Using Hidden Markov Models (HMMs) to Understand Student Activity Data

by
Won Ng

Recording and analyzing log files of student actions generated when using computer-based learning tools is one method of unobtrusive educational assessment meant to complement more traditional methods. Since manually reviewing each log can be a long and involved process, we are investigating the use of the statistical method of hidden Markov modeling to analyze these logs. Our primary goal is to explore and discover what kind of information about students' learning patterns HMMs can extract from these logs. We are experimenting with and modifying the model parameters to determine the types of events in the log files that will yield interesting or valuable results.

Advised by Steve Tanimoto

EE1 037
Wednesday
April 21, 2004
3:30 - 4:20 pm