The first university for Data Science in the Netherlands, JADS, is up and running. Team Rhiknows won the first hackathon for finding ways to protect rhinos against poachers! The information presented in Brabant Brand Box is offered free-of-charge for positive stories about Brabant.
How can we protect rhinos against poachers using big data? This is the theme of the very first Wildlife Hackathon organised by Jheronimus Academy of Data Science (JADS) on 7 July 2017 at the Mariënburg Campus in ’s-Hertogenbosch. More than fifty students and researchers are to work for ten hours non-stop with data sets provided by Welgevonden wildlife reserve in South Africa. Four students of Econometrics and Operational Research at Tilburg University are among the many data scientists. Can they succeed in coming up with the winning solution?
This is the first time that any of the four have participated in a hackathon. “We wanted to give it a try and were curious about how we would perform amongst all those data scientists”, say Belle de Veer, the only woman in ‘team RhiKnows’. “But the most important reason for applying was the opportunity of developing something capable of saving wildlife and making the world a slightly better place.”
The data that is to be used by the students and researchers is presented to them at Mariënburg Campus (a former monastery) at 8 o’clock in the morning on 7 July. The data relates to the movements of prey animals such as zebra, wildebeest, eland and impala in Welgevonden wildlife reserve. In collaboration with Wageningen University & Research, the park has attached tracking collars to these animals, which transfer data, including GPS locations, to an ICT platform. Displays of unexpected behaviour by the animals may indicate the presence of poachers. Because no real data sets of any usable size exist as yet, the ‘hackers’ work with simulated data. For example, a data set that simulates 10,000 time units and contains the location of the animals per time unit and the distance they travelled per second. A further data set records the times when the animals change direction and the angle at which they turn.
Most teams start by designing a model, but team RhiKnows adopts a different approach. The four start with a data analysis: what kind of data do we have available to us? They soon discover that the animals start to cover significant distances when poachers are nearby. Belle: “We designed our model based on that discovery. And because we discovered this relationship early on, we were able to spend a relatively large amount of time on improving the model.” However, that early Eureka moment also raises doubts: surely it cannot be that simple? Furthermore, they find it difficult to identify other variables that predict the presence of poachers with any reliability. Belle: “We really struggled with that problem: how can we make our model truly effective?” Finally, they succeed in developing a model that not only allows Welgevonden to see when poachers are actually in the reserve, but also identify their location.
The session has almost ended and things are getting tense: ‘training’ the model has gone very well, but testing turns out to be more difficult. Only at the very last moment do the students succeed in generating the test data they require to show the accuracy of their results. The idea that they may actually win the competition never enters their heads. “That possibility only dawned on us after the judges’ initial assessment, when we heard that we were through to the next round.” Heartbeats rocket again just before the final presentation: the team's choice of data visualisation method suddenly refuses to work. “The four of us sat nervously around my laptop. Everything depended on me now; if I failed to find the right button, our chance of winning this hackathon would evaporate into thin air.”
Belle presses the right button. And to everybody's huge surprise, team RhiKnows wins the first JADS Wildlife Hackathon. Belle: “We could hardly believe it. Winning a hackathon is pretty special anyway. But winning a prize like this is really something…” And that prize is definitely worth all the effort. In October 2017, Daan Marechal, Maurice Peters, Joshe Klaver and Belle de Veer pay a visit to Welgevonden wildlife reserve. A fantastic experience, says Belle. “We went on safari several times every day. We saw countless magnificent animals and learned so much from the biologists and ecologists who work at the reserve.” Team RhiKnows intends to carry on supporting Welgevonden and now handles data analysis for the wildlife reserve. Belle: “South Africa has won a special place in my heart. If I am ever offered the opportunity of using my expertise to make a difference there, I certainly will.” She is also enthusiastic about JADS. “The hackathon was well organised, the atmosphere friendly and relaxed and the Master courses offered by JADS are highly contemporary and innovative. I’ve even started to nag my brother now: ‘Go to JADS if you want to study data science!’
JADS is a collaboration in the field of big data that has been set up by four parties in North Brabant: Tilburg University, Eindhoven University of Technology, the provincial authority of North Brabant and the municipality of ’s-Hertogenbosch. The university course for a degree in data science – the profession of the future - links technical innovation to innovative trends in the social sciences. JADS offers Bachelor and Master courses at three locations. The emphasis lies on combining big data and business administration at the Mariënburg Campus in ’s-Hertogenbosch. Students and researchers collaborate closely with businesses at this campus. The arrival of JADS has brought many more (international) students to the provincial capital.
This website uses functional, static and social media and other cookies. If you want to change the cookies and scripts that are used, you can change your settings below.