Financial risk control method based on big data
By collecting data from social network platforms and combining complex network analysis technology to establish a risk estimate model, the problem of insufficient supervision capabilities of financial risk control systems for fewer financial activities is solved, and more comprehensive risk identification and intelligent control are achieved. Reduces operating costs and enhances decision support.
Patent Information
- Application Number
- CN202411119194.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-15
- Publication Date
- 2024-11-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing financial risk control big data system mainly relies on financial data, and its regulatory capabilities are weak and insufficient comprehensiveness is insufficient for users with less financial activities.
By collecting user data from social network platforms, performing data preprocessing and graph database modeling, using complex network analysis technology to identify key nodes and community structures, combining financial risk indicators, using multiple linear regression algorithms to establish a risk estimate model, and establishing an early warning mechanism , monitor user behavior in real time to warn of potential risks.
It improves the comprehensiveness and intelligence of risk identification, realizes real-time risk monitoring and early warning, reduces operating costs, and enhances decision-making support, providing financial institutions with more efficient and accurate risk control solutions.
Smart Images

Figure CN118967305A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of financial risk control, and in particular to a financial risk control method based on big data. Background Art
[0002] In financial credit business, the repayment ability of high-risk customers and the identification of fraudulent customers are the main risk points. Therefore, a risk control decision engine is needed for the risk control link of financial credit business. The principle is to derive basic indicators into variables, rules, and rule sets, and combine them into strategies to conduct risk control audits and output risk control conclusions.
[0003] With the development of big data technology, financial risk control based on big data technology has developed rapidly. Big data technology can effectively improve the accuracy of financial data risk control, and the risk control process is efficient and highly intelligent.
[0004] However, the current financial risk control big data system mainly relies on financial data for risk control management. For those users with less financial activities, its supervision ability is relatively weak and its comprehensiveness is insufficient. To this end, we propose a financial risk control method based on big data. Summary of the invention
[0005] The purpose of the present invention is to solve the problem that the current financial risk control big data system mainly relies on financial data to carry out risk control management, and its supervision ability is relatively weak for users with less financial activities, and its comprehensiveness is insufficient. The present invention provides a financial risk control method based on big data.
[0006] In order to achieve the above-mentioned purpose, the present invention specifically adopts the following technical solutions:
[0007] A financial risk control method based on big data includes the following steps:
[0008] S1, data collection, collects user data from social network platforms;
[0009] S2, data preprocessing, cleaning the collected data, removing irrelevant information, and processing missing values and outliers. Then, the user data is mapped to nodes and edges in the graph database to ensure that the data is converted into the format required by the graph database;
[0010] S3. Establish a graph database, select a graph database system, and import the preprocessed data into the graph database. In the graph database, each user can be represented as a node, and the relationship between users is represented as an edge between nodes.
[0011] S4, risk analysis, using complex network analysis techniques to analyze the network structure in the graph database, calculate the centrality index of the network, and identify key nodes and potential community structures;
[0012] S5. Establishment of risk prediction model: Based on the results of complex network analysis and combined with financial risk indicators (such as credit scores, trading behaviors, historical default records, etc.), a multivariate linear regression algorithm is used to establish a financial risk prediction model. In the model, financial risk indicators are used as dependent variables, while key nodes and community structure indicators are used as independent variables.
[0013] S6. Establish an early warning mechanism. Establish an early warning mechanism based on the risk prediction model. When a change in network structure or user behavior pattern is detected, the risk prediction model is used to calculate the changed network structure change or user behavior pattern and output the corresponding predicted financial risk index. When the financial risk index is within the risk range, the early warning system can issue a risk warning in a timely manner.
[0014] S7. Early warning implementation: Integrate the early warning mechanism into the financial risk management platform to monitor social network data and user behavior in real time. Once the early warning conditions are triggered, the system automatically notifies relevant personnel to take measures.
[0015] Furthermore, the user data includes basic user information and user behavior data, the user behavior data includes friendships and interactive behaviors between users, and the interactive behaviors include commenting, forwarding, liking, sharing and publishing content.
[0016] Furthermore, the graph database system selects ArangoDB, which can establish a multi-model database for distributed graph data.
[0017] Furthermore, the centrality indicators include degree centrality, closeness centrality and betweenness centrality, and the calculation of the centrality indicators of the network includes counting the number of connections of each node (i.e., the number of edges) to calculate the degree centrality, measuring the average shortest path length from a node to all other nodes to calculate the closeness centrality, and measuring the frequency of a node appearing in all shortest paths to calculate the betweenness centrality, and the calculation of the centrality indicators of the network is calculated using the NetworkX graph analysis library provided by the graph database.
