With the explosion of online learning environments available in the classroom, in afterschool programs, and at home, there is promising potential for large-scale data collection and analysis in education research. In this session, we discuss the kinds of educational and learning problems that big data can help us better understand. The first talk will discuss the identification of learning pathways and strategies used by students in an online math game, and what this means for personalized learning for students and customized feedback for teachers in math and extending to other STEM fields. The second talk will illustrate how the use of back-end log data can illuminate patterns of studying activity employed by students, and the opportunities this improved understanding of different strategies provides to students and researchers. The final talk will discuss the impact of affective states and student engagement on learning. It will discuss how affect can be sensed via student-system interaction patterns and how learning technologies can dynamically tailor instruction in a manner that is sensitive to both affective and cognitive states.