데이터 구조와 수요예측 정확도와의 상관관계

초록

Purpose This paper analyzes that the correlation between the hierarchy of the entity’s product sales data and the accuracy of demand forecasting was studied to reduce the entity’s demand forecast error. Because each hierarchy of sales data has different characteristics, we propose a new approach for hierarchical forecasting, which represents that each tier characteristic should be considered. Design/Methodology/Approach Based on sales data from consumer goods A company and the hierarchy in ERP, the accuracy of hierarchical forecasting such as Top-down, Bottom-up and Middle-out approaches was compared, and the accuracy of time series analysis methods such as Exponential Smoothing and ARIMA was also done. Findings In order to estimate the accuracy of the proposed hierarchy, the simulation study is performed. Results demonstrate that when comparing existing and optimal hierarchy, “optimal” middle-out approach significantly outperforms existing bottom-up one. Beside comparison with hierarchical forecasting, ARIMA model shows the lowest forecast error. Thus, we reveal that the optimal hierarchy taking into account the characteristics of each level could improve the accuracy of demand forecasting over the existing hierarchy. Research Implications This study reviews the existing studies on identifying factors that intensifies forecasting inaccuracy and the adopted hierarchical approach method. Then we suggest a forecasting model considering the data with hierarchical features and structure.

키워드