UNSW’s Data Science Society DataSoc organised the competition jointly with Tsinghua University’s Institute of Data Science, and their Big Data Association, seeking to unite and foster growing international talent in the rapidly evolving field of data science.
“We wanted to bring together students from diverse backgrounds with a wide range of skillsets to solve two simulated real-world problems using data science techniques,” says Christopher Shi, DataSoc President.
“The goal of the datathon is to help students build the skills they need out there in the workforce – and that’s not just coming up with a fantastic solution to the challenge, but also being able to present its benefits in a convincing manner.
“As a larger aim, we want to promote diversity and collaboration on an international scale, establishing cross-border relationships between UNSW and Tsinghua University.”
On the morning of the first competition day, the teams headed to Servian HQ to be given a huge, complex dataset containing 60,000 photos of clothing. They were tasked with developing a model that would be able to take any image and identify what kind of clothing it was among 12 categories (shorts, shirt, skirt, pants, etc).
“We asked the students to run their model on the 60k photos and the scoring system would compare their results to a set of pre-generated (correct) results and determine how good their model is in predicting what piece of clothing is in a picture,” Christopher says.
After the 12 hour-datathon, the work wasn’t done, though – the teams then headed to Atlassian HQ in Sydney to present their solution to a panel of industry judges and university professionals with expertise in data science. The judges looked at how precisely the problems had been solved, and also evaluated the participants’ teamwork, communication and technical skills.
An engaged audience of first-year students watched on as their more senior counterparts presented their data model solutions.
Team Cache, Tsinghua University took home first place, followed by Team Hide (Tsinghua University) and Team Alphamerge (UNSW).
Second year and going strong
The organisers were pleased with this year’s event, saying the whole experience – having talented contestants from both sides, an awesome week hosting and showing them around Sydney, all led by an exceptionally capable team – had been amazing.
“The week has been much more than a competition – because the delegation consisted of students as well as faculty members, we’ve been able to facilitate introductions between the two universities to build further relationships that will help UNSW students become more internationally-minded and well-travelled,” Christopher says.
“The Datathon was a fantastic segue for making the first step in creating a fruitful and promising international relationship between the two universities. It was a fantastic showcase of talent to our industry partners, and to the world. I definitely think it’s one of the highlights of the year.”
Last year, DataSoc put on their first ever data science datathon. After a UNSW student went on exchange to Tsinghua and fostered great relationships with his data science society counterparts there, the datathon became a multi-university collaboration. Earlier this year, both universities held their preliminary rounds, with three teams each making it through to the finals.
“The data science industry is growing and there’s a strong demand for capable and talented graduates entering the workforce,” Christopher says.
“After two years of success, we’ll definitely run our datathon again next year – I can’t wait to see what challenges our participants will be solving then.
“We would like to thank the UNSW Business School, UNSW School of Mathematics and Statistics, Optiver, Advance AI and Syrius for sponsoring the datathon and making it happen. A special mention must also go out to Servian and Atlassian as well as all of our other partners for their immense support this year.”
By Isabelle Dubach