Hands on with Bayesian Knowledge Tracing

Bayesian Knowledge Tracing has been at the fore in assessing student changing knowledge in digital learning environments. While originally put into practice in Intelligent Tutoring Systems in order to assess student skill mastery, it's uses now include evaluation of item diagnostic power, and learning properties of various interventions and patterns of use. In this tutorial I will introduce BKT as it is current used in practice, the basics of the algorithm, and a hands-on tutorial of how to apply BKT analysis to a dataset to assess individual student knowledge and test learning hypotheses. The tutorial will utilize Kevin Murphy's Bayes Net Toolbox which is compatible with Octave, a free alternative to MATLAB.

Please contact Zach Pardos (pardos at berkeley . edu) with any questions.

Links

Octave

  • MATLAB alternative interpreter [link]

Bayes Net Toolbox

  • Kevin Murphy's Bayes Net Toolbox [link]

[More information will be posted to this page leading up to LAK 2015]