Abstract: The process of enterprise digital transformation is the reconstruction of production modes, business models, and industrial forms, as well as the promotion of enterprise core values. How to better implement digital transformation and enable high-quality development in traditional manufacturing industries has always been a challenging issue in both practice and theoretical research of Chinese manufacturing enterprises' transformation. Based on theoretical review and practical induction, this research classifies feasible paths of enterprise digital transformation into three orientations: technological production, business models, and organizational structure. Using text mining methods, this research classifies Chinese manufacturing enterprises.Using the entropy weight-TOPSIS comprehensive evaluation method, this research evaluates the performance increments brought by different transformation paths, and explores the impact of different transformation paths on comprehensive enterprise performance based on the Bayesian linear regression model. The research findings show that the digital transformation path most favored by sample manufacturing enterprises is business model-oriented; however, the technological production-oriented digital transformation path has the most significant positive impact on comprehensive enterprise performance growth, being the source of long-term competitive advantage for manufacturing enterprises. This research expands the technical methods and theoretical achievements in the field of digital transformation research, providing theoretical support and path references for the digital transformation and upgrading of manufacturing enterprises in China.