[0018] Furthermore, the identification of key nodes and potential community structures includes the following steps:
[0019] A1. Identification of key nodes. Identify key nodes based on the ranking of centrality indicators (degree centrality, closeness centrality, and betweenness centrality).
[0020] A2. Community structure discovery: Use the Louvain community detection algorithm to identify the community structure in the network and find densely connected node groups in the network.
[0021] Furthermore, the use of a multivariate linear regression algorithm to establish a financial risk prediction model includes the following steps:
[0022] B1. Data collection: collect user historical data, including credit scores, transaction behaviors, historical default records, and key node and community structure data obtained through complex network analysis;
[0023] B2. Data preprocessing: clean data, process missing values and outliers, standardize or normalize data, and ensure data quality;
[0024] B3. Feature selection: Select one or more features related to financial risk from the collected data, including key node indicators and community structure indicators in network analysis;
[0025] B4. Model construction: Use the multivariate linear regression algorithm to build a model, in which credit score, transaction behavior, and historical default records are used as dependent variables, while key nodes and community structure indicators are used as independent variables;
[0026] B5. Model training: use historical data to train the model and estimate the model parameters, i.e., regression coefficients;
[0027] B6. Model evaluation: the performance of the model was evaluated by cross validation, R-squared value, adjusted R-squared value, mean square error (MSE) and root mean square error (RMSE) statistical indicators;
[0028] B7. Model optimization: adjust model parameters according to the evaluation results, add or delete features, and use LASSO function regularization to prevent overfitting;
[0029] B8. Model deployment: deploy the optimized model into the financial risk control system.
[0030] Furthermore, the LASSO function expression is as follows:
[0031] J(θ)=1 / 2n(Xθ-Y) T (Xθ-Y)+α||θ||1
[0032] Where n is the number of samples, α is a constant coefficient, and ||θ||1 is the L1 norm.
[0033] The beneficial effects of the present invention are as follows:
[0034] 1. The present invention improves the comprehensiveness of risk identification by combining social network data with financial data, and introduces complex network analysis technology to improve the intelligent level of risk control. At the same time, it realizes real-time risk monitoring and early warning, reduces operating costs, and enhances decision support. In general, the present invention provides a more efficient and accurate risk control solution for financial institutions. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 It is the financial risk control flow chart of the present invention;
[0036] Figure 2 It is a flow chart of identifying key nodes and potential community structures of the present invention;
[0037] Figure 3 It is a flow chart of the present invention using a multivariate linear regression algorithm to establish a financial risk prediction model. DETAILED DESCRIPTION
[0038] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.
[0039] See also Figure 1 The present invention provides a financial risk control method based on big data, comprising the following steps:
[0040] S1, data collection, collects user data from social network platforms;
[0041] S2, data preprocessing, cleaning the collected data, removing irrelevant information, and processing missing values and outliers. Then, the user data is mapped to nodes and edges in the graph database to ensure that the data is converted into the format required by the graph database;
[0042] S3. Establish a graph database, select a graph database system, and import the preprocessed data into the graph database. In the graph database, each user can be represented as a node, and the relationship between users is represented as an edge between nodes.
[0043] S4, risk analysis, using complex network analysis techniques to analyze the network structure in the graph database, calculate the centrality index of the network, and identify key nodes and potential community structures;
[0044] S5. Risk estimation model is established based on the results of complex network analysis, combined with financial risk indicators (such as credit scores, trading behaviors, historical default records, etc.). Financial risk indicators obtain data from multiple channels such as internal systems of financial institutions, public data sets, and third-party service providers. A multivariate linear regression algorithm is used to establish a financial risk estimation model. In the model, financial risk indicators are used as dependent variables, while key nodes and community structure indicators are used as independent variables.
[0045] S6. Establish an early warning mechanism. Establish an early warning mechanism based on the risk prediction model. When a change in network structure or user behavior pattern is detected, the risk prediction model is used to calculate the changed network structure change or user behavior pattern and output the corresponding predicted financial risk index. When the financial risk index is within the risk range, the early warning system can issue a risk warning in a timely manner; the risk range includes credit score reduction to bad and transaction default, etc.
[0046] S7. Early warning implementation: Integrate the early warning mechanism into the financial risk management platform to monitor social network data and user behavior in real time. Once the early warning conditions are triggered, the system automatically notifies relevant personnel to take measures.
[0047] The present invention has the following advantages when used:
[0048] 1. Improve the comprehensiveness of risk identification: By combining social network data and financial data, the present invention can conduct a more comprehensive risk assessment on users with less financial activities, thus making up for the shortcomings of traditional financial risk control systems.
