Machine Learning from Schools about Energy Efficiency
E2e Faculty Directors Catherine Wolfram and Cristopher Knittel, in collaboration with E2e research affiliates Fiona Burlig (University of Chicago), David Rapson (University of California, Davis) and Mar Reguant (Northwestern University) used a novel, machine learning approach to analyze schools' electricity consumption after efficiency upgrades. Simply comparing one school that invested in a new air conditioner to one that didn’t can mask important differences that impact their energy use – one school may be in San Francisco and the other in a hot Central Valley town. In this novel study, the researchers essentially compare each school to itself, both before and after upgrades were made. Using this approach ensures that the study is measuring just the effect of the upgrade.
The study shows that energy efficiency improvements lowered electricity consumption on average by 3 percent. However, they also found that these schools received on average only 24 percent of the energy savings that were projected before officials invested in the upgrades. Lighting and heating and cooling upgrades delivered the greatest savings—49 and 42 percent of expected savings on a school-by-school basis, respectively. This study was made possible with funding and support from the California Public Utilities Commission.
Read more about the research and the analysis below.
E2e Working Paper 032, Machine Learning from Schools about Energy Efficiency