To reduce the impact of single indicator analysis, imbalanced weights, and data noise in traditional performance analysis models, a weighted fuzzy mutual information multi granularity performance analysis model is proposed. Firstly, the space of educational output indicators is divided into granularity layers to avoid the problem of local analysis of indicators. The application scope of the weight method has been expanded by combining the priority graph and CRITIC weighting method. At the same time, the fuzzy mutual information method is used to analyze the overall correlation between the input indicators and the granularity layer, and the importance of the input indicators is calculated and ranked based on the combination weight. Taking undergraduate education data from a certain province as an example analysis, the results verify the rationality and effectiveness of this method.