Explanation of smart energy consumption of Hong Kong households via shapelet-based time series clustering
About
Time series shapelets are discriminative subsequences that have been recently found not only effective for time series classification but also powerful for the clustering problem of time series. This project investigates the shapelet-based approach for solving the problem of time series clustering (TSC). In this project, we propose a novel approach, which integrates the advantages of both shapelets and representation learning for determining high-quality shapelets in an unsupervised manner. Next, this project explores the extent to which and how time series clusters (hence, the electricity customer segmentation and their major behaviour derived from shapelets) can enhance the precision and effectiveness of low-carbon policies in supporting Hong Kong to meet its goals of smart energy. We aim to co-develop innovative and practical engagement strategies with communities to enable deep engagement, thereby enhancing community performance on energy-saving and possibly, the use of solar power.
Required skills
Machine learning.
Principal Investigator
Associate Professor, Department of Computer Science
Co-Investigators
Assistant Professor, Department of Geography