With the inevitable trend of grid-connected operation of a high proportion of new energy sources in new generation power system, the actual operation of power systems is highly stochastic and uncertain. Under such circumstances, traditional methods for identifying weak branches in power systems are difficult to be applied in actual operation due to the problems of calculation speed and calculation accuracy, which promote the application of data-driven methods for weak branch identification in power systems. In this paper, a decision tree based method for weak branch identification is proposed, which takes into account the algorithm interpretability, and simulations are conducted in the IEEE39 system to authenticate the effectiveness of the proposed method.