New product forecasting of appliance and consumables : SARIMA model

Abstract

Drinkworks is a joint venture between Anheuser-Busch InBev and Keurig Green Mountain, with a focus on developing an in-home alcohol system that can prepare different alcoholic beverages. The goal of this project is to forecast the demand for their new product, consisting of appliance and pods, without historical data. For appliance forecast, this paper focuses on an operational level model, SARIMA, which is a time series analysis that considers seasonality and has high accuracy in forecasting. The SARIMA model is implemented with grid search in Python via a demand planning tool, which saves client's time. Weighted consumption rate will be utilized with number of appliance sold to forecast future pods sales. SARIMA model proved to be an effective approach for appliance forecast within client's expectation. A systematic way to forecast pods is also proposed and demonstrated. It is hoped that the results presented here can serve as a basis and help the client with their new product launch.