Abstract: Advanced learning technologies are transforming education as we know it. These systems provide a wealth of data about student behavior. However, extracting meaning from such datasets is a challenge for researchers; and often impossible for the instructors. Understanding learner behavior is critical to finding, extracting, and acting on insight found in educational data. It is equally important to have strong evaluative methodologies to explore the effectiveness of new interventions, and pinpoint when, where, and precisely what students are learning.

This talk covers the ways I have combined human modeling, qualitative, quantitative (statistical and machine learning) methods enable researchers to make sense of behavior and to produce data-driven personalization. I will have a focus on modeling of humans in interactive problem-solving environments, such as intelligent tutoring systems, online courses, and educational video games. Combining results from experimental design, machine learning, and cognitive models results in large improvements to existing learning systems, as well as powerful insights for instructors and researchers on how students behave and learn in interactive environments.

Short Bio: Michael is currently a postdoctoral researcher at Carnegie Mellon University’s Human-Computer Interaction Institute. Michael’s research focuses on deriving understanding from complex interaction data from intelligent tutors and video games. He has worked in data science at Blizzard Entertainment and Warner Bros. Interactive Entertainment (Turbine Inc.). Michael received a NSF GRFP Honorable Mention award, a GAANN fellow, and Freeman-ASIA recipient. Michael was also the PI on a NSF EAPSI grant, in which he traveled to Japan and collaborated with Japanese researchers also working in the educational data mining field. Michael has also performed research in the computer science education and educational video game design, having published a series of educational games with strong empirical studies showing their effectiveness in teaching introductory computer science. Michael has worked on Genetics Cognitive Tutor and has pioneered new methods of individualized student modeling; as a result of this research we have received on Best Paper Award at UMAP 2016 as well as a Best Paper Nomination at ITS 2016.