A Comprehensive Synthetic Dataset of Simulated RWH User Daily Activities and Preferences

Abstract

The global work environment has experienced a transformation in recent years, which has been greatly accelerated by technological advances and the growing importance placed on the benefits of remote working, such as reduced emissions, time savings and improved mental health. This shift has contributed to the grow ing popularity of Remote Work Hubs (RWHs)/ Coworking Space, which combine traditional office infrastructures with the flexibility required for modern remote work, catering to a diverse group of people such as entrepreneurs, freelancers and remote workers. This study presents a pioneering approach to generating syn thetic datasets using Large Language Models (LLMs) via APIs to bridge the gap of accessible user data. Leveraging the ability of Large Language Models to gen erate contextually rich and diverse data, we simulate the nuanced activities and decision-making processes of RWH users. This synthetic dataset provides foun dational insights for coworking space design, management, and policy support through extensive market research and personal development. Through a well designed methodology, including persona generation and diary entry synthesis, we provide a comprehensive picture of the daily activities, workplace decisions, and commuting preferences of shared workspace users based on real-world data sources and advanced model configurations.