Credit Risk Evaluation of Science and Technology Finance Based on Artificial Intelligence and Bayesian Algorithm

Authors

  • Yang Gao School of Economics and Management, Jiangsu Vocational College of Finance and Economics, Huai ’an 223003, Jiangsu, China
  • Lei Sun School of Economics and Management, Shazhou Professional Institute of Technology, Suzhou 215600, Jiangsu, China

Abstract

Since the US subprime mortgage crisis (2007–2010), the prevention of financial systemic risks has been a top priority of all regulatory authorities. In the technology finance industry, new technologies based on big data and underpinned by artificial intelligence are infiltrating the technology finance field. Due to the objective and superior ability of AI data processing, credit risk can be predicted to a certain extent. Based on the Bayesian method, this paper discusses the risk spillover effect of the science and technology finance industry. When carrying out Bayesian quantile regression, two main tasks need to be done: first, determine the prior distribution of each parameter; second, obtain the posterior parameter distribution of samples. The
experimental results show that the maximum value of parameter Alpha1 reached 0.02762 at 75%. The maximum value of parameter Alpha2 reached 0.3031 at 75%, and the value of parameter Alpha2 was larger than that of parameter Alpha1 on the whole. The posterior simulation method not only does not need to assume that all parameters follow the normal distribution; it can also correct them during simulation. The use of artificial intelligence to analyze any changes of debt yield helps to give a comprehensive indication of the overall risk. In addition to the analysis of volatility, it can also more accurately predict the probability of default.

Keywords: Technology finance credit, risk evaluation, artificial intelligence, Bayesian algorithm

Cite As

Y. Gao, L. Sun, "Credit Risk Evaluation of Science and Technology Finance Based on Artificial Intelligence
and Bayesian Algorithm", Engineering Intelligent Systems, vol. 32 no. 5, pp. 445-455, 2024.






Published

2024-09-01