SIADS 680: Learning Analytics
Note: This course is still under development (to be launched in March 2021).
Course Overview
SIADS 680 provides an overview of a key application domain for data scientists—education. In this course students will examine the application of data science as a means to better understand and improve learning. Specifically, students will think critically about the ways in which data scientists can support research and improvement in educational organizations of all types. Anchored in the fields of learning analytics and educational data mining, this course analyzes the unique opportunities and challenges associated with applying data science methods to data stemming from schools, universities, and a myriad of learning opportunities. The course will cover the history of learning analytics, typical data and methods used, the importance of measurement, and the implementation of learning analytics products.
Prerequesites
- SIADS 521, Visual Exploration of Data
- SIADS 532, Data Mining I
- SIADS 542, Supervised Learning
- SIADS 543, Unsupervised Learning
- SIADS 632, Data Mining II
Instruction team
Instructors
- Dr. Andrew Krumm, Assistant Professor, School of Information & Medical School
- Dr. Stephanie Teasley, Research Professor, School of Information
Course Assistant
- Dr. Quan Nguyen, Research Fellow & Lecturer, School of Information
Learning objectives
- Understand the application of data science in educational research - learning analytics
- Develop meaningful metrics to measure student engagement and student success in higher education
- Build prediction models of student success based on academic records and click-stream data generated from Learning Managemente System
- Investigate the algorithmic biases in prediction models
- Build a learning analytics dashboard to communicate results to non-technical audience
- Understand ethical concerns associated with collecting, communicating, and taking action on educational data.
Weekly Office Hours via Zoom (Ann Arbor, Michigan time):
Your instructor will hold weekly, synchronous office hours using the video-conferencing tool, Zoom. The schedule of office hours can be found by clicking on the Live Events link in the left-hand navigation menu. Additionally, all office hours will be recorded and archived so that you can retrieve them at a later date. Archived office hours can be found by clicking on the Resources link in the left-hand navigation menu then clicking the Archived Sessions link.
Communication Expectations
- Please only use the course channel in Slack
- Please ask all questions in public if possible so others can learn from your question.
- If you need to ask a private question please direct message both the instructor and the course assistants.
- Email response time: 48 hours
- Slack response time: 48 hours
- Office hours: see the Google Calendar.
Assignments
In this course, you will be working with a wide variety of data about students and their learning processes. In each week, you will be tasked with different questions that requires extracting insights from the data, in order to inform future teaching and learning practices. Here's an overview of the next four weeks:
| Week | Jupyter Notebook (20% each) | Reflection (5% each) | Tools/libraries | Weight |
|---|---|---|---|---|
| 1 | Exploratory data analysis for learning analytics | What are some historical trends in students learning behavior and academic performance? | numpy, pandas, matplotlib, statsmodels | 25% |
| 2 | Building prediction models of student success | Which students are likely to dropout or fail? Can we build an early at-risk detection model for in time interventions? | scikit-learn | 25% |
| 3 | Investigating the biases of prediction models | Are the prediction models biased toward certain population of students? Are the data biased? | matplotlib, scikit-learn | 25% |
| 4 | Building a learning analytics dashboard | How can we communicate learning analytics insights to a non-technical audience (instructors, students, managers)? | jupyter-dash | 25% |