Analysis code and derived results for Measuring the Effect of Instrumentation Coverage on Appliance Disclosure Using Gradient-Boosted Trees across 25 Households
Published: 15 September 2026| Version 1 | DOI: 10.17632/7mgw79zynr.1
Contributors:
Ninni Singh, Vinit Kumar GunjanDescription
The article asks what a smart meter record reveals about household activity, and shows that the answer depends heavily on something nobody reports: the fraction of a household's load that the dataset actually submeters. Appliances left unmetered still draw power, still reach the aggregate channel, and there they mask the appliances a classifier is trying to detect. Across 25 households that fraction ranges from 0.003 to 0.805, and completing a household's instrumentation raises measured disclosure in every one of them, by amounts that differ from home to home.
Files
Categories
Computer Science, Privacy