Improvement of Demand Forecasting Accuracy: A Method Based on Region-Division

The key factor to increase enterprise profits and reduce the logistic costs is scientific and reasonable logistics demand forecasting, the accuracy of which can directly influence the effect of decision-making for an enterprise. In some instance, a single customer’s demand is irregular, while the demand in a region is comparatively more stable, so it can be better forecasted. In this paper, we have proposed a method to form regions in order that the demand in these regions can be more efficient forecasted while the error of transportation cost caused by replacement of customers by regions can be controlled. Each formed region consists of adjacent customers with similar unit transportation cost to all distributors. Numerical test with data form Northeast Subsidiary Company of China National Petroleum Corporation shows that the method can improve forecast accuracy efficiently.