A Weighted Additive Model for Comparative Analysis on Demand Factors among Multiple Seasonal Products: Nationwide Demand of Japanese Alcoholic Beverages

Abstract

The climate has recently been fluctuating globally. Such climate, which affects the demand for seasonal products, varies depending on geographical conditions. However, analysis models of previous studies on nationwide demand for seasonal products have not considered differences in climate attributed to different regions in the country. Therefore, in this study, focusing on the relationship between climate and region, we extract common characteristics on the nationwide demand factors for a seasonal product group (Japanese alcoholic beverages). We propose a new demand analysis model that enables a comparative analysis among multiple items, more accurately, cross-sectionally, and concisely by using two independent factors: meteorological factors by item and regional weights (demand). We develop a new algorithm using alternating least squares for the estimation problem of inseparable parameters generated by expressing these two factors in the form of products. The validity and effectiveness of the proposed model and algorithm were confirmed by empirical analysis using public data. This makes it possible to theoretically consider regional differences (climate and demand) for nationwide demand of seasonal products. Consequently, the proposed model can be used to conduct multifaceted analysis to situation changes such as climate fluctuations to realize effective sales and operation planning of seasonal products.