In the winter of my first year at the University of Minnesota, a fellow student proposed screening his carefully made VHS recordings of the 16-hour Live Aid “super concert.”
Seen by more than a billion people when it was broadcast earlier that year, the support for rebroadcast in the TV lounge of the Middlebrook Hall honors dorm on the eve of finals week was as unanimous as it was inexplicable.
A mature observer may have been able to predict with some accuracy the odds of who would maintain their academic honors GPA in the semesters to come by noting who joined the standing-room-only crowd instead of studying.
Today, the use of university-produced academic video brings the promise of redemption and prescient help for students of all stripes through the science of predictive analytics.
Predictive analytics is a statistical technique used to extrapolate future trends from historical data. In business, this means assigning a score rating the likelihood of action or preference based upon patterns found in previous transactions. I
It’s the technique that allows retail giant Target to be the “second to know” when your family is expecting based solely on subtle changes in your buying patterns.
Watching what we watch
Colleges and universities are experimenting with ways to use student data to support academic success. The challenge is accurately identifying opportunities to provide proactive help. GPA trends, class attendance, study group participation and assignment completion statistics have a proven track record in predicting success.
Recent studies confirm that reading and posting messages, consuming content, and viewing files online also successfully predicted the majority of the variance in the final student grades.
Translating the benefits of predictive analytics to lecture capture systems is both an opportunity and a challenge. The high-quality reproductions provided by automatic lecture capture systems have replaced bespoke, manually annotated content.
The high-fidelity reproduction of the classroom experience is a runaway hit with students—for review as well as a substitute for missed classes.
The leading systems in this market provide a data set of viewing statistics that can be accessed via dashboards, reports and API calls, and in some cases direct integration with the analytical functions of a CMS.
Lou Harrison, director of education technology services for DELTA at North Carolina State University, says there is significant potential in the data logged by students within lecture capture systems.
“Predictive analytics is almost all correlation-based,” he says. “For example, our predictive models may someday find a correlation between the performance of a student who watches a video at normal speed versus one who ‘chipmunks’ through it at double-speed.”
Room for improvement
Automatic lecture capture systems produce content at a fraction of the cost-per-hour of videos, but this process is also opaque. An hour of recorded lecture may be dense with discrete concepts driving viewership, but the absence of thematic segmentation hides potentially significant insights about the consumption of content.
Some academic leaders believe the most abundant vein of metadata for predictive analytics may lie entirely outside the video player window.
“The circumstances that surround a student as they watch a video are as valuable from a data perspective as which video they select to view,” says William Lindner, director of the Campus Reimagined Initiative at Florida State University.
“Where, when and why a student watched a video can be as valuable as what they watched in many cases. The essential information for accurate prediction may turn out to be as much about the surrounding environment the viewer is in as the number of rewinds or selected viewing speed.”
Watching students watch video holds great promise as a new area to identify and support opportunities to increase success on campus. And, perhaps someday, video analytics may alert a student who could benefit from urgent advisor intervention in the Middlebrook Hall TV lounge.
Sean Brown helped develop and launch disruptive academic technologies at firms such as Apple, IBM, Oracle and Sonic Foundry.