Understanding Acquisition Prices at Swiss Hypermarket: A Time Series Approach on Predicting Acquisition Prices

In this case study consulted at a Swiss hypermarket, the problem of decreasing profit margins is addressed by focusing on the constantly changing acquisition prices in the buying process. The increasing uncertainty of the acquisition prices makes it difficult to set the selling prices, and plan the buying process and therefore heavily affects the resulting profit margins at the hypermarket. This project serves as a pilot project, in which three product categories are analyzed with respect to their acquisition prices and which outside factors influence them, to understand and eventually anticipate future acquisition prices. The analysis is done as a set of time series analyses, comparing the effectiveness of univariate and multivariate machine learning models. The models aim to improve the understanding of the acquisition prices’ behavior as well as answer the question of whether and to what extent the prices can be predicted using the pricing data and external data. The project is a combined analysis of data and theoretical insights given by the subject matter experts of the three product categories and other main players in the hypermarket. This report offers a variety of insights about the nature of the three product categories, which factors impact the acquisition prices, and to what degree. Moreover, it shows that good predictions using relatively simple machine learning models are possible and that selecting the right external factors can improve predictions considerably. Eventually, it offers a range of recommendations and advice on how to continue this pilot project and how to largen its scope effectively.