This week,an article on multi-scale adaptive graph neural network for multivariable time series prediction is readed. Multivariable time series (MTS) prediction plays an important role in the automation and optimization of intelligent applications. This is a challenging task. In this paper, a multi-scale adaptive graph neural network (MAGNN) is proposed to solve the above problems. MAGNN uses multi-scale pyramid network to maintain potential time dependence on different time scales. The author also developed a scale-based fusion module, which effectively promoted the cooperation on different time scales and automatically captured the importance of the contribution time pattern. Experiments on six real data sets show that the performance of MAGNN is better than the most advanced methods in various settings.