Saga Hackaton 2019 | Saga - New Frontier Group

The second 24-hour Saga Hackaton has been finished

The main theme of this year's hackathon was Forming customer segments in line with the strategy

Satisfied with last year’s hackathon, this year we have decided to support the Brand New Engineers project, organized by EESTEC LC Belgrade.

At this year’s Hackathon held on December 14th and 15th at the premises of our company, 27 participants were divided into 6 teams. The participants were from the Faculty of Electrical Engineering, the Faculty of Technical Sciences, the Faculty of Organizational Sciences, the Faculty of Mathematics and the Faculty of Information Technology.

Initially Milen Janjic, Director of the Own Software Products Development Division, and Milan Jevtic, Director of the Software Division have spoken to the present.

Milen emphasized the importance of Hackathon and the education that participants receive through participation, but not the education that is taught in college, through the participants had the opportunity to learn how to create software that is used, stating that International Property Software involves a high level of creativity and the creation of new values .

The main theme of this year’s hackathon was Forming user segments in line with strategy (Unsupervised Learning). The Hakaton participants had mentors who were from our Machine Learning team and were available to them throughout the Hackaton: Milos Bijanic, Tamara Ciric and Srdjan Jovanovic. The very idea was that the participants were working on two realistic machine learning foundation projects that our colleagues were working on. Competitors were provided with the materials they needed to create a business case and technical solution.

The first part of the Hackathon was related to the banking industry, where they were challenged to deal with the problem that our colleagues were internally working on, where candidates had the opportunity to work on real data from the banking industry, that is, data used for our CRM platform Selecta.

The second part of the assignment was based on the location data of our G # project, where the idea was to create segmentation and movement of one user through the mall. Based on the behavior of shopping mall visitors, a segment is created to aggregate data on their movement, seasonality of movement, retention, the zones they visit.

The purpose of the data used during the hackathon was to create customer segments of the banking sector and shopping center, in accordance with a specific marketing strategy. In addition to the ML part being focused on the use of specific models to perform segmentation, some business value had to be created, that is, whether the solution could be used, upgraded and implemented. When voting, 30% of the rating was related to that business value, and 70% of the rating was a technical solution and how they handled the ML structure and technology, as well as an interface that displayed performance.

Although all the teams presented good ideas and solutions to the task, for which they had only 24h, the jury unanimously and with the highest rating declared the winner team Jagoda, whose members were Iva Milojković, Gojko Čutura, Nikola Kesić, Dušan Jovanović and Jana Kovačević . In the second place was the Otpisani team and the third place was taken by the Šurikeni team. All praise also for the other teams that participated in our Hackathon: Lads, Prirodna Neinteligencija and Sagsofon.