Scholarlink

<p>Organizations depend on accurate enterprise financial risk predictions because these predictions help maintain stability during periods of market instability and environmental uncertainty which organizations experience. The existing statistical methods together with single-model machine learning techniques face challenges when they try to model financial interactions that display nonlinear relationships and operate in high-dimensional spaces. The study introduces FinRiskNet as a solution to existing problems through its hybrid ensemble learning framework which combines Artificial Neural Networks and Support Vector Machines and Logistic Regression. The proposed framework includes a complete preprocessing system which uses Min–Max normalization, label encoding, SMOTE-based class balancing, median imputation, and Fourier Transform frequency-domain feature extraction. The model conducts an assessment using a financial risk evaluation dataset which contains demographic information and financial data and behavioural characteristics. The experimental results show that FinRiskNet performs better in classification tasks than traditional machine learning systems because it can generalize well and uses SHAP-based analysis for understanding model predictions. The proposed framework provides an enterprise financial risk management solution which delivers high-accuracy results while maintaining explainable operations and scalable capabilities.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enterprise financial risk prediction model based on machine learning algorithm

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月3日

查看全文
  • Jie Yang,
  • Yan

摘要

Organizations depend on accurate enterprise financial risk predictions because these predictions help maintain stability during periods of market instability and environmental uncertainty which organizations experience. The existing statistical methods together with single-model machine learning techniques face challenges when they try to model financial interactions that display nonlinear relationships and operate in high-dimensional spaces. The study introduces FinRiskNet as a solution to existing problems through its hybrid ensemble learning framework which combines Artificial Neural Networks and Support Vector Machines and Logistic Regression. The proposed framework includes a complete preprocessing system which uses Min–Max normalization, label encoding, SMOTE-based class balancing, median imputation, and Fourier Transform frequency-domain feature extraction. The model conducts an assessment using a financial risk evaluation dataset which contains demographic information and financial data and behavioural characteristics. The experimental results show that FinRiskNet performs better in classification tasks than traditional machine learning systems because it can generalize well and uses SHAP-based analysis for understanding model predictions. The proposed framework provides an enterprise financial risk management solution which delivers high-accuracy results while maintaining explainable operations and scalable capabilities.