Posts

Will You Graduate? Ask Big Data

Journal of Learning Analytics: Commentary

New MOOC, Practical Learning Analytics, Encourages Group Participation

Professor Tim McKay has created a new massive open online course (MOOC) called Practical Learning Analytics to address the typical challenges faculty, staff, administrators and students face to gather, access and analyze data in direct, practical ways.

The course builds upon U-M’s leadership in learning analytics and is the next innovation in a series of initiatives that have sought to make personalized data actionable on a large scale to empower students, faculty, and staff to make better informed choices to improve learning.

Based on Michigan’s on-campus Learning Analytics Fellows program, the course focuses on analytical approaches that anyone can take, and offers a number of innovative approaches that differentiate it from other open online courses. For instance, the course offers a practical approach by providing students with real sample data and code to help facilitate analysis.

Practical Learning Analytics is also the first U-M MOOC to encourage group participation and dialogue as an ideal model for participation. The course is structured to accommodate as many tastes and needs as possible, allowing learners to select their own level of involvement. Professor McKay recommends dividing course content among a group of peers and  exploring notes and key themes in weekly ongoing discussions to help students explore the full content of the course while learning from the perspectives and insights from peers.

Practical Learning Analytics is also bridging impact with global lifelong learners with unique on-campus opportunities for current students. On October 20, the Digital Innovation Greenhouse within the Office of Digital Education & Innovation (DEI) will host a student hackathon where students will utilize data sets from the online course.

Enrollment is currently open for Practical Learning Analytics, which launches on October 5. For additional information and/or to enroll, please visit https://www.coursera.org/course/pla.

Precision Learning?

Gus Evrard, Thurnau Professor of Physics and Astronomy

In a 2013 Nature Medicine article, Alla Katsnelson noted a shift in the lexicon of modern clinical medicine; the framework once known as personalized medicine had morphed into precision medicine. Might educators soon be following suit?

The idea that knowing an individual’s genomic structure may lead to improved medical treatment holds great promise, but the fact that all humans share a common basic physiology poses fundamental limits on how personalized medicine may become. Instead, the concept that each of us is a machine with a slightly different biochemical instruction set means that: i) we can be categorized based on those differences and, ii) treatments can be made more effective through precise understanding of how biochemical variations drive physiological changes associated with disease. The practice of precision medicine will be powered by data (it’s sexier to say Big Data), including laboratory data on underlying biochemistry and clinical data on treatment outcomes.

In the world of education, the concept of personalized learning is now associated with a host of efforts that seek to promote better outcomes for individuals across the full spectrum from kindergarteners to life-long learners. At Michigan, DEI is promoting a number of strategic projects that seek to develop and apply new technological solutions aimed at enhancing teaching and learning for all individuals.

Academic Report Tools (ART) is one such project. Michigan is a very large university, and no one can hold in their heads the rich and ever-evolving detail of the curriculum offered by its 19 schools and colleges. Think of ART as providing a “paint-by-numbers” approximate view of this complex reality. The service currently allows faculty and staff to easily view historical enrollment and grading patterns (including co-enrollment data among pairs of courses), providing fragmentary pictures to be painted that impart some knowledge. Are Engineering students outperforming LSA students in the first-semester course in Statistics? ART slices into the historical data and quickly provides the answer.

We are now beginning to plan the next generation service that will include tools aimed at our students. Personalizing these services is important, in that the needs of (say) a Master’s student in the School of Information are likely to differ from those of (say) a first-year undergraduate in LSA who’s thinking of a dual philosophy and physics degree but might want to do biopsychology instead. What can connect these students is the desire to view some historical collection of data that paints a useful fragmentary picture, one that may inform an impending decision.

Most students (and tuition-paying parents) would ideally like a very high degree of personalization, something along the lines of, “Based on your current standing, taking this particular set of N courses in the following order will essentially (>95% confidence) guarantee a career in discipline X and a high lifestyle happiness index five years after graduation.” Even in our overtly data-rich world, this is currently too much to ask.

Instead of this high degree of personalization, tools for precision learning could follow the lead of medicine. Every student is a proverbial snowflake (unique in detail and, thereby, above average in some particular measure), but viewed from a distance she or he can be classified by a relatively modest set of attributes. Opening windows into how students with different attributes have tied into curricular and co-curricular activities at Michigan, and to careers beyond campus, is a service of great interest to educators, students and parents alike.

Of course, higher education is a dynamic enterprise, so using past or current trends as an indicator of future performance will entail risks.

Being precise means having little uncertainty. Thankfully, the large size of the existing student record database at Michigan means that there is low-hanging fruit that ART services for students can pick. We can paint courses by their numbers, allowing views of student attribute composition, instructor evaluations, and historical grade distributions. Another practical issue students need help with is a better way to formulate and choose among options for multi-term course selection. This is particularly important for juniors and seniors who have declared multiple majors.

While precision learning services don’t aspire to the higher ideals of personalized learning, their utility may render the subtle shift in lexicon of interest only to academics.

 

Events

AIM Analytics: Some Challenges for the Next 18 years of Learning Analytics

Join us on Monday, February 25 from 12:00 p.m. to 1:30 p.m. in the West Conference Room (4th Floor) of Rackham Graduate School (915 E Washington St) for AIM Analytics as we welcome in Ryan Baker, Associate Professor in the Graduate School of Education at the University of Pennsylvania.  

AIM Analytics is a bi-weekly seminar series for researchers across U-M who are interested in learning analytics. The field of learning analytics is a multi and interdisciplinary field that brings together researchers from education, learning sciences, computational sciences and statistics, and all discipline-specific forms of educational inquiry.

To register for this event, please RSVP HERE. Lunch will be provided.

Ryan Baker

Ryan Baker, Associate Professor in the Graduate School of Education at University of Pennsylvania

Title: Some Challenges for the Next 18 years of Learning analytics

Abstract:

After nine years of learning analytics conferences, we have accurate models of constructs many didn’t think we could model, dashboards and interventions and (some) evidence they work, and scaled solutions that are being used to change student outcomes. Learning analytics has been unusually successful in a short time.

Let’s pat ourselves on the back. And after that, let’s reflect.

We have solved some challenging problems. So, what’s next?

Where should we go — and are we actually going there?

I have a few thoughts. And a few concerns.

In this talk, I’ll discuss a few hard problems that I see looming in the path of an optimally beneficial learning analytics; some of the big goals I think we can strive to achieve; some of the grand challenges we will need to — and I think can — solve; and perhaps most importantly — how we’ll know if we’ve gotten there.

RSVP:

To register for this event, please RSVP HERE. Lunch will be provided.

Portfolio Items