automatically executed through the rule engine to minimize its credit risk; differentiated credit limits and management and control strategies are applied to credit groups, and by deploying a real-time risk control system, the credit performance changes of each group are monitored. When the risk indicator exceeds the warning threshold, the credit limit and management and control strategy are adjusted in time to form a closed-loop feedback mechanism.

[0078] Example 2

[0079] like Figure 2 As shown, a personnel credit evaluation and control system based on big data includes:

[0080] The data processing module 101 is used to collect personnel credit data, extract target features reflecting credit risk to obtain an initial feature vector; and obtain a personnel credit feature vector based on the initial feature vector;

[0081] The weight calculation module 102 is used to calculate the target weight of each target feature in the personnel credit feature vector based on the personnel credit feature vector, and obtain the credit risk scoring index;

[0082] Clustering module 103, performing cluster analysis on personnel based on the scoring index to obtain personnel risk grouping;

[0083] The grade classification module 104 is used to analyze the risk characteristics of each person in each group based on the personnel risk grouping to determine the credit risk level of the person;

[0084] The control module 105 is used to determine a differentiated control strategy based on the credit risk level to complete personnel credit control.

[0085] The embodiments described above are merely descriptions of preferred implementation modes of the present invention and are not intended to limit the scope of the present invention. Without departing from the principles and essence of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the scope of protection determined by the claims of the present invention.

Claims

1. A personnel credit assessment and control method based on big data, characterized in that: The specific steps include: S1. Collect personnel credit data, extract target features reflecting credit risk and obtain initial feature vectors; S2. Obtaining a personnel credit feature vector based on the initial feature vector; S3. Calculate the target weight of each target feature in the personnel credit feature vector based on the personnel credit feature vector, and obtain the credit risk scoring index; S4. Performing cluster analysis on personnel credit data based on the scoring index to obtain personnel risk grouping; S5. Based on the personnel risk grouping, risk characteristic distribution analysis is performed on the personnel in each group to determine the credit risk level of the personnel; S6. Determine a differentiated management and control strategy based on the credit risk level to complete personnel credit management and control.

2. A personnel credit assessment and control method based on big data, characterized in that: The steps S1 to S2 specifically include: 1) Collecting personnel credit data and preprocessing the personnel credit data to obtain a credit data set; 2) Based on the requirements of credit risk assessment, determine the target characteristic dimensions of credit assessment; 3) extracting multi-dimensional target features from the credit data set to obtain an initial feature vector; 4) Performing dimensionality reduction processing on the initial feature vector to obtain the personnel credit feature vector.

3. A personnel credit assessment and control method based on big data, characterized in that: The step S3 specifically includes: 1) Calculate the initial weight of the target feature using the hierarchical analysis method; 2) obtaining the experience and knowledge of experts in the field of credit assessment, and adjusting the initial weights to obtain the target weights; 3) Sorting the target weights to obtain the scoring index.

4. A personnel credit assessment and control method based on big data, characterized in that: The step S4 specifically includes: 1) Building a clustering model and obtaining scoring index data, and inputting the scoring index data into the clustering model to calculate the credit similarity of the personnel; 2) Based on the credit similarity, the personnel are divided into risk groups.

5. A personnel credit assessment and control method based on big data, characterized in that: The step S5 specifically includes: 1) Grouping personnel based on personnel risk groups to obtain grouping results; 2) Count the grouping results and analyze the risk characteristic distribution of different groups to obtain the risk characteristic distribution map; 3) Using a clustering algorithm to perform risk clustering on the risk feature distribution map to obtain a clustering result; 4) Setting a risk threshold, dividing personnel based on the clustering results and the risk threshold to obtain the personnel's credit risk level; wherein the credit risk level includes: an acceptable risk level and a high risk level.

6. A personnel credit assessment and control method based on big data, characterized in that: The step S6 specifically includes: 1) For personnel with acceptable risk levels, the credit limit range is set by adopting a classification regression model to complete differentiated credit limit management for personnel with acceptable risk levels; 2) For personnel with high risk levels, rules are extracted based on historical risk events, and the personnel with high risk levels are managed and controlled based on the extracted results.

7. A personnel credit assessment and control method based on big data, characterized in that: The step S6 also includes: real-time monitoring of the credit data of personnel with different credit risks, and dynamically adjusting the management and control methods for different personnel.

8. A personnel credit assessment and management system based on big data, characterized in that: The system is used to implement the method according to any one of claims 1 to 7, comprising: A data processing module (101) is used to collect personnel credit data, extract target features reflecting credit risk to obtain an initial feature vector; and obtain a personnel credit feature vector based on the initial feature vector; A weight calculation module (102) is used to calculate the target weight of each target feature in the personnel credit feature vector based on the personnel credit feature vector, and obtain a credit risk scoring index; A clustering module (103) performs cluster analysis on personnel based on the scoring index to obtain personnel risk groups; A grade classification module (104) is used to analyze the risk characteristics distribution of each person in each group based on the personnel risk grouping to determine the credit risk grade of the person; The control module (105) is used to determine a differentiated control strategy based on the credit risk level to complete personnel credit control.

Images