The impacts of weather attributes on commercial and residential electricity demands and their components in the northwestern United States were examined. Two machine learning methods, regression tree (RT), and random forest (RF), were integrated and compared. Both RT and RF models provide reliable prediction of commercial cooling load. RF models particularly yield higher accuracy with reduced overfitting.
Revised: November 27, 2019 |
Published: June 28, 2019
Citation
Zhou H., Z. Hou, P.V. Etingov, and Y. Liu. 2019.Machine Learning of Commercial and Residential Load Components in Northwestern United States. In The ACM e-Energy 2019 Conference, 385-387. New York, New York:ACM.PNNL-SA-142105.doi:10.1145/3307772.3330160