Online Program

A Framework for Unsupervised Outlier Ensemble Detection: A Case Study of the Nigerian Football Players’ Performance Statistics (303471)

Recruitment of players/athletes into the Nigerian National team is made solely subjective and heavily reliant on the instinct of the person doing the selection. Instinct, though important, may not always be enough for consistently fool proof selection process. In this work, we demonstrate that, outlier detection analysis can supplement instinctive judgement with evidence rooted in available data, by recognizing players that stand out or have exceptional skills. An unsupervised ensemble-based outlier detection method is constructed by unifying outputs from three (3) outlier detection methods, Local outlier factor (LOF), Angle-based outlier degree (ABOD) and Subspace outlier degree (SOD) via Regularization and Gaussian scaling. We also present a heuristic framework for prediction and quantitative performance evaluation of the Ensemble. The Ensemble is applied to the Nigerian football players’ performance statistics data. The detected outlier instances were qualitatively evaluated by a sports analyst (expert) confirming the usefulness of the proposed framework in identifying even the unexpected instances as well as unusual special cases.