[0049] 2. Improve the intelligence level of risk control: Based on complex network analysis technology, the present invention can identify key nodes and potential community structures, thereby more accurately estimating financial risks and improving the intelligence level of risk control.
[0050] 3. Real-time risk monitoring and early warning: The present invention integrates the early warning mechanism into the financial risk management platform, which can monitor social network data and user behavior in real time, promptly discover and warn of potential risks, and effectively reduce the probability of financial risk events.
[0051] 4. Reduce operating costs: Through automated early warning mechanisms, financial institutions can reduce the need for manual monitoring, reduce operating costs, and improve the efficiency of risk control.
[0052] 5. Enhanced decision support: The risk prediction model of the present invention can provide financial institutions with more scientific and accurate risk assessment results, helping decision makers to formulate more reasonable credit policies and risk control strategies.
[0053] To sum up, the financial risk control method based on big data proposed in the present invention can not only improve the comprehensiveness and intelligence of risk identification, but also realize real-time risk monitoring and early warning, reduce operating costs, and enhance decision support, thereby providing financial institutions with more efficient and accurate risk control solutions.
[0054] In this embodiment, preferably, the user data includes basic user information and user behavior data, the user behavior data includes friendships and interactive behaviors between users, and the interactive behaviors include commenting, forwarding, liking, sharing and publishing content.
[0055] In this embodiment, preferably, the graph database system selects ArangoDB. ArangoDB can establish a multi-model database for distributed graph data. The ArangoDB graph database system not only allows the use of one database for all operations, but also allows you to run special queries on data stored in different models. It can minimize network hopping problems and run queries efficiently, even for distributed graph data.
[0056] In this embodiment, preferably, the centrality indicators include degree centrality, closeness centrality and betweenness centrality, and the calculation of the centrality indicators of the network includes counting the number of connections of each node (i.e., the number of edges) to calculate the degree centrality, measuring the average shortest path length from a node to all other nodes to calculate the closeness centrality, and measuring the frequency of a node appearing in all shortest paths to calculate the betweenness centrality, and the calculation of the centrality indicators of the network is calculated using the NetworkX graph analysis library provided by the graph database.
[0057] See also Figure 2 In this embodiment, preferably, the identifying key nodes and potential community structures includes the following steps:
[0058] A1. Identification of key nodes. Identify key nodes based on the ranking of centrality indicators (degree centrality, closeness centrality, and betweenness centrality).
[0059] A2. Community structure discovery: Use the Louvain community detection algorithm to identify the community structure in the network and find densely connected node groups in the network.
[0060] Python code example for identifying key nodes and potential community structures:
[0061] ```Python
[0062] import community as community_l ouva in
[0063] part iti on=community_l ouva i n.best_part iti on(G)
[0064] for ii n range(l en(part it i on)):
[0065] pr i nt("Node{}be l ongs to community{}".format(i,part iti on[i]))
[0066] ```
[0067] See also Figure 3 In this embodiment, preferably, the use of a multivariate linear regression algorithm to establish a financial risk prediction model includes the following steps:
[0068] B1. Data collection: collect user historical data, including credit scores, transaction behaviors, historical default records, and key node and community structure data obtained through complex network analysis;
[0069] B2. Data preprocessing: clean data, process missing values and outliers, standardize or normalize data, and ensure data quality;
[0070] B3. Feature selection: Select one or more features related to financial risk from the collected data, including key node indicators and community structure indicators in network analysis;
[0071] B4. Model construction: Use the multivariate linear regression algorithm to build a model, in which credit score, transaction behavior, and historical default records are used as dependent variables, while key nodes and community structure indicators are used as independent variables;
[0072] B5. Model training: use historical data to train the model and estimate the model parameters, i.e., regression coefficients;
[0073] B6. Model evaluation: the performance of the model was evaluated by cross-validation, R-squared value, adjusted R-squared value, mean square error (MSE) and root mean square error (RMSE) statistical indicators;
[0074] B7. Model optimization: adjust model parameters according to the evaluation results, add or delete features, and use LASSO function regularization to prevent overfitting;
[0075] B8. Model deployment: deploy the optimized model into the financial risk control system.
[0076] In this embodiment, preferably, the LASSO function expression is as follows:
[0077] J(θ)=1 / 2n(Xθ-Y) T (Xθ-Y)+α||θ||1
[0078] Where n is the number of samples, α is a constant coefficient, and ||θ||1 is the L1 norm.
