Hurdle Models for Residential Solar Photovoltaic Systems Installations

• Activity Number: 603 - The Importance of Statistical Analysis in Energy Modeling
• Type: Topic Contributed
• Date/Time: Thursday, August 3, 2017 : 8:30 AM to 10:20 AM
• Sponsor: Government Statistics Section
• Author(s): Janice Lent*
• Companies: U.S. Energy Information Administration
• Keywords: hurdle model ; count data ; negative binomial regression ; logistic regression ; GAMLSS

Abstract

We describe research on the use of hurdle models for projecting the number of installations of residential solar photovoltaic (PV) systems in the United States. The U.S. Energy Information Administration (EIA) publishes detailed energy-related projections annually in its Annual Energy Outlook (AEO). The 2017 edition of the AEO provides projections to 2050. The EIA uses its own National Energy Modeling System (NEMS) to produce AEO projections. Zero-inflated models and hurdle models have been used to model count data in various settings, including public health and econometric applications. Rothfield (2010) used hurdle models to identify significant drivers of residential PV installations in California, including economic and social effects. We use the gamlss package in R to fit hurdle models, incorporating logistic and negative binomial components, to zip code level residential PV installation data from selected states. We combine the model coefficients with projected variables from the NEMS to project future PV installations. The projections are aggregated to the Census Division level for publication.