Predicting Inventory Shipments of Oracle EBS Systems

摘要

Predicting inventory shipments can be useful for lean inventory management such as inventory planning. In this paper, we propose approaches to predict inventory shipments based on the data extracted from the inventory management module of Oracle EBS systems of a GPS-manufacturing company. First, we introduce the process to extract the inventory shipment data from the Oracle EBS system. Then, we adopt time series forecasting algorithms (i.e., ARMA) and Primitive KNN algorithms to predict the future shipments for a group of inventory items. At last, by discovering the patterns of parameter settings when optimal prediction accuracy is achieved, we develop new algorithms to reduce runtime at different levels with trade-offs in prediction accuracy.