Abstract: Traditional agricultural policy analysis pays more attention to the relationship between variables and estimation of model parameters, while “machine learning” focuses on the accuracy of predictions, which is precisely the purpose of agricultural policy research. Because “machine learning” has enormous data collection and storage capabilities, strong learning and analysis capabilities, and more intelligent language analysis capabilities, it could have a revolutionary impact on agricultural policy research. The agricultural economists should pay attention to the development of “machine learning”, introduce “machine learning” in scientific research and teaching, and combine “machine learning” with traditional agricultural economic analysis methods to make agricultural policy formulation more precise, scientific and powerful, and make policy communication more efficient.
