Retention | News

Marist College Deploys Analytics for Retention

• By Leila Meyer
• 09/25/14

Marist College in Poughkeepsie, NY has developed an open source predictive analytics model that uses big data to help identify students at risk of not completing their coursework.

Students, faculty and staff at the college researched, developed and implemented the model as part of the college's Open Academic Analytics Initiative (OAAI), which was founded with the support of a $250,000 grant from Educause's Next Generation Learning Challenges program. Two major objectives of the project were to release the predictive models through an open source license and to collect significant research data on the effectiveness of predictive analytics and intervention strategies on student performance.

The OAAI used Pentaho Business Analytics, an open source business analytics platform from Pentaho, to support its academic early alert system. The predictive model mines student aptitude data, learning management system event-log data and electronic gradebook data to identify at-risk students. Within two weeks of the start of a course, the model can predict which students are unlikely to complete the course successfully.

According to the company, the model uses Pentaho's plug-ins and filters to automatically collect data from multiple sources. The system uses that data to generate course-specific Academic Alert Reports, which alert faculty members to at-risk students. The faculty members can then intervene by providing tutoring, online educational resources or other supports.

The OAAI has deployed its academic early alert system to more than 2,200 students at community colleges and historically black colleges and universities (HBCUs) around the United States. The predictive analytics model was able to identify at-risk students successfully 75 to 79 percent of the time.

The OAAI has released its predictive models under an open source license and it has been implemented at four colleges and universities so far.