Clients' Segmentation Based on Their Purchase History in E-grocery and Offline Stores

Customer classification is a critical aspect of modern retail business strategy. This project employs advanced machine learning techniques to classify customers of a major retail chain in Uzbekistan based on their purchasing behavior, aiming to identify distinct shopping missions. By leveraging natural language processing (NLP) methods and clustering algorithms, this project captures the nuanced goals behind each shopping trip. Word2Vec and FastText models were used to vectorize items and entire orders, with FastText showing better clustering quality. The findings highlight the potential of mission-focused customer classification to drive personalized marketing, improved customer satisfaction, and strategic resource optimization. This approach provides a deeper understanding of customer behavior, enriching traditional segmentation methods.