This year saw the introduction of the annual RSS Statistical Analytics Challenge, a new joint venture by the Young Statisticians Section and the Research Section of the Society. The 2014 Challenge was kindly sponsored by Select Statistics. For six weeks beginning in May, teams of intrepid analysts from across the globe battled it out in the “ultimate test of analytic skill, teamwork and creativity”.
The Challenge brief sets out an open-ended research question relating to a large and complex data set. The goal of the Challenge is for each team to collaboratively explore and analyse the data – using any methods and tools that they wished – and submit a short report summarising their main findings.
Nineteen teams from around the world registered for the 2014 RSS Challenge. Teams were encouraged to include a diverse range of members – male and female, career-young and career-not-so-young – and this diversity was clearly evident in the majority of teams.
This year, the brief focused on the ‘Perils of Data Dredging’, and the Challenge was to produce the most innovative analysis of a set of neuroimaging data (specifically, resting state fMRI data), whilst “highlighting along the way the potential pitfalls for blind or naïve analysis”. No previous knowledge of brain imaging was needed to analyse the data, and teams were free to use as much or as little of the data as they liked (though a minimal data set was established). This freedom was embraced by all teams, and a wide variety of approaches were considered.
The winners of the 2014 Challenge were announced on 4 August 2014.
The Overall Winner was Team UCLStats (Beate Franke, Sam Livingstone, Alfredo Kalaitzis from University College London and Michael Betancourt from Warwick University) with their report entitled “Long-Range Spatiotemporal Dependence of fMRI Imaging Data”.
Team SODA (Chris Brignell, Phillip Pane, Wilhelm Braun and Heather Pettitt) from the University of Nottingham was awarded “Highly Commended Runner-Up” with their report on the “Effect of smoothing halfwidth on estimated network structure in single-subject fMRI data”.
Both teams were invited to present their findings at the YSS and RSS Statistical Analytics Challenge Session at the RSS 2014 conference in Sheffield on Wednesday 3 September 2014. Certificates were presented to both teams by Prof. Chris Jennison at the beginning of the session; this prize-giving was followed by excellent presentations from both teams and a lively discussion with the audience.
With such a excellent response and exciting findings, we are very much looking forward to the RSS Statistical Analytics Challenge 2015 – so assemble your team, embrace the competitive spirit, and get ready to accept the Challenge!
Reviewed by Órlaith Burke.