Occupants’ behavior has a significant impact on building energy consumption, and data mining methods are commonly used to extract their behavior patterns. However, clustering algorithms based on “Euclidean” distance such as K-means have limitations in characterizing occupants’ behavior over time. Although some studies have begun to use time-series clustering algorithms, the resulting data characteristics differ due to occupants’ persistent (energy demand) and non-persistent (heat demand, cold demand, and lighting demand) features.
Previous studies focused more on analyzing household energy data using time-series algorithms than non-persistent behavior. Thus, this study compared three time-series clustering algorithms (K-shape, Dynamic Time Warping (DTW), and Derivative Dynamic Time Warping (DDTW)) in terms of occupants’ heat demand behavior while proposing three evaluation methods.
The results showed that although all three algorithms achieved clustering effects for non-continuous behavior (heat demand), DTW captured heating demand pattern changes more accurately. However, these three types of time-series algorithms did not significantly outperform the K-means algorithm in clustering effects due to the discontinuity of heat demands. Additionally, the proposed evaluation methods performed relatively well compared to commonly used metrics for evaluating clustering effects.
Overall, this study provides insights into selecting appropriate clustering algorithms for studying non-persistent residential behavior and has potential value in improving energy prediction and management.
