Course Description
Focuses on measurement, collection, analysis and reporting of data about learners and their contexts for purposes of understanding and optimizing learning and the environments in which learning occurs: Educational Data Mining. Introduction to theory and application of common techniques along with current issues in the field.
Athena Title
ED DATA MINING
Prerequisite
ERSH 8320
Semester Course Offered
Not offered on a regular basis.
Grading System
A - F (Traditional)
Course Objectives
Educational Data Mining (EDM) draws from several fields, including statistics, psychometrics, machine learning, information retrieval, and scientific computing, to analyze data generated by systems that support learning or education. Such data might include the movement of patrons through a museum or the sequence of actions that students take while solving problems in an adaptive computer tutoring system. In this course, students will learn to use common statistical learning and data mining methods, learn the theory behind these methods, and gain an overview of current problems in EDM. After completing this course, students will demonstrate knowledge of the theoretical foundations and be able to appropriately use common supervised and unsupervised learning methods, including: - cross-validation - the bootstrap - Gaussian mixture models - k-means - principal components analysis - hierarchical clustering - variable selection methods, such as penalized regression - linear discriminant analysis - logistic regression - separating hyper-planes, such as support vector machines Students will examine the current literature in educational data mining on topics such as: - Modeling student learning in intelligent tutoring systems - Modeling student affect, engagement, self-regulation, and metacognitive strategies - Analyzing complex behavioral data, such as participation in online discussion forums - Readings will be drawn from sources such as the Journal of Educational Data Mining (EDM), the International Conference on Artificial Intelligence in Education (AIED) and Neural Information and Processing Systems (NIPS)
Topical Outline
The course will be divided into two parts, as outlined in the course objectives: Part 1 - Study of Common Methods: - Review of regression and joint probability distributions - Clustering/Unsupervised learning methods, includes: Gaussian mixture models, k-means, principal components analysis, and hierarchical clustering - Regression-based supervised learning methods, includes: variable selection methods and linear discriminant analysis - Linear classification methods, includes: separating hyper-planes and logistic regression Cross-validation and the bootstrap are tools for assessing the fit of models and predictive error. The theme of model assessment will be woven throughout the course. Part 2: Study of Current Literature Readings will be drawn from the literature and discussed in class based on student interests.
Syllabus
Public CV