In order to reduce the financial risks existing in enterprises, this paper studies the risk identification of enterprise accounting information fraud with weighted key algorithm. Based on the study of fuzzy association rules, this study analyzes the financial risks of enterprises, puts forward an analysis model of enterprise financial risks based on fuzzy association rules mining algorithm, and makes a qualitative analysis of enterprise financial crises, so as to predict enterprise financial risks and provide reference for enterprises to avoid risks. Comparing the correlation algorithm with the linear regression model, the result shows that the prediction error rate of the linear regression model is higher than that of the correlation algorithm model. Class I misjudgment rate is 40% higher, class II misjudgment rate is flat, overall misjudgment rate is 20% higher, and accuracy linear regression is 20% lower than correlation algorithm. It can be seen that the misjudgment rate of the correlation algorithm model is low, the prediction accuracy is high, and the discrimination and prediction ability of the linear regression model is poor, which proves the effectiveness of the application of the correlation algorithm in this paper.