Data science competitions, such as hackathons, offer dynamic, hands-on learning experiences that can significantly enhance students' skills and knowledge. This presentation explores how a data science competition facilitated collaborative learning, skill acquisition, and the application of known skills in new contexts. To investigate these phenomena, I conducted a detailed study of one hackathon, video documenting participant groups from beginning to end. Additionally, I collected surveys to gauge participants' prior experiences and motivations towards data science, examining how these factors influenced their engagement during the event. A critical aspect of this study focuses on the instructional structure of the competition as well. Despite often receiving minimal instructions, participants frequently benefit from specific content types provided by organizers. This research also aims to identify characteristics of minimal instructions that are promising in fostering an effective learning experience. In this presentation, I will detail my study's methodology and share insights on how to study learning experiences during short-term data competitions. I will also share preliminary results, highlighting key findings on student collaboration and learning processes from my study. By understanding these elements, we can better design data science competitions to maximize educational outcomes and student engagement.