Advanced household profiling using digital water meters

Abstract

Advanced householder profiling using digital water metering data analytics has been acknowledged as a core strategy for promoting water conservation because of its ability to provide near real-time feedback to customers and instil long-term conservation behaviours. Customer profiling based on household water consumption data collected through digital water meters helps to identify the water consumption patterns and habits of customers. This study employed advanced customer profiling techniques adapted from the machine learning research domain to analyse high-resolution data collected from residential digital water meters. Data analytics techniques were applied on already disaggregated end-use water consumption data (e.g., shower and taps) for creating in-depth customer profiling at various intervals (e.g., 15, 30, and 60 min). The developed user profiling approach has some learning functionality as it can ascertain and accommodate changing behaviours of residential customers. The developed advanced user profiling technique was shown to be beneficial since it identified residential customer behaviours that were previously unseen. Furthermore, the technique can identify and address novel changes in behaviours, which is an important feature for promoting and sustaining long-term water conservation behaviours. The research has implications for researchers in data analytics and water demand management, and also for practitioners and government policy advisors seeking to conserve valuable potable-water resources.