SeQual: an unsupervised feature selection method for cloud workload traces

Abstract One challenge of studying cloud workload traces is the lack available users’ identities. Therefore, clustering methods were used to address this through extracting these identities from traces. For better extraction, it beneficial select attributes (columns in traces) for by using feature selection methods. However, use general requires details that are not (e.g. predefined number clusters). paper, we present an unsupervised method identify good candidate clustering. This uses Silhouette coefficients rank best extraction The performance our SeQual evaluated comparison with commonly (supervised and unsupervised) help quality metrics (i.e. adjusted rand index, entropy precision). results show can compete supervised perform than ones, average accuracy between 90% 99%.