Abstract

In this manuscript we explore European feature importance in Day Ahead Market (DAM) price forecasting models, and show that model performance can deteriorate when too many features are included due to over-fitting. We propose a greedy algorithm to search over candidate countries for European features to be used in a DAM price forecasting model, that can be applied to several regression and machine learning modelling methodologies. We apply the algorithm to build price forecasting models for the Dutch market, using candidate countries selected through an integrated analysis based on open-source European electricity market data. Feature importance is visualised using an Auto Regressive and Random Forest model. We explain these results using cross-border flow and DAM price data. Two types of models (LEAR and the Deep Neural Network) are considered for the DAM price forecasting with and without European features. We show that in the Dutch case, taking European market integration into account improves the Mean Absolute Error (MAE) of the best performing DAM price forecasting model by 3.1%, the relative MAE (rMAE) by 3.85%, and the Symmetrical Mean Absolute Percentage Error (sMAPE) by 0.31 p.p., compared to the best forecasting model without European features. Through statistical testing we show that European features improve the accuracy of the forecasts with statistical significance.

Highlights

  • D EMAND RESPONSE (DR) is an energy management technique where energy consumers are incentivised to shift their energy use in time [5], [6]

  • In order to include EU market integration in a bidding zone’s Day Ahead Market (DAM) forecasting model, we propose a greedy algorithm as defined in Algorithm 1

  • In this manuscript, we propose a method for searching optimal combinations of European features in Day Ahead Market (DAM) price forecasting models

Read more

Summary

INTRODUCTION

D EMAND RESPONSE (DR) is an energy management technique where energy consumers are incentivised to shift their energy use in time [5], [6]. Inter-connectivity enables cross-border electricity trading, opening up national electricity markets to foreign demand These facts already make a strong case for the consideration of European market integration in a DAM price forecasting model. Many statistical methods rely on the calibration of linear relationships While they can still be powerful modelling approaches, they might not perform well with high-resolution data like hourly prices with high volatility [21]. It is possible that as renewable energy penetration increases, price volatility would increase and ML methods would increasingly outperform statistical methods In this manuscript we perform an EU wide, data-based analysis of European market feature (e.g. price and load) importance in DAM price forecasting models of European bidding zones. Temporal variations in relative model forecasting performance will be analysed (Section V-D) using univariate Diebold-Mariano-tests (DMtests) on the hourly forecasts, in combination with Kernel Density Estimates of the daily DAM prices per hour of the day

ELECTRICITY PRICE FORECASTING METHODS
LASSO ESTIMATED AUTO REGRESSIVE MODEL
DEEP NEURAL NETWORK
DIEBOLD-MARIANO TEST
GREEDY ALGORITHM FOR EUROPEAN FEATURE SEARCH
DATA AND TOOLS
EUROPEAN FEATURE IMPORTANCE ANALYSIS
DUTCH DAM PRICE FORECASTING WITH EU MARKET INTEGRATION
Findings
CONCLUSION
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call