Web and Internet Science, University of Southampton

Web and Internet Science

ECS Students win UK Data Science Challenge

Extract from HypeTrain's winning entry
Extract from HypeTrain's winning entry

A team of MSc Data Science students from Electronics and Computer Science (ECS) at the University of Southampton have recently competed against 170 students from across the UK to win the UCL Data Science Student Challenge.

The competition a collaboration between UCL (University College London) Department of Computer Science and Microsoft – and sponsored by Bloomberg – and is designed to ignite a passion for the amazing things made possible by Data Science. It was themed around the development of innovative data science solutions with large financial data-sets in order to give Bloomberg a competitive advantage in the finance industry, while demonstrating use of Microsoft’s Cortana Intelligence Suite. Students had the opportunity to work on deep learning, text-mining and NLP to extract information from news articles in real-time.

The challenges also focussed on the use of machine learning and classification of news articles for ranking and sentiment trend analysis, while demonstrating the impact of news and events on asset prices, as well as understanding the trends between various instruments in financial markets.

Teams were judged on their ability to identify significant relationships between instruments, news reports and trends, as well as illustrating the strength of those relationships. How that information could be used to enhance decision making, and identify generalised patterns of events were also important considerations.

The winning team, HypeTrain – Uktu Ozbaluk, Alex Young and Lukas Weiss, from ECS, came up with a forward indicator, which isn’t currently offered by Bloomberg. Their solution can be seen here on GitHub.

Team member Utku is clear that information studied during the course contributed to their success: “We mainly followed what we had been taught through group project meetings and visualisation lectures. We focused heavily on an interesting story and presented it in a simple and straightforward way which distinguished our approach from other groups.”

Posted by gb2r13@sot on 10 May 2017.