Using BLE beacons and Machine Learning for Personalized Customer Experience in Smart Cafes

Despite recent advances in technology, providing personalized experiences to customers in physical environments like retail shops, restaurants and cafes remains challenging. This paper proposes a pervasive environment that utilizes Bluetooth Low Energy (BLE) beacons in conjunction with unsupervised machine learning to personalize a customer's visit to a coffee shop. Some key aspects of the proposed solution include personalized content delivery, providing customers with an automatic table reservation based on their preferences, using a barista interface to provide personalized interaction with customers, and allowing customers to monitor real-time coffee shop conditions. An architecture utilizing MQTT and an NSQL database was implemented. Historical traces of customer's physical behaviour acquired using beacons in the coffee shops were used to create clusters of similar customers. The best performing clustering algorithms were K-Medoids and Hierarchical clustering using the Optimal Matching (OM) distance resulting in a Purity of 0.910 and 0.941 respectively.