クラスタリングにより選定したサンプリングポイントの需要データを用いた商品マーケットシェア推定の一方法

抄録

Demand forecasting is a critical success factor for effective supply chain systems. Many demand forecasting methods based on time-series data have been investigated. However these methods can not consider the on-going market change because of its usage of past time series. In this paper, a new method of estimating the product market share based on real time demand sensing data is proposed. This method consists of the two parts, which are (1) sampling point selection using clustering analysis and (2) estimating product market share using sampled demand data. Numerical experiments show the effectiveness of the proposed method and recent network technology makes it easy to implement this method in real demand forecasting systems.