Predicting logistics delivery demand with deep neural networks

Abstract

Delivery time affects the logistics route, depending on the needs of the place and quantity. An efficient prediction of delivery demand would help the construction of logistics model. The data on delivery demand are time-dependency and space-correlation. Modeling the multidimensional sequence or making the prediction based on it would be a computation consuming work. Our research is based on deep learning to propose an efficient procedure to predict delivery demand. With the simulation study, the prediction performance of the proposed procedure is acceptable. This is conducive to the further study of logistics decisions making.