In this study, data was acquired by fitting GPS sensors to Gaelic football players which are worn during game time. These wearable devices collect internal (heart rate) and external (gps positioning) data related to the movement of players. However, this form of raw data is not well suited to machine learning algorithms as it lacks the necessary semantics which can identify the type and duration of movements. The dataset presented here is created by a data engineering exercise, driven by domain experts, to transform the GPS coordinates into a series of (player) actions. The end result is a database comprising 12 variables and almost 160k actions. It’s reuse potential is targeted at machine learning researchers, sport scientists and coaches who are seeking to understanding the effort and load of players during game time. Analysis is enables across five dimensions: games, players, actions, duration and speed. In addition, the concept of an event groups together actions that belong to the same sequence and enables analysis at a different level of abstraction.