This module introduces structure discovery in learning analytics.
Let’s explore key techniques and tools.
Learning Objectives
By the end of this module, you should be able to:
Define structure discovery in the context of learning analytics.
Identify methods such as clustering and factor analysis.
Apply structure discovery to educational data.
Structure Discovery
A framework in unsupervised machine learning where, in contrast to supervised learning, algorithms learn patterns exclusively from unlabeled data.
These models explore the data to discover hidden patterns, structures, and relationships on their own.
Other frameworks include weak- or semi-supervision, and a small portion are considered self-supervision (but many scientists consider this unsupervised learning).
Structure discovery methods
Why Unsupervised Learning?
Discover hidden structures or data groupings
Ideal for exploratory data analysis
Data preparation for supervised learning
Reduce dimensionality (simplify data while retaining meaning)
Clustering
Data mining technique used to form groupings
Structure discovery methods
Factor Analysis
Dimension reduction when we have lots of variables
Structure discovery methods
Q-Matrix
Knowledge inference
Skill-item mapping or knowledge component (KC) models
Structure discovery methods
This week
We will go in-depth on clustering, factor analysis, and Q-matrix methods
How to use them to discover insights from educational data
How to avoid obtaining meaningless findings
We will cover examples of each of these three forms of unsupervised learning in the code along
What applications are you interested in?
Who here has already used clustering, factor analysis, or Q-Matrix (or something like it)?
What applications are you interested in?
Who here has already used clustering, factor analysis, or Q-Matrix (or something like it)?
Tell us more – about the data, about the goal of your analysis
