We introduce GroupSet, a technique to facilitate the exploration of temporal charts using a set-based approach. GroupSet operates in a twofold way: first it classifies temporal data into categories (sets) for each time point, second it enables to explore such membership to categories (sets) over time. This approach enables to reveal temporal similarities of elements by categories (sets) memberships, usually hidden by overplot. We demonstrate the applicability of the technique to two case studies (traffic data and sport data) and report on usability feedback of an interactive prototype implementing the technique. Our code and datasets are published as an open-source project and we expect further research towards efficient set creation and temporal manipulation, which remain under-explored areas in the domain of set visualization and interaction.