A Machine Learning Approach for Prediction of Corrugated Cases Prices

Abstract

Even though machine learning is widely spread among different industries, its application in the fast-moving consumer goods (FMCG) business is not a common practice and even today it remains in its early stages. Moreover, to our knowledge, there has never been a systematic approach to predict packaging materials costs in this kind of markets using machine learning algorithms, from the buyer’s perspective. On the other hand, the FMCG business is a highly competitive environment, in which profitability depends not only upon sales, but also upon keeping healthy product margins. This means not only setting the right prices that consumers are willing to pay, but also getting the lowest possible costs in the supply chain. Cases usually represent between 15% and 25% of the total packaging cost, being a material with functional requirements that usually does not add value to the consumer. Therefore, it is of high importance to maintain low cases prices to achieve competitivity. In this work we propose a machine learning approach for prediction of prices of a corrugated cases portfolio of a big FMCG firm in LATAM, to understand if real prices are higher or lower than what is suggested by the model. In this way, anomalies in the dataset will be unveiled, which might become opportunities for further negotiations and costs reductions.