Performance evaluation is not only a reflection of the results of the enterprise’s operational performance, but also an important reference for the optimization and improvement of enterprise performance management. This paper takes enterprise performance evaluation as an entry point, describes the design idea of enterprise performance management evaluation index system, and compares and analyzes the focus and shortcomings of two kinds of performance management tools, namely, balanced scorecard and key performance indicators. The uncertainty of the evaluated object in the process of performance management work is regarded as a random variable, and entropy is used to measure it. After completing the research preparation on the mathematical definition of entropy, the performance evaluation idea of CART decision tree algorithm is discussed by taking the example of R&D project of enterprise A. The performance evaluation model based on CART decision tree algorithm is constructed through three steps of defining variables, generating decision tree and pruning decision tree. Compared with similar modeling algorithms, the evaluation accuracy of the designed performance evaluation model on different performance data sets is stable at 95.00% and above, and up to 100.00%.
