Credit risk assessment system for small and medium-sized enterprises in production and manufacturing industry

Abstract

The invention discloses a credit risk assessment system for small and medium-sized enterprises in the production and manufacturing industry, and the system comprises a data collection and preprocessing module which is used for collecting and preprocessing data; the data enhancement module is used for training the generated adversarial network model and performing data generation and fusion; the credit risk assessment model construction and training module is used for constructing and training a credit risk assessment model; and the credit risk prediction and application module is used for credit risk prediction and result application. According to the invention, through a data enhancement technology, robust model construction and integration of multiple data sources, the accuracy, robustness and adaptability of credit risk assessment of small and medium-sized enterprises are significantly improved, and the problems of data imbalance, poor model generalization ability and the like in the prior art are effectively solved.

Description

Technical Field

[0001] The invention belongs to the technical field of risk assessment, and in particular is a credit risk assessment system for small and medium-sized enterprises in the manufacturing industry. Background Art

[0002] In the financial field, credit risk assessment is the core link of credit decision-making. Especially for small and medium-sized enterprises, due to the lack of complete financial statements and credit records, traditional credit risk assessment methods face many challenges. Existing credit risk assessment models are mainly based on the following technologies:

[0003] 1. Statistical model: Traditional credit scoring models (such as logistic regression models) rely on the statistical laws of historical data. These models are simple and easy to explain, but they often show low prediction accuracy when faced with complex nonlinear relationships and high-dimensional data.

[0004] 2. Machine learning models: including support vector machines (SVM), random forests, gradient boosted decision trees (GBDT), etc. Although these models are superior in dealing with nonlinear relationships and high-dimensional data, they still rely on a large number of high-quality data samples for training and testing. If the data samples are scarce or unbalanced, the prediction effect of these models may drop significantly.

[0005] 3. Deep learning model: With the advancement of deep learning technology, models such as neural networks (NN) have achieved great success in areas such as image recognition and natural language processing. Some studies have begun to apply deep learning technology to credit risk assessment, but deep learning models require massive data for training, and their "black box" nature makes the model less interpretable.

[0006] Although existing technologies have achieved certain results in credit risk assessment, they still have the following major shortcomings:

[0007] 1. Strong data dependence: Traditional statistical models and machine learning models show obvious performance degradation when data samples are insufficient or the data quality is not high. This is particularly prominent in the credit risk assessment of small and medium-sized enterprises, because small and medium-sized enterprises often lack sufficient historical credit data.

[0008] 2. Data imbalance: In the credit data set of small and medium-sized enterprises, there are often more good credit records and fewer default records. This data imbalance will cause the model to tend to predict the good credit of the enterprise and reduce the recognition rate of default risk.

[0009] 3. Poor model generalization ability: Existing credit risk assessment models have weak generalization ability when dealing with new data or abnormal situations, and are prone to overfitting problems, especially when faced with sparse data and highly heterogeneous data sets.

[0010] 4. Lack of innovative data enhancement methods: Although deep learning technology has performed well in many fields, in the field of credit risk assessment, how to effectively use deep learning models, especially how to perform data enhancement to improve the robustness of the model, remains a challenge. Summary of the invention

[0011] In view of the above problems, the present invention is proposed to provide a credit risk assessment system for small and medium-sized enterprises in the manufacturing industry that overcomes the above problems or at least partially solves the above problems.

[0012] To achieve the above object, the present invention adopts the following technical solutions:

[0013] A credit risk assessment system for small and medium-sized enterprises in the manufacturing industry, the system comprising:

[0014] Data collection and preprocessing module, used to collect and preprocess data;

[0015] Data augmentation module, used to train the generated adversarial network model and perform data generation and fusion;

[0016] Credit risk assessment model construction and training module, used to build and train credit risk assessment models;

[0017] Credit risk prediction and application module, used for credit risk prediction and result application.

[0018] Optionally, the data collection and preprocessing module includes a data collection unit and a data preprocessing unit.

[0019] Optionally, the data collection unit includes:

[0020] Collect historical credit data of SMEs from the databases of financial institutions;

[0021] Obtain supplementary data from external public data sources.

[0022] Optionally, the data preprocessing unit includes:

[0023] Clean the collected data to remove noisy data, duplicate data, and missing values;

[0024] Perform feature selection on the data to identify features associated with credit risk;

[0025] Categorical data is encoded and numerical data is standardized or normalized.

[0026] Optionally, the data enhancement module includes a generative adversarial network training unit and a data generation and fusion unit.

