Better understanding and prediction of team performance from collaborative patterns can aid team management and decision on personnel selections. The dissertation adopts data analytics to study team performance in two different contexts. The first project reveals how the research team compositions are associated with team performance in an emerging research area. The second project focuses on predictive models that learn representations of individual members and teams to predict the sports team performance with hierarchical structures. This dissertation benefits my research field of people analytics by proposing modeling frameworks to be potentially applied to team research in other domains.