Development of a Forecasting Model for Coffee and Carbonated Drink Sales

The expansion of health-conscious consumers and the launch of decaffeinated and zero-sugar products have led to an increase in coffee and carbonated beverage sales. This study developed a sales forecasting model for key items of interest using category sales data collected from January 2021 to December 2023 from the high-demand distributor POS index dataset provided by Market Link Co., Ltd. In particular, the analysis focused on coffee and carbonated beverage sales across all channels, hypermarkets, and convenience stores. Among time series analysis techniques, the ARIMA model and Holt-Winters exponential smoothing method were employed, and the AIC and SBC indicators were used for model comparison. Since all analyzed distribution channels exhibited some degree of trend and seasonality, trend differencing and seasonal differencing were performed. As a result of the analysis, the ARIMA model with the constant term removed, which had the lowest AIC and SBC values, was selected as the optimal model. However, since this study was conducted in the post-COVID-19 period, external factors such as economic changes may have influenced sales, indicating the need for additional research that accounts for such factors. Moreover, this study is limited to a specific period and beverage category. Future research should consider a broader range of beverage categories and long-term data to improve long-term sales forecasting and analyze interactions between different beverage types.