[0027] Optionally, the generative adversarial network training unit includes:

[0028] Use the preprocessed data to train a generative adversarial network model, which consists of a generator and a discriminator;

[0029] The parameters of the generator and discriminator are continuously adjusted until the generator can generate synthetic data that is very close to the real data, and the discriminator cannot effectively distinguish between the generated data and the real data.

[0030] Optionally, the data generation and fusion unit includes:

[0031] Use the trained generator to generate a large amount of synthetic data to supplement the original dataset;

[0032] The generated synthetic data is fused with the original dataset to form an enhanced dataset for subsequent model training.

[0033] Optional modules for building and training credit risk assessment models include:

[0034] A model building unit, which uses a random forest algorithm to build a decision tree model and integrates the results of these decision trees;

[0035] The model training unit uses the enhanced data set to train the random forest model, evaluates the performance of the model through cross-validation, and continuously adjusts the model parameters to optimize the prediction effect of the model;

[0036] The model evaluation unit uses the test data set to evaluate the trained random forest model and calculate the model's accuracy, recall, and F1 value.

[0037] Optional credit risk prediction and application modules include:

[0038] The risk prediction unit inputs the current financial and operational data of SMEs into the trained random forest model to generate credit risk prediction results and output the credit score and risk level of the enterprise;

[0039] In the result application unit, the forecast results are used to guide the risk management decisions of financial institutions, including determining loan amounts and formulating risk response strategies.

[0040] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0041] 1. The present invention introduces generative adversarial networks for data enhancement, generates high-quality synthetic data, and makes the data set richer and more balanced, thereby improving the training effect and prediction accuracy of the model. The present invention uses a random forest algorithm to build a credit risk assessment model, and improves the robustness of the model by integrating multiple decision trees, avoiding overfitting and improving the accuracy of the model when processing complex data.

[0042] 2. The present invention not only utilizes the internal financial and operational data of the enterprise, but also obtains supplementary information from external public data sources, further enriching the dimensions of the data and enhancing the predictive ability and reliability of the model. The present invention can predict credit risk based on real-time data and allow the model to be updated regularly to adapt to new data changes, thereby improving the dynamic and timely nature of risk management. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A flowchart of a credit risk assessment system for small and medium-sized enterprises in the manufacturing industry provided in an embodiment of the present application. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0045] See also Figure 1 This embodiment provides a credit risk assessment system for small and medium-sized enterprises in the manufacturing industry, the system comprising:

[0046] The data collection and preprocessing module is used to collect and preprocess data.

[0047] The data collection and preprocessing module includes a data collection unit and a data preprocessing unit.

[0048] The data collection unit comprises:

[0049] Collect historical credit data of small and medium-sized enterprises from the databases of financial institutions. These data include the company's financial data, operational data, historical credit records, etc.

[0050] Obtain supplementary data from external public data sources (such as government open data, industry reports, etc.) to increase the breadth and depth of the dataset.

[0051] The data preprocessing unit comprises:

[0052] The collected data is cleaned to remove noisy data, duplicate data, and missing values.

[0053] Perform feature selection on the data to determine the features related to credit risk, such as the company's debt ratio, return on assets, cash flow, etc.

[0054] The categorical data is encoded and the numerical data is standardized or normalized to ensure the effectiveness of subsequent algorithms.

[0055] Data augmentation module for training the generated adversarial network model and performing data generation and fusion.

[0056] The data enhancement module includes a generative adversarial network training unit and a data generation and fusion unit.

[0057] The generative adversarial network training unit includes:

[0058] The preprocessed data is used to train the generative adversarial network model, which consists of a generator and a discriminator. The generator generates synthetic data similar to real data by learning the distribution of input data; the discriminator is responsible for determining whether the input data is real data or generated data.

[0059] The parameters of the generator and discriminator are continuously adjusted until the generator can generate synthetic data that is very close to the real data, and the discriminator cannot effectively distinguish between the generated data and the real data.

[0060] The data generation and fusion unit includes:

[0061] Use the trained generator to generate a large amount of synthetic data to supplement the insufficient or unbalanced parts of the original dataset.

[0062] The generated synthetic data is fused with the original dataset to form an enhanced dataset for subsequent model training.

[0063] The credit risk assessment model construction and training module is used to build and train the credit risk assessment model.

[0064] The credit risk assessment model construction and training modules include:

[0065] The model building unit uses the random forest algorithm to build a decision tree model and integrates the results of these decision trees to improve the accuracy and stability of the model.

[0066] The model training unit uses the enhanced data set to train the random forest model, evaluates the performance of the model through the cross-validation method, and continuously adjusts the model parameters (such as the number of trees, maximum depth, etc.) to optimize the prediction effect of the model.

