Extreme Event Forecasting for Postal Logistics Service
Eunhye Kim, Hoon Jung · 2021
Forecasting demand is one of the main challenges in supply chain management. Accurate demand prediction plays a vital role in achieving operational optimization for logistical resources. Especially, in special periods when the demand extremely increases compared to normal, it becomes more important to establish the forecasting-based operation plan for logistics service reliability. This study addresses a prediction problem of postal parcel that arises at the logistics infrastructure of Korea Post. The main purpose of this paper is to develop an extreme event forecasting model for postal parcel logistics based on feature engineering and ensemble method. The proposed scheme consists of three main phases. The first phase is to analyze the characteristics of the postal parcel volume and generate the internal and external factor-based features. The second phase is to develop the internal and external ensemble predictive models. The third phase is to construct the hybrid model for extreme event prediction. The experiment with data supplied by Korea Post demonstrates the advantage in terms of prediction performance compared with other methods.