Thermostat metrics derived from HVAC cycling data for targeted utility efficiency programs

New data streams of household energy consumption offer opportunities for utilities to understand their customer base and differentiate between customers. From residential HVAC cycling data and weather data, we derive metrics that can be used to indicate out of a population which households are most likely to have extraordinarily high or low thermostat setpoints, which households are most likely to have thermostat setpoints that change at fixed times each day, and which households could benefit from a change in the hysteresis setting of a thermostat. We demonstrate these metrics using air conditioner data from a set of houses in Austin, Texas. With surveys as validation, we show metrics with accuracy greater than 70% that can be used for targeting utility energy efficiency programs. We show two examples where targeting based on these metrics can substantially increase the mean percentage energy savings achieved for a fixed cost.