The paper “Water Consumption Pattern Analysis Using Biclustering: When, Why and How”, authored by LASIGE’s PhD student Miguel G. Silva has been published in the Water Journal, a top-ranked journal in Water Supply and Treatment (h5-index 63; Google Scholar #5). The paper co-authors are LASIGE’s integrated researcher Sara C. Madeira and Rui Henriques from INESC-ID and Instituto Superior Técnico.
The deployment of smart grid meters in water distribution systems (WDSs) has increased data collection, enabling data analytics tools to extract essential information. Clustering techniques can be used to model end-use water consumption profiles, allowing water companies and municipalities to know better what to expect from customers. Nevertheless, due to user diversity and the local nature of consumption profiles, some challenges limit its effectiveness, as clustering techniques focus on global consumption patterns. To this end, in this work, biclustering approaches are suggested for identifying statistically significant, actionable, and interpretable consumption patterns because they offer unique benefits over traditional clustering techniques. The authors demonstrate the utility of biclustering approaches for analyzing water consumption profiles and present a taxonomy for a structured view, principled application and critical assessment of biclustering water consumption data. This is the first work applying biclustering on water consumption data.
The proposed approach for pattern discovery has been validated using data from a real-world water distribution system in Quinta do Lago, Portugal. Experimental results show the relevance of biclustering for the analysis of consumption time series data, with statistical significance and robustness guarantees, which water management entities can use to model water consumption profiles, opening up new opportunities in this field.
The paper is available here.