Proposed in this paper is an idea of analyzing the correlation degree between static loss factors and line loss rate through gray relation（GR） with variable weights,and a GM（1,N） model for predicting the line loss rate.In this method,first,the impact weights of various factors on the line loss are determined based on the loss mechanism.Then,the relationships between various factors and the line loss rate are compared via the GR method.Finally,strongly-correlated factors are picked to establish a loss-rate model for the prediction of distribution line loss rate and loss reduction potential due to relevant factor optimization.Case analyses show that both the GR analysis and the prediction is of high accuracy,and that the proposed method requires fewer samples and less computation,so that it provides a useful guidance for the energy-saving evaluation and planning of power network.