Inventory Management for Automated Convenience Stores in Brazil

Abstract

Today, small retailers in Latin America account for 70% of the market share. Convenience stores play a crucial role, as people look for more convenience with modern lifestyles. Most of these stores are managed by people without experience or formal education in business management. A challenging problem for small retailers is inventory management. We developed this project for Onii, a Brazilian startup company with over 300 convenience stores in the country, run by small businesspeople in a franchise-like model. Its stores are entirely automated; thus, there are no cashiers or employees inside the store. This project aims to develop inventory management policies for Onii store operators and help them manage their stocks better. We use unsupervised Machine Learning techniques like k-means clustering and principal component analysis to identify patterns and segment stores and items. Then various inventory policies were computed to look for the lowest cost for each combination of clusters of stores and items. The best policy for the Onii store's reality is the Periodic Review model, with different period parameters (R) for each combination. At last, sensitivity analysis was conducted to determine the impacts of each parameter used in the model, such as ordering cost, holding cost, and inventory cost. The result is a robust model that Onii can apply to their current and future stores.