Stellenbosch University
Welcome to Stellenbosch University
Competition creates opportunities for students to develop practical skills
Author: Pia Nänny
Published: 02/11/2017

Stellenbosch students who participated in a data science competition hosted by Capitec in conjunction with the Department of Statistics and Actuarial Science not only had to create models to solve a problem, but also had to present and “sell" those models to the industry partner.

Students received a sanitised data set and were told to solve a real-world problem identified by Capitec. After creating their models, the participants with the top five scores had to present their findings to a panel consisting of four Capitec representatives and four SU lecturers. Their presentation scores were added to their initial score to determine their positions.

Brother and sister Elan and Chloe van Biljon joined forces to take first prize. Chloe has an undergraduate background in Actuarial Science and is currently busy with a master's degree in Economics while Elan is a fourth-year Electrical and Electronic Engineering student (Informatics stream).

“I have not found many problems where I can apply what I already know about machine learning while still learning new things. I found both of those in this competition and that is what made it worth it for me," said Elan.

In second place was Jandré Marais, a master's student in Mathematical Statistics. He valued the opportunity to apply theory on a real-world problem.

“It is very rewarding to find a solution to a problem and more so if someone can gain from it. It's an excellent opportunity to apply what you've learned and to see how your capabilities compare to others'.

“Hit-and-miss is a big part of the process. It's fun when one of your stranger ideas improves the model, because then you can investigate why it happened and learn something new in the process," he added.

According to Capitec's August Carstens this competition is an extension of their collaborative efforts with universities, providing more students with the opportunity of practical exposure.

“It was also an opportunity to share and spread our excitement and enthusiasm for data science," he said. “It was evident that all the participants put a great deal of time and effort into solving the problem at hand. Students showed interest in understanding the intricacies of the problem and came up with creative ideas on how to solve it."

According to him they ran one of the earlier versions of their model on the competition data. Two the solutions actually outperformed Capitec's earlier model.

“We have developed the first version of the solution and we are looking to continuously improve it in future. We definitely got some helpful insights from the students' work," he added.

Carel van der Merwe from the Department of Statistics and Actuarial Science regarded this competition as an excellent opportunity to apply theory in practice.

“Students also had to develop the additional skill of presenting their model to an industry partner as they would in a real-life situation."