Adaptive Grey Prediction Model with Application to Demand Forecasting of Chinese Logistics Industry

摘要

Accurate prediction from small data is a fundamental yet challenging problem for market planning and management analysis. In this field, grey models provide a new class of short-term predictions from small-scale data and have been successfully applied on various applications. This paper introduces a novel adaptive grey model, namely AGM(1, 1, 2), aiming to improve forecast accuracy of time critical problem that are always faced in demand planning. The proposed model has three improvements: for one thing, we construct the self-adjust grey equation to adapt data moving tendency; second, a novel evolutionary algorithm is used to search the optimal solution of key parameters in the adjusted grey equation; third, the adjusted time response function is derived to produce forecasting values. Afterwards, in order to evaluate the performance of the proposed model and explain how it can be used in practice, application of Chinese logistics demand is used as illustrative example. The new approach forecast future scenario of express delivery volume in China. And the comparative study shows that the proposed model outperforms the others and is effective for short term predictions.