The Great Energy Predictor Shootout II: Measuring retrofit savings - Overview and discussion of results
- Additional Document Info
- View All
A second predictor shootout contest was held to evaluate the effectiveness of empirical or inverse models for hourly whole-building energy baselines to measure savings from energy conservation retrofits. The accuracy of the contestants' model was evaluated by determining their abilities to predict data that were carefully removed from the training or pre-retrofit period. Results from the contest showed that neural networks provided the most accurate model of a building's energy use. Second contest's results showed that cleverly assembled statistical models appeared to be as accurate or in some cases, more accurate than some of the neural network entries.
author list (cited authors)
Haberl, J. S., Thamilseran, S., & ASHRAE, ..