[0079] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A financial risk control method based on big data, Features: It includes the following steps: S1, data collection, collects user data from social network platforms; S2, data preprocessing, cleaning the collected data, removing irrelevant information, processing missing values and outliers, and then mapping the user data to nodes and edges in the graph database to ensure that the data is converted into the format required by the graph database; S3. Establish a graph database, select a graph database system, and import the preprocessed data into the graph database. In the graph database, each user can be represented as a node, and the relationship between users is represented as an edge between nodes. S4, risk analysis, using complex network analysis techniques to analyze the network structure in the graph database, calculate the centrality index of the network, and identify key nodes and potential community structures; S5. Establishment of risk prediction model. Based on the results of complex network analysis and combined with financial risk indicators (credit score, transaction behavior and historical default record), a multivariate linear regression algorithm is used to establish a financial risk prediction model. In the model, financial risk indicators are used as dependent variables, while key nodes and community structure indicators are used as independent variables. S6. Establish an early warning mechanism. Establish an early warning mechanism based on the risk prediction model. When a change in network structure or user behavior pattern is detected, the risk prediction model is used to calculate the changed network structure change or user behavior pattern and output the corresponding predicted financial risk index. When the financial risk index is within the risk range, the early warning system can issue a risk warning in a timely manner. S7. Early warning implementation: Integrate the early warning mechanism into the financial risk management platform to monitor social network data and user behavior in real time. Once the early warning conditions are triggered, the system automatically notifies relevant personnel to take measures.
2. According to the financial risk control method based on big data as described in claim 1, Features: The user data includes basic user information and user behavior data. The user behavior data includes friendships and interactive behaviors between users. The interactive behaviors include commenting, forwarding, liking, sharing and publishing content.
3. According to the financial risk control method based on big data as described in claim 1, Features: The graph database system selects ArangoDB, which can establish a multi-model database for distributed graph data.
4. According to the financial risk control method based on big data as claimed in claim 1, Features: The centrality indicators include degree centrality, closeness centrality and betweenness centrality. The centrality indicators of the network are calculated by counting the number of connections of each node to calculate degree centrality, measuring the average shortest path length from a node to all other nodes to calculate closeness centrality, and measuring the frequency of a node appearing in all shortest paths to calculate betweenness centrality. The centrality indicators of the network are calculated using the NetworkX graph analysis library provided by the graph database.
5. According to the financial risk control method based on big data as described in claim 4, Features: The identification of key nodes and potential community structures includes the following steps: A1. Identification of key nodes. Identify key nodes based on the ranking of centrality indicators (degree centrality, closeness centrality, and betweenness centrality). A2. Community structure discovery: Use the Louvain community detection algorithm to identify community structures in the network and find densely connected node groups in the network.
6. According to the financial risk control method based on big data as claimed in claim 1, Features: The use of the multivariate linear regression algorithm to establish a financial risk estimation model includes the following steps: B1. Data collection: collect user historical data, including credit scores, transaction behaviors, historical default records, and key node and community structure data obtained through complex network analysis; B2. Data preprocessing: clean data, process missing values and outliers, standardize or normalize data, and ensure data quality; B3. Feature selection: Select one or more features related to financial risk from the collected data, including key node indicators and community structure indicators in network analysis; B4. Model construction: Use the multivariate linear regression algorithm to build a model, in which credit score, transaction behavior, and historical default records are used as dependent variables, while key nodes and community structure indicators are used as independent variables; B5. Model training: use historical data to train the model and estimate the model parameters, i.e., regression coefficients; B6. Model evaluation: the performance of the model was evaluated by cross-validation, R-squared value, adjusted R-squared value, mean square error (MSE) and root mean square error (RMSE) statistical indicators; B7. Model optimization: adjust model parameters according to the evaluation results, add or delete features, and use LASSO function regularization to prevent overfitting; B8. Model deployment: deploy the optimized model into the financial risk control system.
7. According to the financial risk control method based on big data as claimed in claim 1, Features: The LASSO function expression is as follows: J(θ)=1 / 2n(Xθ-Y) T (Xθ-Y)+α||θ||1 Where n is the number of samples, α is a constant coefficient, and ||θ||1 is the L1 norm.
Citation Information
Patent Citations
Financial risk identification method and device, electronic equipment and storage medium
CN117611335A
Intelligent prediction method and system for personal credit default
CN118195760A
Power financial service risk control method and system based on GRU-LSTM neural network
WO2023142424A1
Cited By
Risk control behavior early warning prediction method and system based on Al identification
CN119379427A
Financial risk early warning method and system based on big data technology
CN119624661A
Financial case fund flow traceability analysis method
CN120106956A
