Congestion remains one of the most prevalent transport problems in major cities. Recent approaches to managing demand aim to make working hours flexible to reduce congestion during peak periods. However, these approaches must also address synchronization needs at the employer, household, and individual levels. This study presents a framework to identify critical commuters who can adjust their arrival times, benefiting individual motivations, and managing arrival time demands. We explore multiple machine learning approaches to model and predict an individual's ability to shift their workplace arrival times. Ultimately, we choose gradient boosting due to its superior performance. Utilizing this ensemble approach on an employee survey data from Rennes Metropole in France, we analyze the factors that influence individual's flexibility in their arrival times. Key determinants identified include regular school drop-offs, theoretical arrival time contract with employer, and, to a lesser extent, age and income. Our findings demonstrate the need for a bi-level framework that incorporates both arrival time demand management and social justice analyses to ensure effective and equitable outcomes.