This design project is a collaboration between Hunter Montgomery, Sumant Agrawal, Alex Qin, and The Wine Merchant, a premier wine retailer in the St. Louis Area.

Abstract

The wine industry’s volatility due to external influences causes inconsistencies in purchasing behavior for merchants. These inconsistencies lead to suboptimal business decisions, such as over/under-stocking and accumulation of idle merchandise leading to loss of profit. While each small business faces unique challenges, our client (a St. Louis wine merchant) desires a model that predicts both type and quantities of wines to purchase to reduce incurred business costs and improves store performance and inventory management. Our group used applications of forecasting techniques (namely, the Holt-Winters and Triple Exponential Smoothing methods) and linear programming to produce a recommendation that accounts for recent sales history as well as the physical constraints of their store. These reports can be used and evaluated over time to assess the predictability of products as well as determine which products (namely the wines with poorest sales) are no longer profitable to keep within the store. This is accomplished in three stages: data aggregation, forecasting demand, and using the predicted demand to determine store arrangement.