<p>Driven by the global rise of cloud computing technologies, cloud-related enterprises are actively pursuing technological transformations. To assist investors in understanding the internal control factors influencing these companies, this study employs annual data from 2013 to 2023.We dissect financial performance metrics through three dimensions—Shareholding, Directorate, and Management Layer—further breaking them down into 15 sub-indicators to identify the governance characteristics most likely to influence corporate performance. We find that the executive compensation, executive shareholding, the shareholding of the largest shareholder, and board shareholding emerge as the most influential factors, suggesting they may significantly impact financial performance. Methodologically, goodness-of-fit estimates for both in-sample and out-of-sample data demonstrate that non-linear algorithms outperform their linear counterparts. Among non-linear methods, the Random Forest algorithm outperforms XGBoost. Finally, the study conducts three sets of robustness tests, including rolling window analysis, reclassification of training and testing sets, substitution of indicators, and segmentation of samples into state-owned versus non-state-owned categories, as well as by three major economic regions. These robustness tests provide strong evidence for the model’s reliability in terms of mitigating overfitting and ensuring data consistency.</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

Governance Factors Influencing Financial Performance in Cloud-Based Enterprises: A Machine Learning Analysis

  • Ziling Huang,
  • Lichao Lin,
  • Xiaofei Jia

摘要

Driven by the global rise of cloud computing technologies, cloud-related enterprises are actively pursuing technological transformations. To assist investors in understanding the internal control factors influencing these companies, this study employs annual data from 2013 to 2023.We dissect financial performance metrics through three dimensions—Shareholding, Directorate, and Management Layer—further breaking them down into 15 sub-indicators to identify the governance characteristics most likely to influence corporate performance. We find that the executive compensation, executive shareholding, the shareholding of the largest shareholder, and board shareholding emerge as the most influential factors, suggesting they may significantly impact financial performance. Methodologically, goodness-of-fit estimates for both in-sample and out-of-sample data demonstrate that non-linear algorithms outperform their linear counterparts. Among non-linear methods, the Random Forest algorithm outperforms XGBoost. Finally, the study conducts three sets of robustness tests, including rolling window analysis, reclassification of training and testing sets, substitution of indicators, and segmentation of samples into state-owned versus non-state-owned categories, as well as by three major economic regions. These robustness tests provide strong evidence for the model’s reliability in terms of mitigating overfitting and ensuring data consistency.