"The Enterprise Performance Puzzle - Predictive Modeling Based on Interpretable Machine Learning"

Corporate performance is a mature and popular research topic, and this study focuses on the influence of internal factors on corporate performance, and constructs a systematic corporate performance prediction model through the tree model and SHAP algorithm in machine learning. Unlike previous studies that explored corporate performance in a single dimension, this study covers corporate characteristics and executive characteristics, and explores the contribution of internal factors in corporate performance prediction. We compare and analyze firm-level and executive-level characteristics to discover the importance and mode of their roles in predicting firm performance. The results show that firm-level characteristics have a more direct and significant impact than executive-level characteristics in predicting firm performance. Among them, total corporate assets play the most important role in predicting corporate performance, and this study not only overcomes the limitations of traditional measurement models, but also provides new perspectives and tools for corporate performance prediction and management practices.