Day-Ahead Price Forecasting in Great Britain’s BETTA Electricity Market

Daniel McGlynn, Sonya Coleman, Dermot Kerr, Catherine McHugh

Research output: Contribution to conferencePaper

Abstract

The characteristics of commodities such as
electricity, natural gas and oil mean that standard statisticsbased
pricing and prediction models that are typically applied
in financial markets cannot readily be transferred and used as
energy pricing models. Therefore, we investigate the use of
computational intelligence-based approaches for electricity
price forecasting. This paper compares two models for dayahead
electricity price forecasting, an AdaBoosted ensemble
of the Extra-Trees algorithm and a Generalized Regression
Neural Network (GRNN). In this work both forecasting
models were applied to the national electricity market of Great
Britain, the British Energy and Electricity Trading
Arrangements (BETTA). The models were evaluated using the
mean absolute percentage error (MAPE) statistic and the
results show that the GRNN yielded a comparable forecasting
error to the AdaBoosted algorithm with a significantly faster
computation time.

Conference

ConferenceSYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE
Abbreviated titleCiFER
CountryIndia
CityBenguluru
Period18/11/1820/11/18

Fingerprint

Electricity
Error statistics
Trees (mathematics)
Natural gas
Power markets
Costs

Keywords

  • Extra-Trees
  • Generalized Regression Neural Networks
  • British Energy and Electricity Trading Arrangements (BETTA)
  • Machine Learning
  • Price Forecasting

Cite this

McGlynn, D., Coleman, S., Kerr, D., & McHugh, C. (Accepted/In press). Day-Ahead Price Forecasting in Great Britain’s BETTA Electricity Market. Paper presented at SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE, Benguluru, India.
McGlynn, Daniel ; Coleman, Sonya ; Kerr, Dermot ; McHugh, Catherine. / Day-Ahead Price Forecasting in Great Britain’s BETTA Electricity Market. Paper presented at SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE, Benguluru, India.
@conference{b065d27dcb9f49c09de252d7f052b008,
title = "Day-Ahead Price Forecasting in Great Britain’s BETTA Electricity Market",
abstract = "The characteristics of commodities such aselectricity, natural gas and oil mean that standard statisticsbasedpricing and prediction models that are typically appliedin financial markets cannot readily be transferred and used asenergy pricing models. Therefore, we investigate the use ofcomputational intelligence-based approaches for electricityprice forecasting. This paper compares two models for dayaheadelectricity price forecasting, an AdaBoosted ensembleof the Extra-Trees algorithm and a Generalized RegressionNeural Network (GRNN). In this work both forecastingmodels were applied to the national electricity market of GreatBritain, the British Energy and Electricity TradingArrangements (BETTA). The models were evaluated using themean absolute percentage error (MAPE) statistic and theresults show that the GRNN yielded a comparable forecastingerror to the AdaBoosted algorithm with a significantly fastercomputation time.",
keywords = "Extra-Trees, Generalized Regression Neural Networks, British Energy and Electricity Trading Arrangements (BETTA), Machine Learning, Price Forecasting",
author = "Daniel McGlynn and Sonya Coleman and Dermot Kerr and Catherine McHugh",
year = "2018",
month = "9",
day = "1",
language = "English",
note = "SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE : IEEE Symposium on Computational Intelligence for Financial Engineering and Economics, CiFER ; Conference date: 18-11-2018 Through 20-11-2018",

}

McGlynn, D, Coleman, S, Kerr, D & McHugh, C 2018, 'Day-Ahead Price Forecasting in Great Britain’s BETTA Electricity Market' Paper presented at SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE, Benguluru, India, 18/11/18 - 20/11/18, .

Day-Ahead Price Forecasting in Great Britain’s BETTA Electricity Market. / McGlynn, Daniel; Coleman, Sonya; Kerr, Dermot; McHugh, Catherine.

2018. Paper presented at SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE, Benguluru, India.

Research output: Contribution to conferencePaper

TY - CONF

T1 - Day-Ahead Price Forecasting in Great Britain’s BETTA Electricity Market

AU - McGlynn, Daniel

AU - Coleman, Sonya

AU - Kerr, Dermot

AU - McHugh, Catherine

PY - 2018/9/1

Y1 - 2018/9/1

N2 - The characteristics of commodities such aselectricity, natural gas and oil mean that standard statisticsbasedpricing and prediction models that are typically appliedin financial markets cannot readily be transferred and used asenergy pricing models. Therefore, we investigate the use ofcomputational intelligence-based approaches for electricityprice forecasting. This paper compares two models for dayaheadelectricity price forecasting, an AdaBoosted ensembleof the Extra-Trees algorithm and a Generalized RegressionNeural Network (GRNN). In this work both forecastingmodels were applied to the national electricity market of GreatBritain, the British Energy and Electricity TradingArrangements (BETTA). The models were evaluated using themean absolute percentage error (MAPE) statistic and theresults show that the GRNN yielded a comparable forecastingerror to the AdaBoosted algorithm with a significantly fastercomputation time.

AB - The characteristics of commodities such aselectricity, natural gas and oil mean that standard statisticsbasedpricing and prediction models that are typically appliedin financial markets cannot readily be transferred and used asenergy pricing models. Therefore, we investigate the use ofcomputational intelligence-based approaches for electricityprice forecasting. This paper compares two models for dayaheadelectricity price forecasting, an AdaBoosted ensembleof the Extra-Trees algorithm and a Generalized RegressionNeural Network (GRNN). In this work both forecastingmodels were applied to the national electricity market of GreatBritain, the British Energy and Electricity TradingArrangements (BETTA). The models were evaluated using themean absolute percentage error (MAPE) statistic and theresults show that the GRNN yielded a comparable forecastingerror to the AdaBoosted algorithm with a significantly fastercomputation time.

KW - Extra-Trees

KW - Generalized Regression Neural Networks

KW - British Energy and Electricity Trading Arrangements (BETTA)

KW - Machine Learning

KW - Price Forecasting

M3 - Paper

ER -

McGlynn D, Coleman S, Kerr D, McHugh C. Day-Ahead Price Forecasting in Great Britain’s BETTA Electricity Market. 2018. Paper presented at SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE, Benguluru, India.