Educational games help reinforce educational concepts. They help students learn through hypothesizing, probing, and reflecting upon the game environment. Understanding the impact of a game is important before deploying it in a class. Recent studies in learning analysis describe methodologies and approaches for analyzing educational games. However, richer and deeper analyses are underexplored, especially in games with various decisions and gameplay. Designing games based on provenance appears to be a great approach, but not deeply explored in the educational game environment. In this article, we present ProBee, a provenance-based model for educational game analytics. ProBee uses a provenance data-driven approach that allows for extracting all actions performed by the player, including temporal aspects. The method traces the game's time series, which indicates the player's behavior and adopted strategy during the game session. In addition, we evaluate our approach in a case study with the Control Harvest educational game, which helps teachers present the Biological Control subject