FS Data Science Hackathon with ECB

From Friday, September 20, 2019, until Saturday, September 21, 2019, Frankfurt School of Finance & Management hosted its first overnight hackathon in cooperation with ECB. The 25-hour event, supported by Google Cloud, took place at Frankfurt School’s AI Lab.

40 participants, students from Frankfurt School and external students as well as young professionals, were divided into eight teams with mixed skills levels to foster learning, collaboration and exchange. The teams were asked to develop an intelligent way to automatically assign categories, according to the COICOP classification scheme, to supermarket products already scraped from online sources. Their solutions have been evaluated on the basis of four different criteria: Accuracy, Generalisability, Innovation and AI Monitoring & Business Process Integration. The winning team's solution is a combination of simple machine learning models and a rule-based system designed to be highly accurate and easily generalizable without being hugely susceptible to classification errors that can arise from few training examples.

Judges and mentors were impressed by the high level of accuracy and praised the creative solutions of the teams. For participants, the hackathon was not only challenging but also created an opportunity to network.

The two winning teams presented their solutions at ECB, following the conference “Inflation in a changing economic environment” on Tuesday, September 24, 2019. "I was impressed by the interest of the ECB conference participants in our solutions", said Chukwuma Dim, PhD student at Frankfurt School.