Modeling short-term energy load with continuous conditional random fields

  1. Get@NRC: Modeling short-term energy load with continuous conditional random fields (Opens in a new window)
DOIResolve DOI:
AuthorSearch for:
TypeBook Chapter
Proceedings titleMachine Learning and Knowledge Discovery in Databases : European Conference, ECML PKDD 2013, Prague, Czech Republic, September 23-27, 2013, Proceedings, Part I
Series titleLecture Notes In Computer Science; Volume 8188
ConferenceEuropean Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD 2013), September 23-27, 2013, Prague
Pages433448; # of pages: 16
SubjectBenchmarking methods; Conditional random field; Discriminative approach; Energy demands; Mean absolute percentage error; Predictive accuracy; Predictive performance; Root mean square errors; Electric load forecasting; Forecasting; Learning systems; Mean square error; Random processes
AbstractShort-term energy load forecasting, such as hourly predictions for the next n (n ≥ 2) hours, will benefit from exploiting the relationships among the n estimated outputs. This paper treats such multi-steps ahead regression task as a sequence labeling (regression) problem, and adopts a Continuous Conditional Random Fields (CCRF) strategy. This discriminative approach intuitively integrates two layers: the first layer aims at the prior knowledge for the multiple outputs, and the second layer employs edge potential features to implicitly model the interplays of the n interconnected outputs. Consequently, the proposed CCRF makes predictions not only basing on observed features, but also considering the estimated values of related outputs, thus improving the overall predictive accuracy. In particular, we boost the CCRF's predictive performance with a multi-target function as its edge feature. These functions convert the relationship of related outputs with continuous values into a set of "sub-relationships", each providing more specific feature constraints for the interplays of the related outputs. We applied the proposed approach to two real-world energy load prediction systems: one for electricity demand and another for gas usage. Our experimental results show that the proposed strategy can meaningfully reduce the predictive error for the two systems, in terms of mean absolute percentage error and root mean square error, when compared with three benchmarking methods. Promisingly, the relative error reduction achieved by our CCRF model was up to 50%. © 2013 Springer-Verlag..
Publication date
PublisherSpringer Berlin Heidelberg
AffiliationInformation and Communication Technologies; National Research Council Canada
Peer reviewedYes
NPARC number21270677
Export citationExport as RIS
Report a correctionReport a correction
Record identifierc4c72ccd-7e97-4e26-b3b8-51ed3f13a2e7
Record created2014-02-17
Record modified2016-07-11
Bookmark and share
  • Share this page with Facebook (Opens in a new window)
  • Share this page with Twitter (Opens in a new window)
  • Share this page with Google+ (Opens in a new window)
  • Share this page with Delicious (Opens in a new window)