This document discusses customer segmentation using an unsupervised machine learning model. It analyzes transaction data from a brewery's 99,708 customers over 2 years to group them into meaningful marketing segments without prior customer information. A self-organizing map (SOM) model was used to iteratively cluster customers based on similarities. The optimal number of clusters was determined to be 3 based on cluster distances and averages. The 3 clusters identified were "Promising", "Explorers", and "High Value", with the latter found to be the best target segment for marketing due to its high sales percentage and visit frequency.