This research focuses on a new method to evaluate the financial credit risk of manufacturing enterprises in a heterogeneous data environment, taking Chinese manufacturing enterprises as an example. This paper uses principal component analysis and K-means clustering technology to quantitatively score and grades the credit status of enterprises and systematically evaluates the relative importance of each credit index. To solve the challenge caused by the imbalance of data categories, the study introduced the SMOTE oversampling method to optimize the prediction accuracy of the machine learning model. After comparing the performance of various machine learning models, the best-performing model is selected as the core tool of stress testing, which is used to evaluate the coping ability of enterprises in various industries under different stress scenarios. The results show that: (1) there are significant differences in the effects of various credit indicators on the financial credit risk of manufacturing enterprises. The solvency of the industry is the most critical risk factor, while the impact of enterprise operation ability is relatively small; (2) Under the test condition of increasing pressure step by step, the MLP model shows the optimal stability and prediction accuracy, which is better than other models; (3) Enterprises in different subsectors show obvious differences in anti-pressure ability when facing external pressure. General equipment manufacturing enterprises have strong anti-pressure ability, while special equipment manufacturing enterprises show greater vulnerability. The results of this study have improved the understanding of risk assessment mechanisms, expanded the scope of application of methods, and made a significant contribution to the theoretical basis of financial credit risk assessment, which has greatly contributed to the development of this area. These discoveries have led to the creation of several innovative strategies designed to effectively manage credit risks within model companies, which provide stakeholders with valuable insights into the complexity of navigation solutions in a multifaceted environment. Using these ideas, stakeholders can make more accurate and wise decisions, effectively reducing the risk of exposure and contributing to the sustainable growth and stability of enterprises. This doubling of theoretical understanding and practical application enhances the overall effectiveness of credit risk management practices, strengthening a strong and adaptive strategy to ensure that the subtle and ever-changing challenges of today's financial landscape are specifically designed.

Yunpeng Zhao. Research on Financial Credit Risk of Manufacturing Enterprises under Heterogeneous Data Based on Machine Learning. Academic Journal of Business & Management (2024) Vol. 6, Issue 11: 214-218. https://doi.org/10.25236/AJBM.2024.061131.

[1] Wang F, Hou Z, Juan R, et al. Research on Influencing Factors and Countermeasures of Risk Control of State-Owned Enterprises under the Background of Big Data [J]. Proceedings of Business and Economic Studies, 2023, 6(6): 128-133.

[2] Peng C, Da F, Wu J, et al. China's local government debt risk assessment and countermeasures under the influence of COVID-19[J]. Procedia computer science, 2022, 199: 354-360. DOI:10.1016/j. procs. 2022.01.043.

[3] Fan T R, Liu M S, Tong K Z, et al. Research of interoperation model for heterogeneous database based on Internet[C]. International Conference on Machine Learning & Cybernetics, 2003. DOI:10. 1109/ ICMLC.2003.1259838.

[4] Kamm S, Veekati S S, Mueller T,et al. A survey on machine learning based analysis of heterogeneous data in industrial automation[J]. Computers in Industry, 2023.

[5] Chen H, Ma K, & Shen J. Interpretable Machine Learning Facilitates Disease Prognosis: Applications on COVID-19 and Onward[J]. International Journal of Computer Science and Information Technology, 2024, 3(3), 428-436.

[6] Chen H, Varatharajah Y, De Ramirez S S, Arnold P, Frankenberger C, Hota B, & Iyer R. A retrospective longitudinal study of COVID-19 as seen by a large urban hospital in Chicago[J]. MedRxiv, 2020, 11.