关键词: 贝叶斯网络，结构学习，概率推理，成绩预测

Abstract: Students' achievement is an important basis for the evaluation of teaching quality in universities. This paper presented a construction method for main courses Bayesian networks,which use course score as data sample. After the network parameter learning, we have inference and prediction to course score. The results show that the model can reveal effectively the key factors of which affect the course score, and to provide a basis for decision-making.

Key words: Bayesian network, Structure learning, Probability inference, Achievement prediction