Performance management is core part of enterprise human resources management, for any organization or enterprise a good, the secret to its success lies in the effective performance management. Through the enterprise performance management, it can motivate employees, improve employee satisfaction, will help the staff's personal goals combined with enterprise strategy, realize the goal of the enterprise and the balance of personal development, promote enterprise realize sustainable development. In this paper, we conduct research on enterprise performance management information system development and the robustness optimization based on data regression analysis and mathematical optimization theory. We firstly review the modern regression analysis approaches for large-scale data with the introduction of our proposed revised regression model for enterprises. Later, we summarize the state-of-the-art mathematical optimization models to select the most feasible one to serve as the tool for our system. Then, we analyze the performance of the current existing performance management information system and adopt the prior knowledge to optimize the system. The experimental result proves the feasibility and effectiveness of our system. The performance evaluation accuracy is enhanced from 86.7% to 93.5% and the system robustness is enhanced as well.