[0067] The model evaluation unit uses the test data set to evaluate the trained random forest model and calculates indicators such as the model's accuracy, recall, and F1 value to ensure that the model has good generalization capabilities.

[0068] Credit risk prediction and application module, used for credit risk prediction and result application.

[0069] Credit risk prediction and application modules include:

[0070] In actual applications after deployment, the risk prediction unit inputs the current financial and operational data of SMEs into the trained random forest model to generate credit risk prediction results, and outputs the credit score and risk level of the enterprise to help financial institutions make decisions such as loan approval and risk control.

[0071] The result application unit uses the forecast results to guide the risk management decisions of financial institutions, including determining loan amounts and formulating risk response strategies. The risk assessment system can also be regularly updated and retrained based on new data to improve the real-time and accuracy of the forecast.

[0072] This embodiment introduces a generative adversarial network for data enhancement, generates high-quality synthetic data, and makes the data set richer and more balanced, thereby improving the training effect and prediction accuracy of the model. This embodiment uses a random forest algorithm to build a credit risk assessment model, and integrates multiple decision trees to improve the robustness of the model, avoid overfitting, and improve the accuracy of the model when processing complex data.

[0073] This embodiment not only uses the internal financial and operational data of the enterprise, but also obtains supplementary information from external public data sources, further enriching the dimensions of the data and enhancing the predictive ability and reliability of the model. This embodiment can predict credit risk based on real-time data and allow the model to be updated regularly to adapt to new data changes, thereby improving the dynamic and timely nature of risk management.

[0074] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. A credit risk assessment system for small and medium-sized enterprises in the manufacturing industry, characterized in that: The system comprises: Data collection and preprocessing module, used to collect and preprocess data; Data augmentation module, used to train the generated adversarial network model and perform data generation and fusion; Credit risk assessment model construction and training module, used to build and train credit risk assessment models; Credit risk prediction and application module, used for credit risk prediction and result application.

2. A credit risk assessment system for small and medium-sized enterprises in the manufacturing industry as claimed in claim 1, characterized in that: The data collection and preprocessing module includes a data collection unit and a data preprocessing unit.

3. A credit risk assessment system for small and medium-sized manufacturing enterprises as claimed in claim 2, characterized in that: The data collection unit comprises: Collect historical credit data of SMEs from the databases of financial institutions; Obtain supplementary data from external public data sources.

4. A credit risk assessment system for small and medium-sized manufacturing enterprises as claimed in claim 2, characterized in that: The data preprocessing unit comprises: Clean the collected data to remove noisy data, duplicate data, and missing values; Perform feature selection on the data to identify features associated with credit risk; Categorical data is encoded and numerical data is standardized or normalized.

5. A credit risk assessment system for small and medium-sized manufacturing enterprises as claimed in claim 1, characterized in that: The data enhancement module includes a generative adversarial network training unit and a data generation and fusion unit.

6. A credit risk assessment system for small and medium-sized manufacturing enterprises as claimed in claim 5, characterized in that: The generative adversarial network training unit includes: Use the preprocessed data to train a generative adversarial network model, which consists of a generator and a discriminator; The parameters of the generator and discriminator are continuously adjusted until the generator can generate synthetic data that is very close to the real data, and the discriminator cannot effectively distinguish between the generated data and the real data.

7. A credit risk assessment system for small and medium-sized manufacturing enterprises as claimed in claim 6, characterized in that: The data generation and fusion unit includes: Use the trained generator to generate a large amount of synthetic data to supplement the original dataset; The generated synthetic data is fused with the original dataset to form an enhanced dataset for subsequent model training.

8. A credit risk assessment system for small and medium-sized manufacturing enterprises as claimed in claim 1, characterized in that: The credit risk assessment model construction and training modules include: A model building unit, which uses a random forest algorithm to build a decision tree model and integrates the results of these decision trees; The model training unit uses the enhanced data set to train the random forest model, evaluates the performance of the model through cross-validation, and continuously adjusts the model parameters to optimize the prediction effect of the model; The model evaluation unit uses the test data set to evaluate the trained random forest model and calculate the model's accuracy, recall, and F1 value.

9. A credit risk assessment system for small and medium-sized manufacturing enterprises as claimed in claim 1, characterized in that: Credit risk prediction and application modules include: The risk prediction unit inputs the current financial and operational data of SMEs into the trained random forest model to generate credit risk prediction results and output the credit score and risk level of the enterprise; In the result application unit, the forecast results are used to guide the risk management decisions of financial institutions, including determining loan amounts and formulating risk response strategies.

Images