Anonymous transactional dataset in a local food and beverages (F&B) micro-small-medium enterprise (MSME) for...

The dataset presented in this article comprises anonymous transactional records and associated product metadata collected from a local Food and Beverages (F&B) Micro-Small-Medium Enterprise (MSME) operating in a local city in Indonesia. This data can be used by researchers, data scientists, and industry professionals using various techniques in recommender system and machine learning. The data acquisition process involved the passive logging of sales events through the business's internal Point-of-Sale (POS) system from January 2025 to September 2025. The raw data, initially containing a comprehensive transaction log and a detailed product catalog, underwent a cleaning and structuring protocol. The process needed the elimination of seven unneeded features which included blank customer details and duplicate monetary entries and uniform product description formatting. The dataset lacks any distinctive customer identification numbers because it does not contain Customer IDs. The RFM analysis of 1000 transactions through session context grouping produced 14 pseudo-profiles which showed stability as behavioral indicators for unidentifiable users. The final data package consists of two relational tables which include the Transactions Table with 11 core features (Outlet, Date, Time and Total Amount) and the Products Metadata Table with definitions for 96 individual products. The available data in the dataset allows researchers to perform studies about retail analytics and recommender systems. Specifically, it supports the development and benchmarking of algorithms designed for session-based recommendation and the creation of user segmentation models in anonymous, data-sparse environments typical of the MSME retail sector.