New York Tech Journal
Tech news from the Big Apple

Promoting #solar technology diffusion through data-driven behavior modeling

Posted on March 19th, 2015

NYC EnergyData

03/19/2015 @Urban Future Lab, 15 MetroTech, Brooklyn, NY

20150319_185434[1] 20150319_190656[1]

Kiran Lakkaraju @Sandia Labs presented models to better understand when people install #SolarPanels on their homes. He emphasized the importance of understanding the economic, social, technical, and cognitive factors when people decide to purchase solar panels.

He first talked about experimental studies showing different arguments for solar power appeal differently depending on one’s political view.

Kiran then proceeded to the main study which modeled household adoption rates for 440,000 homes near San Diego between 2007 and 2011 based on energy usage, demographics, size of house, unemployment rate, neighbors with solar panels, etc. When information was only available for those who had adopted solar panels, a model of factors were created to fill in the missing data. To create a model he projected the characteristics backwards in time to create a monthly-by-month data set and fit a logistic regression (hazard model) to the factors.

Significant variables predicting higher adoption rates included: owner occupy, household size, peer at ¼ mile, have a pool. These findings are noteworthy since financial incentives such as energy rebates are not in the list. This indicates that a policy of increased rebates will have only a marginal effect on adoption rates.

Kiran concluded by talking about another policy that might in theory cause a greater increase in adoptions: seed neighborhoods by randomly installing free systems. This would jump start the peer effect and would assist in the dissemination of information about home solar power.

He encouraged the audience to contact him (klakkar@sandia.gov) about the studies he presented and contact Eugene Vorobeychik @ Vanderbilt Univ (eug.vorobey@gmail.com) for copies of the behavioral studies.

posted in:  NYC Energy Data    / leave comments:   No comments yet