After four months of competition, the winners of the Data Challenge have
been decided. 58 teams from 37 different universities participated in the
project. The objective: To create an algorithm capable of identifying the same
user on several different devices. The Parser team, a partnership between
William Parfitt of the University College of London and Robert Fraser of Oxford
University, won the first place on the podium.
MEC Belgium and DATAA organized the award ceremony of the MEC
Data Challenge on Friday the 14t of July. From February 24th to June 25th, no less than 37
universities mainly from Europe, 58 teams and 74 participants competed to win this
challenge. Originally, however, the MEC Data Challenge challenged only four universities:
the VUB, the UGent, the ULB and the University College of London. Damien D'Ostuni,
Innovation & Activation Expert of MEC Belgium, explains: "We are surprised at the
enthusiasm that the challenge has brought about in so many universities. Especially with a
challenge this size! The students had to create an algorithm to be able to follow the same
user on the different devices (tablet, desktop, smartphone) used during the day. The
objective of this algorithm is to allow us to optimize the digital campaigns and thus avoid
irritating the user with the same advertising message. In addition to this media-related
objective, the algorithm will also improve the overall future of IoT-related technologies.
With an accuracy rate of 84.6%, the algorithm created by William Parfitt of the University
College of London and Robert Fraser of Oxford University was the most effective. The two
Englishmen climbed to the top of the podium and won the sum of 3,000 euros. They were
followed by the FiKoD team, made up of three students from Wroclaw University in Poland,
with 81.8% accuracy. Finally, the final step of the podium is attributed to Tom Bolton of the
University of Oxford, with an accuracy rate of 80%. The first Belgian university comes in fifth
position with the UGent.
At the same time, DATAA's Data Scientists also took part in the out-of- competition
challenge. Their solution reached 90.4% accuracy. François Bordes, CEO of DATAA,
confirms: "We succeeded in developing an efficient solution, surpassing those of the
participants both in terms of accuracy and in terms of speed. However, the three winning
algorithms of the MEC Data Challenge have been tested by our team and have already been
implemented on our servers. The results of this challenge are very promising for the future of
innovation in big data. The algorithm will allow us to better target individuals and is an
important step in our development in Predictive Marketing as well as in Artificial Intelligence."
This collaboration between MEC Belgium and DATAA definitely marks a further step
forward in the development of Predictive Marketing and in the creation of new media
solutions for the agency.
