The Impact of Temporal Granularity on Socio-Demographic Inference from Household Load Profiles
Researchers have explored how the temporal granularity of household electricity usage data impacts the ability to infer socio-demographic characteristics. Their study, using a year of data from 1,589 households, found that predictive accuracy remains stable between 15-minute and 1-hour intervals, and again between 1-day and 7-day intervals, suggesting opportunities for data minimization without significant utility loss. The research also indicated that handcrafted features and XGBoost classifiers performed competitively, and that different types of socio-demographic attributes require varying levels of data granularity for accurate inference. AI
IMPACT Provides insights into privacy-utility trade-offs in smart metering data, informing how granular data collection impacts inference capabilities.