Almost half of the students who begin college at a two- or four-year institution fail to earn a degree within six years. Can analytics help?
Big Data's Surprising Uses: From Lady Gaga To CIA
(click image for larger view and for slideshow)
Measuring academic performance isn't new. Consider the venerable report card, which gives student and school alike a way to monitor progress during a single class, an academic year or across generations.
But the advent of computer-mediated and online instruction, especially massive open online courses (MOOCs) with their tens of thousands of students per class, are changing what's possible. That's because they provide so much data about a student interactions, not only with the course material but with teachers and even other students. This massive amount of data can be parsed, compared, merged, modeled and analyzed, with the goal of improving educational outcomes.
Proponents say these systems will be able to detect struggling students sooner than traditional means, triggering a teacher intervention or even making an automatic change in a lesson plan.
An IBM white paper on the topic, "Building a Smarter Campus: How Analytics is Changing the Academic Landscape," also notes that once predictive analytics take current and past data to figure out what will happen next, prescriptive analytics can be applied: "Prescriptive answers the ensuing question: In light of what we believe is going to happen, here are recommendations on how to best respond."
The IBM white paper goes on to discuss other uses of analytics by schools, such as targeting student scholarships and identifying key donors and investors.
Why have predictive and prescriptive analytics become hot topics in educational circles? Undoubtedly one reason for all the attention comes from declining college graduation rates in the United States. Almost half of the students who begin college at a two- or four-year institution fail to earn a degree within six years.
In fact, just last week, the National Commission on Higher Education Attainment report "College Completion Must Be Our Priority" called for raising college graduation rates, noting that college populations are changing and will increasingly include older and part-time students.
"The number of Americans attending college is at a historic high, but far too many never make it to graduation," the report begins. "This is an unacceptable loss of human potential -- a waste of time, resources, and opportunity. Left unaddressed, it will hinder social mobility and impede the nation's economic progress."
The report also explicitly called for postsecondary institutions to seek ways to "enhance student learning outcomes," and referred to a number of projects using computer technology for creating individualized teaching plans, combined with ongoing evaluations.
"Within higher education, the emergence of massive open online courses may change how we serve students to accommodate their schedules using novel new platforms, social media, and predictive analytics," the report says.
Indeed, vendors of educational platforms have been promoting their analytical strengths.
"In the last three or four years, we built a team of five PhDs, who've built algorithms and models to predict student performance," John Baker, CEO of learning management system (LMS) company Desire2Learn told InformationWeek in a phone call.
Last fall, Desire2Learn ran an analytics beta.
"Our predictive algorithm makes dynamic predictions beginning with the first week of term," Baker wrote in a follow-up email. "Depending on the availability of historical data associated with a specific course we are able to achieve accuracy rates approaching 95% as early as week two and three," Baker said, adding this was validated with research data sets, including one from the University of Wisconsin.
Desire2Learn created a "risk quadrant," a visual representation of how each student is likely to do in the course. The predictions are made dynamically on a week-to-week basis.
In his email, Baker explained the quadrants this way:
"The first quadrant displays the students who are on-track (likely to pass the course) and fully engaged. The second quadrant displays the students who are on-track (likely to pass the course) but not fully engaged. The third quadrant displays the students who are likely to drop out or withdraw from the course. The fourth quadrant displays the students who are likely to fail the course or get a very low grade."
When the feature goes into production in March, interventions will be incorporated into the model as well. "And we'll learn from those interventions what works and what doesn't, which will provide more insight," Baker said.
Desire2Learn's predictive models can be customized by course and by institution. "Through a simple plug-and-play interface, we allow the institution to build and tune the model themselves without having to rely on Desire2Learn," Baker wrote.
Separately, Desire2Learn added to its analytic portfolio in January, announcing the acquisition of Degree Compass, a predictive analytics technology that helps students select the courses to take that will drive completion of their degree program. The Degree Compass product will be generally available in spring 2013. Financial terms of the acquisition were not disclosed.
The MOOC vendors have been promoting their analytical strengths too.
Daphne Koller, a founder of Coursera, a leading provider of free online courses in the United States with 30-plus university partners, said her platform and others do more than simply scale instruction to larger student populations or reduce the cost per student with things like automatic grading.
Speaking at a day-long conference last month on the impact of MOOCs on California's education system, Koller said these systems are ushering in a "brand new pedagogy" and provide important keys about effective teaching and educational design.
"We can now do the kind of rapid evolution in education" that is common at companies like Google, which 'A/B test' their ad positions and user interface elements for effectiveness," she said. "These websites evolve in a matter of days or weeks rather than years."
But other MOOC executives are more cautious.
"Analytics is our biggest department," Advance Learning Interactive Systems Online (ALISON) CEO Mike Feerick told InformationWeek in a phone call. With more than 1.2 million unique visitors per month and 250,000 graduates worldwide, ALISON, founded in 2007, is considered by some the first MOOC.
Nevertheless, Feerick said analytics, while extremely important, can't reform education by itself. "This consuming idea that data will solve everything is erroneous," he said. Talented teachers using these and other new tools are the key, he said.
Also voicing concern about analytics was educator and blogger Will Richardson, whose e-book, "Why School: How Education Must Change When Learning and Information Are Everywhere," was published last September by TED Books.
"I don't have a problem using keystrokes to figure out what a child knows, needs to know or where they're at," Richardson told InformationWeek. "I have problem with who makes the decision about the best path from that point forward."
Richardson has written at length about what he believes is the major flaw in schools today: A focus on passing tests, especially in an online world in which information and answers are abundant. "Learning is a messy process that you can't always predict," he said. "There's a danger doing that, taking out the serendipity."
InformationWeek is surveying IT executives on global IT strategies. Upon completion of our survey, you will be eligible to enter a drawing to receive an Apple 32-GB iPad mini. Take our 2013 Global CIO Survey now. Survey ends Feb. 8.
6 Tools to Protect Big DataMost IT teams have their conventional databases covered in terms of security and business continuity. But as we enter the era of big data, Hadoop, and NoSQL, protection schemes need to evolve. In fact, big data could drive the next big security strategy shift.
Big Data Brings Big Security ProblemsWhy should big data be more difficult to secure? In a word, variety. But the business won’t wait to use it to predict customer behavior, find correlations across disparate data sources, predict fraud or financial risk, and more.