You are here

Anomaly and Event Detection for Unsupervised Athlete Performance Data


Jim O'Donoghue, Mark Roantree, Bryan Cullen, Niall Moyna, Conor O’ Sullivan, Andrew McCarren

Publication Type: 
Refereed Conference Meeting Proceeding
Often the first step in data mining is anomaly detection, where outliers and unusual shifts in the data are identified. This is crucial in avoiding poorly fitting models in subsequent analyses for automatically generated, unsupervised datasets. Data was generated from sensor vests worn by Gaelic Football players over 17 matches, resulting in over 2 million instances. This analysis focused on one match with a 1 second moving average giving roughly 6, 200 instances. Such data renders manual anomaly detection traditionally employed by sports scientists intractable. Our algorithm detects anomalies for unsupervised, time series data and employs univariate and multivariate approaches.
Conference Name: 
CEUR Workshop Proceedings (
CEUR Workshop Proceedings (
Digital Object Identifer (DOI): 
Publication Date: 
Conference Location: 
Research Group: 
Dublin City University (DCU)
Open access repository: