//
you're reading...
Georgia, Higher Education, Oracle, Predictive Analytics, Public Universities, Student Retention, Valdosta State University

Predictive Analytics at Valdosta State University — and New Revelations on the Road to Retention

By 2020, a projected 60 percent of jobs in Georgia will require some form of a college education, whether a certificate, associate’s degree or bachelor’s degree. Today, only 42 percent of the state’s young adults — its burgeoning workforce — qualify.

Yet the answer is not simply opening the doors of academe a little wider. It’s not just about putting bodies in the seats; it’s about keeping them there, encouraging students to stay, providing the guidance that motivates them to complete their degrees.

At Valdosta State University in Georgia, nearly one-third of its freshman class doesn’t return for their sophomore year, darkening the students degree aspirations – and costing the school $6.5 million in lost revenue annually.

Deep analytics could provide an answer to the attrition. Like many higher education institutions, VSU was sitting on a mountain of warehoused data. But the challenge was how to put the information together in a cohesive and interlocked manner, to improve student success by using the data to boost decision making, build predictive models and provide actionable intelligence to deliver personalized support to students.

Says Brian Haugabrook, chief information officer at VSU, “We needed a solution that could rapidly integrate the many data silos and provide the capability to create a comprehensive business intelligence environment for academic, financial and HR information systems.”valdosta gal-1

In planning the implementation of its new business intelligence system, the university conducted a review of all the brand-name solutions available.  “We found the top selling tools at the time advertised out-of-box integration with 200-plus reports and 10-plus dashboards.  This initially became very attractive,” said Haugabrook. But after an in-depth analysis of the reports, what the search committee found was that the out-of-box solutions only provided simple reports that were already available in the university’s student information system. “We also found that creating new reports or expanding the out-of-box data structure was very cumbersome and time consuming.  Some technologies would require us to fund consultants to make major changes,” said Haugabrook.

The answer was found in Oracle’s business information suite of tools, which would provide the university with a foundation that could rapidly deploy new models specific to predictive metrics and, ultimately, improve retention. Combining Oracle Business Intelligence Enterprise Edition with Oracle Endeca Information Discovery allowed VSU to create a complete, actionable intelligence environment. Said Andy Clark, vice president of enrollment, marketing and communications, “We could take the siloed data in the different shops, break down the barriers, and combine structured and unstructured student data—like student surveys and ID card usage data—in one system for analysis.”

New business processes provided actions behind the information. Faculty and advisors could trigger intervention processes within the business intelligence platform. New reports and dashboards could be created in a few hours.  Departments were empowered to extract data themselves without the need for an IT programmer; that benefit alone saved over 45 hours weekly in IT support.

Not only was the university looking at reports and dashboards but actionable alerts that initiated processes to ensure students stayed on a path towards graduation. “Information without action and automation cannot improve the business goals,” said Haugabrook.

The original Oracle business intelligence tool took nine months to install for development; the big data solution, Oracle Endeca Information Discovery, took only three months. New reports and features are constantly added, based on user feedback.

The initial at-risk model was built around high school performance, high school GPA and SAT/ACT scores.  This model gave VSU predictions on students who were at-risk in math- and reading-based courses.  “Our math model restricts registration based on risk.  A student can no longer register for a math course such as pre-calculus if they are predicted to fail,” said Haugabrook. The model improved math pass rates by over 10% in some courses by ensuring students got the skills they needed to be successful.

Said Joe Burkhart, director of Oracle’s education and research global industry solutions, “What the university got was a true data mining tool that calculated probability about several things. They can push the information out on their own. They don’t have to give it to us. Other proprietary solutions would charge the university. What results is an entire ecosystem under their own control”

Now, each department across campus can generate reports and analysis specific to their area. Not only does it boost efficiency in retrieving information, but it automates many of the business processes. What’s more, faculty can easily communicate with each other, quickening the process of flagging issues related to retention and helping to deliver guidance to students in a timely fashion. Moreover, faculty can receive recurring reports by email, eliminating the need for administrators to log in and run reports at specific times during the semester. “We are integrating information into our data warehouse to create a day-to-day at-risk model,” said Haugabrook.valdossta 2

The model monitors when students login, how much time is spent on course topics, the time to complete assignments and other activities. It improves the risk models by being proactive in reaching out to students the same day they get off track for passing each course.

The faculty portal provides an easy alert feature, so if a student is flagged for performing poorly or not coming to class, notifications are sent to everyone on campus that could proactively influence performance, from advisors and the student success center to athletic coaches if the student plays a sport or the housing coordinator if the student lives on campus.  Said Haugabrook, “We have received feedback from students stating that they felt the university cared when so many individuals reached out to them to offer support.”

“Our student portal is also innovative,” said Clark. Personalized information based on large amounts of data is pushed out to each individual.  Advertisements, or blocks of information on the web interface, focus on student success and information they need to keep them motivated to stay in school and move them closer to graduation. “Communications with the student are created on a very granular level, down to the individual. Eye-catching ads talk directly to them, so that, if a student registers for algebra, for example, and there’s a likelihood of them struggling, ads will direct the student to resources that can help them succeed.”

Despite the highly targeted communications, some students still find themselves off course. Identifying these students early and reaching out to them immediately is critical to getting the student back on the road to retention. “We aggressively go after students with our early intervention strategy, where students are directed to the right resources at the right time, as soon as possible, say in the third week of class rather than the eighth,” said Clark.

Combining structured data from the warehouse with unstructured data from various other systems is producing reports that help VSU identify trends and disclose if certain metrics influence student success. Some of the findings verified commonly held beliefs; for example, students who spend time in the library are more successful – a finding found in the Oracle big data tool by filtering students who printed documents in the library. Other reports revealed unexpected connections. “Students who eat breakfast between 6 and 9 a.m. have an 85-90 percent retention rate,” said Haugabrook.  “This is only a correlation as eating breakfast alone does not make a student successful but it was very interesting given our overall retention rate is 67 percent.”  This correlation gives the university a population it can further analyze to find causation.valdosta classroom

Among other revelations, the combined data showed that students who get jobs as freshman have an 83-90 percent retention rate.  “We found that most jobs are advertised over the summer.  New freshman arrive in August and were unaware of the new jobs,’ said Haugabrook. Consequently, the university allocated additional funds for on-campus jobs, began to advertise them more and hold some specifically for freshman.  “This was a low hanging fruit that will provide a huge benefit as all it required was better communication,” said Haugabrook. “One item alone may not improve success but our ability to combine multiple activities allows us to put a value on engagement.”

As a result of the university’s targeted communications program, and the Oracle suite’s combining of structured and unstructured data, VSU has improved retention 4 percent in just two years. “We are currently in year three and project this will be our best year, estimating 2.5-3 percent improvement in retention” said Haugabrook.

Advertisements

About markwalston

Writer, historian, creative director, poet, playwright, author of nine books and nearly 200 essays and articles exploring a broad range of American social, cultural and historical topics.

Discussion

No comments yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: