Predicting baseline for analysis of electricity pricing. (2018)

• Record Type: Journal Article
• Authors: Kim, Taehoon; Choi, Jaesik; Lee, Dongeun; Sim, Alex; Spurlock, C. Anna; Todd, Annika; Wu, Kesheng
• Is Part Of: International journal of big data intelligence. Volume 5:Number 1/2(2018)
• Page Start: 3
• Page End: 20
• Publication Date: 2018
• Subjects: baseline model -- residential electricity consumption -- outdoor temperature -- gradient tree boosting -- GTB -- electricity rate scheme; Big data -- Periodicals; 005.705
• Languages: English
• ISSNs: 2053-1389

Abstract

To understand the impact of a new pricing structure on residential electricity peak demands, we need a baseline model that captures every factor other than the new price. The gold standard baseline is a randomised control trial, however, control trials are hard to design. The alternative to learn a baseline model from the past measurements fails to make reliable predictions about the daily peak usage values next summer. To overcome these shortcomings, we propose several new methods. Among these methods, the one named LTAP is particularly promising. It accurately predicts future usages of the control group. It also predicts the reductions of the peak demands to remain the same, while previous studies have found the reduction to be diminishing over time. We believe that LTAP is capturing the self-selection bias of the treatment groups better than techniques used in previous studies and are looking for opportunities to confirm this feature.