Danijel Kivaranovic, a PhD student at the Department of Statistics and Operations Research, has won a machine learning competition on the popular data science platform Kaggle. He shares the 20.000 USD prize with his teammates.

The goal of the competition, which was hosted by the Los Alamos National Laboratory, was to predict the time to the next laboratory earthquake from laboratory real-time seismic data. Danijel and his team extracted intelligent features from high-frequency acoustic data and trained several complex machine learning models (neural networks and gradient boosting machines) that achieved the best predictive accuracy among the more than 4.500 participating teams. A complete description of their solution, written by his teammate Philipp Singer, was posted here.

This is only the most recent success of Danijel Kivaranovic, who is Kaggle Grandmaster with more than 5 years of experience in machine learning competitions. The Grandmaster tier is the greatest honour that is awarded by Kaggle and is currently held by only 162 data scientists. Danijel has won 6 gold medals, which means that he has achieved top-ranked results in 6 different competitions. (Among these where two 1st place, one 3rd place and two 5th place finishes). His best overall ranking on the Kaggle leaderboard was 41st position. The Kaggle leaderboard reflects the competition performances of more than 100.000 data scientists worldwide.