Quantified self: building a multi-time window analytical workflow for clustering wearable data streams
University of New Brunswick
A new research domain known as the Quantified Self has recently emerged and is described as gaining self-knowledge through using technology to acquire spatio-temporal data on our behavior. Wearable technology is widely used in this domain since it generates a large amount of wearable stream data, which contains information on self-monitoring activities and physical health related problems. However, very little is known about which stream clustering algorithms should be used and which time windows can reveal individuals' spatio-temporal patterns that can yield new self-knowledge insights. This thesis proposes an analytical workflow developed to reveal self-quantified patterns that can be used to understand physical activity behavior. It consists of six phases that are devised to support tasks including retrieving, processing, and clustering wearable data streams. The streaming k-means clustering algorithm, based on an online/offline approach using both sliding and damped time window models, is proposed to uncover self-quantified patterns. An intervention experiment with 15 participants is used to gather Fitbit data logs and implement the proposed analytical workflow. The clustering results reveal the impact of a time window model has on exploring the evolution of micro-clusters and the labelling of macro-clusters to accurately explain regular and irregular individual physical behavior.