Bayesian Modeling of Uncertainties in End-Use Electricity Consumption Amounts Inferences

• Activity Number: 509 - Methodological Innovations and Applications in Government Statistics
• Type: Contributed
• Date/Time: Wednesday, August 2, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Government Statistics Section
• Author(s): Hiroaki Minato*
• Companies: U. S. Energy Information Administration (EIA)
• Keywords: administrative data ; Bayesian multilevel measurement error models ; end-use energy consumption amounts ; Residential Energy Consumption Survey (RECS) ; Stan ; survey weights

Abstract

Minato (2016) approached the problem of statistical calibration of energy-engineering-expert-model (E3M) estimates of residential electricity end-use consumption amounts with Bayesian multilevel models. The Residential Energy Consumption Survey (RECS), conducted by the U. S. Energy Information Administration (EIA), provided the data on building characteristics as well as energy end choices and uses. With the survey data and weather data, engineering models were formulated to estimate various end-use energy consumption amounts. However, the Bayesian multilevel models did not incorporate the engineering models' estimation errors or the administrative billing data's processing errors. In this paper, we directly model the uncertainties in those errors within the Bayesian framework. Survey weighting errors are also modeled for population inferences.