A Modified Pattern Sequence-Based Forecasting Method for Electricity Price

DOI

10.2991/caai-17.2017.76How to use a DOI?

Keywords

locally weighted linear regression; pattern sequence-based forecasting (PSF); electricity price; time series forecasting

Abstract

Electricity price forecasting is a relevant yet hard task in the field of one step time series forecasting. A new approach called Pattern Sequence-based Forecasting (PSF) shows a remarkable improvement in the time series prediction. The PSF consists of four steps: clustering, extraction pattern subsequence, sliding window to find similar subsequence, predicting. However, it doesn't consider the characteristic of correlation which changes with the length of the time gap. In this paper, a modified machine learning method based on similarity of pattern sequences which is combined with the locally weighted linear regression, is proposed to forecast electricity prices. The main novelty is that the modified PSF consider the problem of correlation which changes the mean value into weighted mean value and the PSF is firstly used in the price forecasting track of the Global Energy Forecasting Competition 2014 which shows a significantly better performance.

Copyright

© 2017, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Huaizhi Qiu
AU  - Lingling Zhao
AU  - Xiaohong Su
PY  - 2017/06
DA  - 2017/06
TI  - A Modified Pattern Sequence-Based Forecasting Method for Electricity Price
BT  - Proceedings of the 2017 2nd International Conference on Control, Automation and Artificial Intelligence (CAAI 2017)
PB  - Atlantis Press
SP  - 337
EP  - 341
SN  - 1951-6851
UR  - https://doi.org/10.2991/caai-17.2017.76
DO  - 10.2991/caai-17.2017.76
ID  - Qiu2017/06
ER  -