How IBM Garage helps improve predictive analysis

By | 2 minute read | December 18, 2018

When we started working with the IBM Garage team, we never imagined we’d turn our whole system on its head, find new purposes for data we were about to discard, and improve our predictive capabilities.

We developed the XComP (eXtensible Competencies Platform) system at East Carolina University (ECU) in 2011. The XComP system helps schools assess outcomes by generating cumulative, real-time evaluations of students, faculty, curricula and programs. In 2017, we realized we needed to evolve our original concept into a common platform that used the same data capture and normalization to provide similar reporting across disciplines, and we started a company called XComP Analytics as a technology transfer from ECU.

Visiting the IBM Garage expecting a platform “tune-up” …

I didn’t know what the term “Garage” meant at first, but a vendor we worked with introduced us to IBM and said, “You need to talk to these people, because they know analytics better than anybody.” He was right.

Experts from the Garage team evaluated our working system and found ways to improve the platform and simplify development processes. The IBM Garage Method identified scale issues, found discarded, usable data and determined a better cloud architecture. One of the greatest aspects of working with an experienced corporation was the collaboration among technologies at IBM. Garage Consultants used a microservices architectural style and plugged in IBM Analytics into specific areas of our tool.

… and driving out with a supercharged analytics system

We’d worked with several software companies to solve various issues with the platform, but we finally fixed them with the IBM version of XComP. We also got six new analytics environments because the Garage looked at the data differently. There’s no question that the Garage makes you rethink things that you didn’t have before. For example, the Garage helped me give context to projects we were doing and put them into four big data categories. This helped us streamline what the various minimum viable products (MVPs) needed to be.

We participated in an IBM Design Thinking workshop and restructured the MVP sequencing three times. In terms of speed, I think the most important thing was to measure twice before you cut once. If you didn’t have the design right, then the programming would run into issues. We checked the design and put the sequences where they needed to be several times before we started coding. A lot of the time savings was in up-front thinking.

New predictive power under the XComP hood

Our new system can predict outcomes that weren’t available before we worked with the Garage. Every day, XComP culls thousands of pieces of data from students’ exams, skills assessments and other coursework and displays them on an easy-to-read grid that identifies strengths, weaknesses and deficits. Our system uses this data to predict performance for decades to come, which allows students to be more proactive in their education, instructors to intercept high-risk students, and institutions to improve educational programs.

We’ve been so happy with the first phase of this project that we plan to work with the Garage in Austin, Texas, to further develop each of the MVPs. IBM transformed our tool, and we look forward to our future endeavors with the Garage.

How XComp built its world-class competency-based analytics platform for education and training with the help of IBM Garage:


Most Popular Articles