This study examines college-to-work migration of graduates in the United States using big data from the professional networking website LinkedIn, with a special focus on how the ratio of graduates staying at the areas of study changes over time and how universities affect the stay rate. A multivariate least squares regression model is adopted for analysis. The research finds over 40 percent of graduates graduated 2003-2012 have stayed in the areas of study, with the most recent graduates having the highest stay rate. Graduates from the more prestigious universities are more likely to migrate after graduation. Metropolitan size still dominates variations in stay rates. Strengths and limitations of big data are also explored in this study.