An Optimal Multi-variable Grey Model for Logistics Demand Forecast

The grey system theory, which has been extensively used in many areas, is appropriate for forecasting. It is necessary to put forward new models or algorithms to improve its performance, especially for forecast accuracy. In the forecast process of grey model, the size of data sample and the number of variables can affect forecast accuracy. In this paper, we first put forward a new method to choose optimal forecast variable number and data sample size for multi-variable grey model. Then we establish an optimal multivariable grey model, in which the goal function is the minimum fitting relative error; and one constraint is data sample constraint, and the other is variable number constraint. Finally, we give the algorithm. Case studies of logistics demand forecast indicate that the model can solve the problem of factor choice and data sample size determination with high accuracy, and can fully utilize the sample information.