Is Predictive Analytics a Game Changer for Higher Education?

|

In a previous post, we discussed why big data is a big priority in higher education in 2017. The collection and analysis of big data is critical to improving student success and graduation rates, recruitment, academic advising processes, and the monitoring of student behavior and performance. Data is also at the core of business intelligence, reporting and analytics systems that improve decision-making and support institutional goals.

In addition to using historical data to analyze past behavior, performance and trends, colleges and universities are using predictive analytics to determine what will happen next. Predictive analytics applies a variety of techniques (data mining, statistical algorithms, analytical queries, machine learning, artificial intelligence, predictive modeling, etc.) to historical and new, real-time data sets. This makes it possible to predict the likelihood of future patterns, outcomes or events based on a numerical score.

In higher education, schools are expected to identify students who are likely to succeed, as well as at-risk students. Schools are expected to identify opportunities to help students succeed with targeted support. Predictive analytics can help colleges and universities achieve these goals, using data rather than assumptions.

A recent article in The New York Times points out that slightly less than half of college students graduate in four years. That number only increases to 60 percent after six years, which has created increased pressure on schools to do better. More schools are now relying on predictive analytics to produce better outcomes. The goal is to identify successful student pathways, monitor student progress, and alert advisers when students veer off that path.

Here are a few examples cited in the article:

  • Georgia State is using predictive analytics to identify nursing students who are at risk of dropping out. They’ve found that poor performance in introductory math, not a foundational nursing course, indicated a higher risk of failure and the need for intervention.
  • At the University of Arizona, it was assumed that a grade of C in freshman English composition was acceptable, but data showed that just 41 percent of students who made a C earned a degree. As a result, the school needed to devote more resources to writing, especially for C students.
  • Predictive analytics determined that a D in a required American history course was the most common denominator for students who didn’t graduate from Middle Tennessee State University. Because it is a reading-intensive course, more reading comprehension is required for students who struggle. Predictive analytics enabled the school to go deeper than grade-point averages and monitor performance in specific courses.

The first step to a successful predictive analytics initiative is to identify a problem to be solved and the questions that must be answered to solve that problem. For example, a high percentage of students in a particular major either drop out or transfer. What common thread links the highest number of these students? Next, determine the appropriate predictive analytics technique(s) that will provide you with data that enables you to make the predictions you want to make. You may find that your data needs to be cleaned and normalized to produce the most valuable insights. Lastly, arriving at conclusions is only part of the job. You must be prepared to act upon the insights produced by predictive analytics if you expect to see any improvements.

The ability to analyze past performance has proven valuable for colleges and universities. The ability to predict future success and take steps to reduce the risk of failure is a huge step toward delivering the kind of accountability that parents, students and politicians expect from higher education.