Predicting which students enrolled in graduate online education are at-risk for failure is an arduous yet important task for teachers and administrators alike. This research reports on a statistical analysis technique using both static and dynamic variables to determine which students are at-risk and when an intervention could be most helpful during a semester. Time-series clustering analysis of online teacher education classes revealed that prediction is possible after the 10th week capturing over 78 % of at-risk students. Visual analysis of dynamic student activities shares a number of striking commonalities consistent with EKG charting. The potential exists for instructors to recognize simple graphic patterns that identify and formatively address these issues with their students. Next phases of research will apply further validation of both the models attempted and additional predictor variables.
Available at: http://works.bepress.com/andy_hung/27/