Abstract: In order to solve the problem of liquidation prediction of un-arrived freight invoice carrier system, the forecasting algorithm based on the k-Nearest Neighbor(k-NN) algorithm model was studied. Hadoop technology was applied to construct a liquidation prediction platform for un-arrived freight invoice based on big data. The practice shows that the platform can make the service department master the freight operation situation of the whole railway in time, and clarify the service performance among freight carrier enterprises. It is an important part of the railway freight transportation carrier system liquidation.