Source: Bruce McLaren Educational Data Mining

Download Report

Transcript Source: Bruce McLaren Educational Data Mining

Deutsches Forschungszentrum für Künstliche Intelligenz
Educational Data Mining
WS 2007/08
Introduction to the Seminar
Dr. habil Erica Melis
Dr. Bruce M. McLaren
Paul Libbrecht
Source: Bruce McLaren
Educational Data Mining Seminar 2007/08
1
What is Educational Data Mining (EDM)?
 Making good use of the raw data collected by e-Learning and
educational technology systems
Motivated by:

Proliferation of data from many Internet-based educational systems

Base conclusions and development on real data rather than conjecture and
intuition
 Use educational data to, for example, improve systems, evaluate
student behavior, support teachers
 Interactive Learning Environments: intelligent tutoring systems,
collaborative systems, open inquiry systems
Scaling up - Possibility for large-scale and longitudinal analysis
How are students learning from and reacting to educational
technologies?
Source: Bruce McLaren
Educational Data Mining Seminar 2007/08
2
Uses of Educational Data Mining
 Find common errors committed or help requests made by
students, so that subsequent versions of educational technology
can better address them
 Student modelling
 Learn how to create adaptive systems that change their approach
based on different learning styles
 Discover ways that students “game” the system, i.e., students that
do not seriously try to learn but rather just try to get through the
technology, and how to react to this
 Provide ways for teachers to analyze -- and react to -- student
efforts
Source: Bruce McLaren
Educational Data Mining Seminar 2007/08
3
Educational Data Mining Tools & Techniques
 Machine Learning
Many techniques available -- and have been largely
prepackaged, e.g.,
 Decision Trees
 Support Vector Machines
 Boosting algorithms
Off-the shelf tools
 WEKA (A flightless bird, found in New Zealand)
 YALE (Yet Another Learning Environment)
 Statistical Techniques
Bayesian analysis of data
 Language analysis, esp. for collaborative systems
Off-the shelf tools
 TagHelper
Source: Bruce McLaren
Educational Data Mining Seminar 2007/08
4
Seminar Schedule
 22.10.2007 Introduction - DFKI Bledsoe
 29.10.2007 Introduction to Machine Learning - 16.00 DFKI
Room to be decided and published on website
 05.11.2007 ActiveMath Presentation and Demo - 16.00 DFKI
Room to be decided and published on website
 Work on projects throughout the semester
Meet with your advisor at least twice
 Work on your e-Portfolio
Martin Homik will explain shortly …
 Presentation of student projects
Selected dates: Thursday Feb 28; Friday, Feb 29
If there are any conflicts with these dates, send email to Erica, Bruce
& Paul very soon!
Source: Bruce McLaren
Educational Data Mining Seminar 2007/08
5
Course Requirements - Grading
 Key requirement: Present a paper from the seminar website:
http://www.activemath.org/teaching/eddatamining0708/literature.php
 Papers selected during today’s seminar, if you miss the first seminar contact:
 Dr. Erica Melis ([email protected])
 Dr. Bruce McLaren ([email protected])
 Paul Libbrecht ([email protected])
 Read not only this paper, but important referenced and related papers
 Meet at least twice with your advisor (advisors listed next to each paper on the website)
 Send first version of slides to your advisor at least 2 weeks before presentations
 Present the paper at the final seminar meetings
 Attend the two introductory lectures, plus all student
presentations
 Participate in lecture discussions
 Participate in individual ePortfolios - Martin Homik will explain
http://edm.activemath.org/
Source: Bruce McLaren
Educational Data Mining Seminar 2007/08
6
Any Questions?
Source: Bruce McLaren
Educational Data Mining Seminar 2007/08
7